REPORTS OF THE TIBOR T. POLGAR

FELLOWSHIP PROGRAM, 2018

Sarah H. Fernald, David J. Yozzo, and Helena Andreyko

Editors

A Joint Program of The Hudson River Foundation and The New York State Department of Environmental Conservation

December 2020

i ii ABSTRACT

Eight studies completed within the Hudson River Estuary under the auspices of the

Tibor T. Polgar Fellowship Program during 2018 have been included in the current volume.

Major objectives of these studies included: (1) determining the effects of light, nutrients, and

temperature on cyanobacterial blooms, (2) quantifying the differences in microplastic

concentrations among marsh, tributary, and open water locations in the Hudson River

watershed, (3) determining the effect of microplastic size and shape on the uptake ability of

the Eastern Oyster (Crassostrea virginia), (4) evaluating the effect of salinity on gametogenesis in Eastern Oysters in the Hudson River, (5) determining the effect of the redox environment on anaerobic biodegradability of personal care products by native microorganisms in anoxic estuarine sediments, (6) comparing Vallisneria americana reproduction modes between sites in the Hudson River and Chesapeake Bay,

(7) characterizing habitat use of tidal wetlands by the painted turtle (Chrysemys picta), and

(8) using core sample analysis to determine the environmental history of Ramshorn-

Livingston Marsh.

iii iv TABLE OF CONTENTS

Abstract ...... iii

Preface ...... vi

Fellowship Reports

Controls on Cyanobacteria Growth in the Hudson River Estuary Corey W. Rundquist and Stuart E. G. Findlay ...... I-1

Quantification of Microplastic Content in Surface Water and Sediment within Hudson River Tributaries and Marshes Jason Randall, Zofia Gagnon, and Christopher Bowser ...... II-1

Size Does Matter: Exposure and Effects of Microplastics on the Eastern Oyster (Crassostrea virginica) Erika Bernal, Paul A. X. Bologna, and Beth Sharack ...... III-1

The Effect of Salinity on Eastern Oyster Reproduction in the Hudson River Estuary Kaili M. Gregory and Matthew Hare ...... IV-1

Bugs on Drugs: The Influence of Redox Environments on the Microbial Degradation of Pharmaceuticals in the Hudson River Watershed Michelle L. Zeliph and Max M. Haggblom ...... V-1

Assessing Mode of Reproduction in Vallisneria americana of the Hudson River, NY, and the Chesapeake Bay, MD Carrie E. Perkins and Maile C. Neel ...... VI-1

Painted Turtle Ecology in a Freshwater Tidal Marsh: Concluding Survey Virginia Caponera and Erik Kiviat ...... VII-1

Human Impact on Ramshorn-Livingston, A Hudson River Freshwater Tidal Marsh Elizabeth Thompson and Dorothy Peteet ...... VIII-1

v 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 2018 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.

vi 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 2018. 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.

Sarah H. Fernald New York State Department of Environmental Conservation Hudson River National Estuarine Research Reserve

David J. Yozzo Glenford Environmental Science

Helena Andreyko Hudson River Foundation for Science and Environmental Research

vii

CONTROLS ON CYANOBACTERIA GROWTH IN THE HUDSON RIVER ESTUARY

A Final Report of the Tibor T. Polgar Fellowship Program

Corey W. Rundquist

Polgar Fellow

Cary Institute of Ecosystem Studies Millbrook, NY 12545

Project Advisor:

Stuart E. G. Findlay Aquatic Ecologist Cary Institute of Ecosystem Studies Millbrook, NY 12545

Rundquist, C. W. and S. E. G. Findlay. 2020. Controls on Cyanobacteria Growth in the Hudson River Estuary. Section I: 1-23 pp. In S.H. Fernald, D.J. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

I-1 ABSTRACT

In this study, the effects of light, nutrient and temperature on potential for cyanobacterial bloom formation were tested under simulated Hudson River conditions.

The results from this experiment were then compared to actual Hudson River conditions in order to better predict the formation conditions for harmful cyanobacterial blooms. The first experiment measured the effects of temperature as well as nitrogen and phosphorus levels on the cyanobacteria concentration of small scale cultures with natural inoculants.

The second experiment tested the variables of elevated nutrient levels and depth under field conditions at Norrie Point in Staatsburg, NY. After the cultures grew for three weeks, the results from the first experiment showed that increased temperature had a significant, positive effect on cyanobacterial concentration. Additionally, the results from the second experiment showed that elevated levels of nitrogen and phosphorus significantly increased cyanobacterial concentration. The results from this experiment convey that cyanobacteria can quickly grow up to harmful concentrations, in warm, nutrient-rich water. In the Hudson River, these conditions could be found in a relatively stagnant, open section of river, downstream of a sewage treatment plant, in late summer.

I-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

Greenhouse Experiment ...... I-8

River Experiment ...... I-10

Results ...... I-11

Greenhouse Experiment ...... I-11

River Experiment ...... I-14

Discussion ...... I-16

Acknowledgments...... I-21

References ...... I-22

I-3 LIST OF FIGURES

Figure 1 – Picture of the experimental setup testing the effects of temperature and nutrients on cyanobacteria concentration. The higher temperature treatment is depicted on the left and the lower temperature is depicted on the right...... I-9

Figure 2 – A picture of the experimental rig testing the effects of nutrients and depth on cyanobacteria concentration, placed in the Hudson River at Norrie Point, Staatsburg, NY...... I-11

Figure 3 – Average chlorophyll (µg/L) measurements of four culture treatments grown over three weeks. Temperature was found to have a significant effect on culture growth (p<0.05), and enriched nutrient levels of nitrogen and phosphorus did not have a significant effect (p>0.05)...... I-13

Figure 4 – Average phycocyanin (µg/L) measurements of four different treatments on cyanobacterial cultures, grown over three weeks. Temperature had a significant effect on culture growth (p<0.05), but elevated nitrogen and phosphorus levels did not have a significant effect (p>0.05)...... I-13

Figure 5 – The average chlorophyll (µg/L) measurements for cyanobacterial culture growth with nutrient level and depth variables. Elevated levels of nitrogen and phosphorus had a significant effect on culture growth over two weeks. Depth did not have a significant effect...... I-15

Figure 6 – The phycocyanin (µg/L) measurements for four treatments with nutrient levels and depth on cyanobacterial culture growth over two weeks. Elevated levels of nitrogen and phosphorus had a significant effect on culture growth, and the growth depth variable did not have a significant effect...... I-15

Figure 7 – The comparison of average peak growth for unaltered hot and unaltered cold treatments in the greenhouse experiment, and enriched submerged and unaltered submerged treatments in the river experiment. Temperature had a significant effect on cyanobacterial growth in the greenhouse experiment and enriched nutrient levels had a significant effect on culture growth in the river experiment...... I-16

I-4 INTRODUCTION

Cyanobacteria, or blue-green algae, are an integral part of aquatic ecosystems

(WHO 1999a). Blue-green algae are photoautotrophic bacteria that can live in both salt and freshwater, often becoming a large part of the biomass in rivers and lakes. One key characteristic about cyanobacteria is that under certain conditions, they are able to grow at extremely rapid rates and reach very high abundances in what is known as a bloom.

What is specifically dangerous about these blooms is that they sometimes contain certain types of cyanobacteria that produce toxins, known as cyanotoxins. These cyanotoxins pose a serious threat to bodies of water used for drinking water and recreational uses, causing serious damage to the surrounding ecosystem, citizens, and wild and domestic animals that live near the affected body of water (EPA 2017). Extended exposure can cause serious damage to the liver, including liver cancer, and can even be neurotoxic

(Hitzfeld et al. 2000; Hudnell 2008; Carmichael 1994). These harmful effects are shown in a study of cyanobacteria that took place in Chesapeake Bay. According to the study, blue-green algae were responsible for a series of substantial bird and fish kills. The toxins also caused very real effects on humans, including skin rashes, nausea, and fevers (Tango and Butler 2008). Because of this harmful potential, cyanobacterial blooms are a primary issue in providing people with clean water. In the context of anthropogenic pollution and climate change, clean water is something that will only gain more importance in the coming decades (Hitzfeld et al. 2000). Furthermore, cyanobacterial prevalence has been increasing at a disproportionate rate to that of other phytoplankton, and are becoming more abundant in the U.S. and worldwide (Hudnell 2008; Taranu et al. 2015).

I-5 The lower Hudson River is a tidal, well-mixed, nutrient-rich estuary in eastern

New York State (Fernald et al. 2007).The lower Hudson River has fluctuating values of turbidity, dissolved organic carbon, and nutrient levels, including nitrogen and phosphorus. Recently, the presence of cyanobacterial blooms has become an issue for the

Hudson River Valley. The worst of these blooms mostly occur in the numerous slower- moving tributaries of the Hudson River (Fernald et al. 2007). An example of one of these blooms in the Hudson River valley occurred in the Wallkill River, in which cyanotoxins were reported up to 25 times the DEC’s “High Toxin” threshold (Riverkeeper 2016). The cyanotoxins in this bloom posed a real threat to both the local ecosystem and community, harming aquatic organisms as well as domestic pets. If such a cyanobacterial bloom were to happen in the Hudson River, it would pose a real danger to the estuary ecosystem.

Cyanobacterial blooms are only able to form under certain ranges of water quality and atmospheric conditions (CDC 2017). Three main factors for the formation conditions of algal blooms identified by the World Health Organization are temperature, phosphorus and nitrogen levels, and light intensity (WHO 1999a). Cyanobacteria can withstand a wide range of temperatures from arctic lakes to hot springs; however, blooms only form in warmer conditions during the summer, and will usually last for a couple of weeks at a time (WHO 1999a). The trigger water temperature for cyanobacterial biomass dominance is about 25ºC, which means that blooms most frequently occur in the late summer months in the Hudson Valley (Fernald et. al 2007). Nitrogen and phosphorus are two of the primary nutrients that promote cyanobacterial growth and bloom formation (Davis et al.

2015). These nutrients are an important factor for cyanobacterial blooms for two reasons, the first being that cyanobacteria have a very high nutrient affinity and are able to

I-6 outcompete other phytoplankton for phosphorus and nitrogen. The second is that cyanobacteria are very efficient at nutrient storage, gathering enough at once to sustain up to four cell divisions (WHO 1999a). Sunlight conditions are important for the formation of blooms as most cyanobacteria prefer lower levels of light compared to other phytoplankton (Lopez et al. 2008). The optimal growth conditions of cyanobacteria are approximately half of direct sunlight level, and overexposure to intense light levels can be detrimental to cyanobacteria (WHO 1999b). Furthermore, transient microstratification in the Hudson River can lead to increased temperature, nutrient accumulation, and sunlight exposure. These conditions create a high overall phytoplankton biomass in the euphotic zone, and eventually can increase cyanobacterial dominance when resources become depleted (Fernald et al. 2007).

When trying to predict harmful algal blooms in the actual environment, it is very important to take these different factors of cyanobacterial growth into consideration. The purpose of this experiment was to analyze optimal conditions for the formation of harmful algal blooms in the Hudson River by testing different cyanobacterial growth factors in conditions similar to the Hudson River. The hypothesis that increased temperature, increased nitrogen and phosphorus levels, and higher sunlight exposure lead to increased cyanobacteria growth, was tested in this experiment The first experimental setup tested the effects of temperature, as well as elevated phosphorus and nitrogen levels on cyanobacterial growth in a controlled environment on land. The second experiment tested the variables of elevated nutrients and depth on isolated volumes of Hudson River water. This experiment examined cyanobacteria in conditions as close to the Hudson

River as possible, without actually exposing the river to potentially toxic organisms. The

I-7 degree to which these different variables alter cyanobacterial growth will form a picture of what harmful algal bloom formation conditions look like in the Hudson River, helping to predict these blooms.

METHODS

Greenhouse Experiment

In the first experiment, the effects of temperature and increased nutrient levels on cyanobacterial growth were tested on small cultures in a controlled environment.

To begin culture growth, a natural inoculant was collected from a small pond with a history of algal blooms located in Tivoli, NY. This pond collection was then filtered to

35 µm to isolate the cyanobacteria for inoculation. In the experiment, a total of 16 cultures were prepared and subjected to four different treatments. First, sixteen 500 ml flasks were filled with deionized water. Half of these flasks were then fertilized using potassium nitrate and sodium phosphate. These flasks were raised to nitrogen and phosphorus levels twice that of natural levels documented from the Hudson River by

Fernald et al.(2007), with concentrations of 70 µM nitrogen and 2.24 µM phosphorus.

After, 20 ml of natural inoculant was added to all sixteen flasks and were ready for testing.

The experiment took place in a greenhouse on the Cary Institute of Ecosystem

Studies campus, located in Millbrook, NY. In order to manipulate temperature, two clear

40”x20”x7” containers were placed adjacent to each other on a table to serve as water bath for the algal cultures. One container was subjected to an average ambient temperature of 25.34ºC, while the other was cooled to an average temperature of approximately 20.43ºC. This second container was cooled by a Fischer Scientific Isotemp

I-8 Water Bath, externally pumping cold water through a 2 ft long copper coil, placed in the water container. After this, shadecloth was draped over the experimental setup, controlling the cultures to receive an average sunlight level of 203.24 µE/m2/s during the day.

This experiment ran for three weeks, with the growth of the cultures being measured by a YSI Exo 2 sonde with chlorophyll-a and phycocyanin fluorometric probes in µg/L. The peak concentrations for all four treatments were analyzed by a repeated- measures Analysis of Variance (ANOVA), using the program JMP, version 13. In addition to this, laboratory chlorophyll measurements were taken on Day 11 and Day 14 of the experiment (Yéprémian et al. 2017). Temperature and light data were monitored throughout the experiment using HOBO Pendant water data loggers.

Figure 1: Picture of the experimental setup testing the effects of temperature and nutrients on cyanobacteria concentration. The higher temperature treatment is depicted on the left and the lower temperature is depicted on the right.

I-9

River Experiment

For the second experiment, the effects of elevated nutrient levels and depth on culture growth were tested under realistic Hudson River conditions.

The design of this experiment included a floating frame that was suspended in the

Hudson River, with attached containers full of isolated river water and ambient cyanobacterial communities. To keep the rig afloat a 1m x 1m square was constructed out of 1-1/4” PVC piping. The square was held together by square elbow joints and PVC cement. Across the square, four lengths of nylon rope were strung both lengthwise and widthwise, forming four intersections in the middle of the square. Four 1.5 m long chains were then clipped to the four intersections, able to hang down into the water. On each of these chains, two LDPE 10 L clear cubitainers were fastened at the very top and at 1 m down the chain. In total, the rig had four cyanobacterial cultures growing at the surface level, and four growing at a depth of 1 m.

This experiment took place adjacent to the dock of the Norrie Point

Environmental Center in Staatsburg, NY. The experimental rig was placed about 20 m offshore, upriver of the dock. One side of the PVC square was tied to the dock and the other side was tied to a submerged cinderblock, keeping the rig suspended in the same area during high and low tide. After this, all of the cubitainers were filled with unaltered

Hudson River water. Half of the surface and submerged cultures had their nitrogen and phosphorus levels raised to twice the ambient level, as done in the first experiment. The cultures were sealed in the cubitainers and the whole rig was set to grow for two weeks.

The growth of the cultures was monitored using the same phycocyanin and chlorophyll-a

I-10 fluorometric probes on the YSI Exo 2 sonde, measuring in µg/L. The measurements from these probes were analyzed by running the same repeated measures ANOVA test on the peak concentration for all four treatments. Water samples were also taken on Day 1 and

Day 7 of the experiment, used for laboratory chlorophyll extraction measurements, as done in the first experiment.

Figure 2: A picture of the experimental rig testing the effects of nutrients and depth on cyanobacteria concentration, placed in the Hudson River at Norrie Point, Staatsburg, NY.

RESULTS

Greenhouse Experiment

At the outset of the greenhouse experiment, all four treatments of cultures started very slowly with little to no growth for the first week. After that point, the warmer temperature treatments started to grow abruptly, observed both quantitatively and qualitatively, turning green in color (Figures 1, 3, and 4). The high temperature treatment cultures then rose to a higher concentration than the cold treatments, as shown by the

I-11 error bars on both the chlorophyll and phycocyanin graphs. In addition to this, the higher temperature cultures reached their peak concentration at Day 13, which was slightly earlier than the colder temperature cultures on Day 14 (Figures 3 and 4). On Day 13, a repeated measures ANOVA test was run on the cyanobacteria concentration in relation to temperature, resulting in a p-value of less than .0001, indicating a significant effect of temperature on culture growth for both phycocyanin and chlorophyll-a measurements.

All of the cultures exhibited growth for roughly the same amount of time, with no signs of further cyanobacterial accumulation at three weeks. The temperature variable was found to have a significant effect on cyanobacterial growth, determined through statistical analysis on both chlorophyll and phycocyanin parameters (ANOVA, p<0.05) (Figure 7).

The cultures enriched with nitrogen and phosphorus had greater overall measurements for both chlorophyll and phycocyanin. The cultures with elevated nutrients for both higher and lower temperature treatments had greater peak concentrations than their unaltered counterparts (Figures 3 and 4). This was not consistent among all four cultures within each treatment. The elevated levels of nitrogen and phosphorus did not have a significant effect on peak cyanobacterial culture growth (ANOVA, p>0.05).

Overall, the cyanobacterial growth through all treatments was consistent and steady, verified through the similar measurements of both chlorophyll and phycocyanin pigments.

I-12 Greenhouse Natural Inoculant Culture Growth (Chlorophyll-a) 600

500

400 a µg/L a - 300

200

100 Chlorophyll 0 0 5 10 15 20 25 -100 Time (Days)

Enriched Nutrients Cool Ambient Nutrients Cool Ambient Nutrients Warm Enriched Nutrients Warm

Figure 3: Average chlorophyll (µg/L) measurements of four culture treatments grown over three weeks. Temperature was found to have a significant effect on culture growth (p<0.05), and enriched nutrient levels of nitrogen and phosphorus did not have a significant effect (p>0.05).

Greenhouse Natural Inoculant Culture Growth (Phycocyanin) 20

15

10

5

Phycocyanin µg/L Phycocyanin 0 0 5 10 15 20 25 -5 Time (Days)

Enriched Nutrients Cool Ambient Nutrients Cool Ambient Nutrients Warm Enriched Nutrients Warm

Figure 4: Average phycocyanin (µg/L) measurements of four different treatments on cyanobacterial cultures, grown over three weeks. Temperature had a significant effect on culture growth (p<0.05), but elevated nitrogen and phosphorus levels did not have a significant effect (p>0.05).

I-13 River Experiment

For the river experiment, the cultures grew rapidly overall, with noticeable growth on Day 3. In the beginning, the enriched surface cultures had the highest cyanobacterial concentration in comparison to other treatments, peaking on Day 5 for both chlorophyll and phycocyanin (Figure 5,6). From then on, the enriched surface treatment decreased in concentration, and the enriched bottom treatment exhibited the highest algal concentrations for the remainder of the experiment. In contrast, the two unaltered surface and bottom treatments had the lowest concentrations throughout the whole experiment.

These cultures both peaked early, approximately one week into growth, and exhibited very small changes in concentration throughout the experiment. On Day 6, the enriched bottom culture was comparable to the unaltered bottom culture, but increased from that point on (Figure 5). By Day 8, the elevated levels of nitrogen and phosphorus had a significant effect on cyanobacterial concentration (ANOVA, p<0.05). This effect is illustrated through the comparison of peak concentrations between the enriched bottom and unaltered bottom treatments (Figure 7).

In terms of the growth depth variable, there was no clear trend in either the chlorophyll or phycocyanin data. For the enriched cultures, the surface culture treatments peaked first at Day 5, but then decreased and were surpassed by bottom culture concentrations (Figure 5,6). For phycocyanin, the higher concentration of the enriched bottom treatment was more apparent, as shown by the error bars at the end of the experiment (Figure 6). In contrast, the chlorophyll data had more variance in the measurements. Furthermore, the enriched surface treatment exhibited the highest peak chlorophyll concentration among all other treatments (Figure 5). For the unaltered

I-14 cultures, both the surface and bottom treatments had very similar readings for both probes. Overall, the bottom unaltered treatment had greater values, but these data points often overlapped with the surface unaltered cultures. Depth did not have an effect on the peak concentration of cyanobacterial culture growth (ANOVA, p>0.05).

Isolated Hudson River Water Culture Growth (Chlorophyll-a) 80 70 60

50 a (µg/L) a - 40 30 20

Chlorphyll 10 0 1 4 6 8 11 13 15 Time (Days)

Enriched Nutrients Surface Ambient Nutrients Surface Enriched Nutrients Bottom Ambient Nutrients Bottom

Figure 5: The average chlorophyll (µg/L) measurements for cyanobacterial culture growth with nutrient level and depth variables. Elevated levels of nitrogen and phosphorus had a significant effect on culture growth over two weeks. Depth did not have a significant effect.

Isolated Hudson River Water Culture Growth (Phycocyanin) 6 5 4 3 2

1 Phycocyanin (µg/L) Phycocyanin 0 1 4 6 8 11 13 15 Time (Days)

Enriched Nutrients Surface Ambient Nutrients Surface Enriched Nutrients Bottom Ambient Nutrients Bottom

Figure 6: The phycocyanin (µg/L) measurements for four treatments with nutrient levels and depth on cyanobacterial culture growth over two weeks. Elevated levels of nitrogen and phosphorus had a significant effect on culture growth, and the growth depth variable did not have a significant effect.

I-15

Temperature and Nutrient Effect on Peak Cyanobacteria Abundance 400 350

300

a (µg/L) a - 250 200 150 100 50 0

Greenhouse Experiment River Experiment Average Peak Chlorophyll Peak Average Elevated Nutrients or Temperature Unaltered Cultures

Figure 7: The comparison of average peak growth for unaltered hot and unaltered cold treatments in the greenhouse experiment, and enriched submerged and unaltered submerged treatments in the river experiment. Temperature had a significant effect on cyanobacterial growth in the greenhouse experiment and enriched nutrient levels had a significant effect on culture growth in the river experiment.

DISCUSSION

Overall, in these two experiments, both increased temperature and elevated nitrogen and phosphorus levels had a positive, significant effect on cyanobacterial growth. The effect of the temperature variable was shown in the greenhouse experiment, testing cultures with a natural inoculant in unaltered and enriched nutrient water. At the outset of culture growth, the higher temperature treatments distinguished themselves as having a greater cyanobacterial concentration. This was not only shown in the great increase of chlorophyll and phycocyanin levels, but also noticeable through a visible green hue. In the greenhouse experiment, temperatures above the critical 25ºC point

(Fernald et. al. 2007) made the cyanobacterial cultures grow faster and reach significantly higher concentrations. This was a clear relationship, as the greenhouse experiment used regular deionized water as the base for the cultures, with only a small amount of other

I-16 phytoplankton and zooplankton to interact with the cyanobacteria. In the river experiment, the average water temperature measured at the Hudson River Environmental

Conditions Observing System (HRECOS) Norrie Point hydrological measurement station was 26.84ºC. This value is clearly above the 25ºC point, and while not tested as a variable, could definitely have helped all of the cultures to produce growth over the two week period. This late summer hot temperature is also the cause of more frequent natural cyanobacterial blooms during this time, and could provide an explanation for the substantial amount of cyanobacteria in the river which allowed the cultures to grow without an external inoculant. In summary, this experiment reinforced the findings by

Fernald et al. (2007); the cultures grown in temperatures above 25ºC had significantly higher concentrations than the cultures grown at approximately 20ºC.

In the experiment that took place in the river, the doubled levels of nitrogen and phosphorus had a significant, increasing effect on cyanobacterial concentration. For cultures growing both a meter under the water and at the surface, the enriched cultures were able to better reach and maintain higher cyanobacterial concentrations. Similar to the greenhouse experiment, the enriched cultures in the river experiment showed qualitative and quantitative signs of higher concentrations, early on in the experiment.

The enriched cultures had a more intense green hue and ultimately reached significantly higher peak concentrations than their unaltered counterparts. In the river experiment, it is important to take into consideration that the cultures contained large portions of river water, capturing ambient communities of bacteria, algae, and zooplankton. These other organisms provided competition for the cyanobacteria in terms of gathering nitrogen and phosphorus to grow. An explanation then, for the success of cyanobacterial cultures in

I-17 enriched conditions, could be the high nutrient affinity of cyanobacteria, and their ability to outcompete other organisms and take full advantage of excess nutrients. The positive effect of nitrogen and phosphorus on cultures was also slightly conveyed in the greenhouse experiment. While the elevated nutrient levels were shown to not have a significant effect on culture concentration, almost all of the enriched treatment concentrations were slightly higher than the unaltered treatments. Furthermore, both of the enriched culture treatments had higher peak concentrations, with the enriched hot culture treatment being greater than the hot unaltered culture treatment. In conclusion, when nitrogen and phosphorus were added to Hudson River water, cyanobacteria were able to take advantage of the environment and grow to significantly higher concentrations than unaltered cultures.

In the river experiment, the depth variable modified various growth conditions of cyanobacteria, most notably light level. Sunlight level changes from approximately

3265.89 µE/m2/s at the surface the water, to approximately 914.45 µE/m2/s at one meter below the surface. Overall, depth did not have a significant effect on culture concentration in the river experiment; however, the bottom cultures exhibited high variance, but generally were able to reach and maintain slightly higher cyanobacterial concentrations than the surface cultures. This is especially conveyed in the relationship between the surface and bottom elevated nutrient cultures. The surface culture quickly peaked before the bottom, but then decreased in concentration. In contrast, the bottom culture, showed more steady growth and reached higher concentrations towards the end of the experiment. When analyzing this relationship, the sunlight level is one of the most important to take into consideration, with the surface cultures receiving 75%-50% of

I-18 maximum sunlight and submerged cultures receiving 25%-10% of sunlight. Over the course of the two week experiment, the average sunlight level measured by the adjacent

Norrie Point meteorological HRECOS station during the peak of the day was 5225.429

µE/m2/s. A possible explanation for the relationship of the surface and bottom nutrient treatments could be that the surface treatment grew very quickly due to the large amount of sunlight, but then started to die off because the sunlight was too intense. This would fit the characteristic of cyanobacteria as being sensitive to intense levels of sunlight. In contrast, the bottom culture would receive a lower but sufficient amount of sunlight, allowing the cultures to grow at a steadier rate.

The results of elevated temperature and nutrients increasing cyanobacterial growth become significant when put into the context of the actual conditions of the

Hudson River. From the greenhouse experiment, it was determined that temperatures above 25ºC significantly increased cyanobacterial concentration. This immediately identifies a time window in the Hudson River in which the temperature is above 25ºC, and therefore more prone to cyanobacterial blooms. The average river temperature during the experiment at Norrie Point was approximately 27ºC, corroborating the theory that the late summer has ideal temperature formation conditions for cyanobacterial blooms.

Furthermore, the water temperature at Norrie Point exceeded 25ºC from July 2 through

September 10, showing that there is a large time window for optimal cyanobacterial bloom formation conditions. This temperature threshold also identifies possible areas that are more susceptible to blooms. These would be slow moving, open sections of the river that can be easily heated by the sun for longer periods of time, allowing the temperature to rise. In addition, this experiment conveyed the importance of the effect that

I-19 temperature has on cyanobacterial growth; increased temperature can cause just a small amount of pond water grow to a high level status bloom concentration as classified by the

New York State DEC. At a concentration this high, harmful cyanobacteria would produce enough toxins for the culture to be harmful to human health (NYSDEC 2018). In conclusion, temperature is a vital formation condition for cyanobacteria, with optimal temperature conditions occurring during the late summer in the Hudson River.

The experiment that took place within the Hudson River attempted to replicate, as closely as possible, conditions for growing cyanobacteria without actually releasing cyanobacterial blooms into the river. Thus, the effects of the variables in the river experiment on cyanobacterial growth would be very similar to the conditions of the actual Hudson River. In terms of the elevated nitrogen and phosphorus levels, this means that the addition of nutrients into the Hudson River could easily contribute to cyanobacterial growth.

In conclusion, the formation conditions of increased temperature as well as increased levels of nitrogen and phosphorus were shown to have a significant positive effect on cyanobacterial concentration. Under these conditions, cyanobacterial cultures are able to reach concentrations high enough to be considered dangerous to human health, within just a week of growth. In the Hudson River, these conditions would take the form of a slow moving, open section of river, downstream of a nutrient polluting source. In the warm waters of the late summer, an area like this would be susceptible to cyanobacterial blooms.

I-20 ACKNOWLEDGEMENTS

First and foremost, I would like to thank the Hudson River National Estuarine

Research Reserve, the Hudson River Foundation, and the Polgar Fellowship program for giving me this opportunity to conduct research. I would also like to thank Stuart Findlay for advising my research project. In addition, David Fischer, Heather Malcolm, and the

Cary Institute of Ecosystem Studies were very helpful in providing me assistance and resources with my research. Lastly, I would like to thank Sarah Fernald and the staff at the Norrie Point Environmental Center for their guidance and for allowing me use of their facilities.

I-21 REFERENCES

Carmichael, W. 1994. The toxins of cyanobacteria. Scientific American, 270: 78- 86.

Centers for Disease Control (CDC). 2017. “Cyanobacteria (blue-green algae) Blooms: When in doubt, it’s best to keep out!” Cyanobacteria Blooms FAQs. https://www.cdc.gov/habs/pdf/cyanobacteri_faq.pdf (accessed September 18, 2017).

Davis, T.W., G.S. Bullerjahn, T. Tuttle, R.M. McKay, , and S.B. Watson. 2015. Effects of increasing nitrogen and phosphorus concentrations on phytoplankton community growth and toxicity during Planktothrix Blooms in Sandusky Bay, Lake Erie. Environmental Science & Technology 49: 7197-7207.

Environmental Protection Agency (EPA). 2017. “Toxicology of Cyanobacteria.” https://www.epa.gov/waterresearch/toxicology-cyanobacteria (accessed September 18, 2017).

Fernald, S.H., N. Caraco, and 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.

GLISA. 2018. “Algal Blooms.” http://glisa.umich.edu/climate/algal-blooms (accessed October 22, 2018).

Hitzfeld, B.C., S.J. Höger, and D.R. Dietrich. 2000. Cyanobacterial toxins: removal during drinking water treatment, and human risk assessment. Environmental Health Perspectives 108(suppl 1): 113-122. Hudnell, H.K., editor. 2008. Cyanobacterial harmful algal blooms: state of the science and research needs. Advances in Experimental Medicine and Biology, Vol. 619. Springer Science & Business Media, LLC. Riverkeeper. 2016. Confirmed: Wallkill River Algae Produced High Toxin Levels. https://www.riverkeeper.org/newsevents/news/waterquality/confirmedwallkill river-algae-produced-hightoxinlevels/ (accessed September 18, 2017).

Lopez, C.B., E.B. Jewett, Q. Dortch, B.T. Walton, and H.K. Hudnell. 2008. “Scientific Assessment of Freshwater Harmful Algal Blooms”. Interagency Working Group on Harmful Algal Blooms, Hypoxia, and Human Health of the Joint Subcommittee on Ocean Science and Technology. https://www.whoi.edu/fileserver.do?id=41023&pt=10&p=19132 (accessed October 10, 2018).

I-22 New York State Department of Environmental Conservation (NYSDEC). 2018. “Harmful Algal Blooms (HABS) Program Guide.” http://www.dec.ny.gov/docs/water_pdf/habprogramguide.pdf (accessed April 23, 2018).

Tango, P., and W. Butler. 2008. Cyanotoxins in tidal waters of Chesapeake Bay. Northeastern Naturalist 15: 403-416.

Taranu, Z.E., I. Gregory-Eaves, P.R. Leavitt, L. Bunting, T. Buchaca, J. Catalan, I. Domaizon, P. Guilizzoni, A. Lami, S. McGowan, H. Moorhouse, G. Morabito, F. R. Pick, M.A. Stevenson, P.L. Thompson, and R.D. Vinebrooke. 2015. Acceleration of cyanobacterial dominance in north temperate-subarctic lakes during the Anthropocene. Ecology Letters 18: 375-384. World Health Organization (WHO). 1999a. “Cyanobacteria in the Environment.” http://www.who.int/water_sanitation_healthresourcesquality/toxcyanchap2.pdf (accessed April 23, 2018).

World Health Organization (WHO). 1999b. “Determination of Cyanobacteria in the Laboratory.” http://www.who.int/water_sanitation_healthresourcesquality/toxcyachap12.pdf (accessed April 23, 2018).

Yéprémian, C., A. Catherine, C. Bernard, R. Congestri, T. Elersek, and R. Pilkaityte. 2017. Chlorophyll a Extraction and Determination. pp. 331-334 in J. Meriluoto, L. Spoof and G.A. Codd (Eds.). Handbook of Cyanobacterial Monitoring and Cyanotoxin Analysis. https://onlinelibrary.wiley.com/doi/pdf/10.1002/9781119068761. (accessed October 25, 2018).

I-23

Quantification of Microplastic Content in Surface Water and Sediment within Hudson River Tributaries and Marshes

A Final Report of the Tibor T. Polgar Fellowship Program

Jason Randall

Polgar Fellow

School of Science, Department of Environmental Science Marist College Poughkeepsie, NY 12601

Project Advisors:

Dr. Zofia Gagnon School of Science, Department of Environmental Science Marist College Poughkeepsie, NY 12601 [email protected]

Chris Bowser New York State Department of Environmental Conservation Hudson River National Estuary Research Reserve [email protected]

Randall, J.D., Z.E. Gagnon, and C. Bowser. 2020. Quantification of Microplastic Content in Surface Water and Sediment within Hudson River Tributaries and Marshes. Section II:1-33 pp. In S.H. Fernald, D.J. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

II-1

ABSTRACT

The infiltration of microplastic (MP) pollution into aquatic ecosystems has been discovered in watersheds across the world. MPs have a variety of different origins, compositions, densities, shapes, and a high resistance to degradation. These properties allow for accumulation of MPs in major watersheds and make mitigation extremely difficult. Freshwater marshes are known to be a sink for sediment and certain pollutants, however, the role marshes play in MP accumulation is unknown. The main purpose of this study was to quantify and investigate differences in MP concentrations between marsh, tributary, and open water locations in the Hudson River watershed. The following four Hudson River marshes and their corresponding tributaries were sampled to investigate differences in MP content: Sleightsburg Marsh and Rondout Creek, Tivoli

Bays and Saw Kill Creek, Fishkill Marsh and Fishkill Creek, and Constitution Marsh and

Indian Brook. Surface water and sediment samples were analyzed from each sampling location to account for density differences in polymers present. The findings of this study indicate that Hudson River marshes act as a sink for MP pollution. Marsh samples contained a significantly higher average concentration of MPs in sediment (2.28 MPs/g

MS) than tributary (0.27MPs/g MS) or open water samples (0.42 MPs/g MS) and the highest average surface water concentration (0.30 MPs/L) of all location types. Fishkill

Creek and Sleightsburg Marsh contained the highest average MP concentrations between tributary and marsh sites, respectively. While development of mitigation strategies to eliminate microplastics from bodies of water has been proven nearly impossible without harming the ecosystem, the findings of this study highlight a need for elimination of microplastic use at the consumer level and prevention of infiltration at a regulatory level.

II-2

TABLE OF CONTENTS

Abstract ...... II-2

Table of Contents ...... II-3

List of Tables and Figures...... II-4

Introduction ...... II-5

Methods...... II-8

Results ...... II-20

Discussion ...... II-28

Acknowledgements ...... II-31

References ...... II-32

II-3

LIST OF TABLES AND FIGURES

Figure 1. Map of marsh locations of target sites ...... II-12

Figure 2. Surface water microplastic contents by sample type ...... II-20

Figure 3. Sediment microplastic contents by sample type ...... II-21

Figure 4. Map of target watersheds and SPDES locations ...... II-23

Figure 5. Tributary microplastic contents by location ...... II-24

Figure 6. Marsh microplastic contents by location ...... II-25

Figure 7. Open water microplastic contents by location ...... II-25

Figure 8. FTIR spectra for two identified polymer particles ...... II-27

Table 1. Tributary site characteristics ...... II-22

Table 2. Marsh site characteristics ...... II-22

Table 3. Polymers identified, abundance, and densities ...... II-26

II-4

INTRODUCTION

Microplastics (MPs) are composed of a variety of synthetic polymers and range between 0.1 µm and 5.0 mm in length. MPs can enter waterways as macroplastic waste or litter that overtime continue to fragment from the macroplastic to nanoplastic level and persist in aquatic ecosystems (Espinosa et al. 2016). The origin of these particles can vary greatly from fragmentation of macroplastic waste, such as water bottles to fragments of clothing composed of synthetic fibers, or to modern prefragmented MPs, such as microbeads found in facial cleansers and toothpastes. The New York State Attorney

General’s Office (2015) points out that manufactured microbeads and synthetic fibers are entering waterways through water treatment plants across the country that filter water through screens of 6.0 mm or larger in some cases and 1.5-6.0 mm in other cases. This creates a great concern due to the easy access of industrially engineered MPs into nearly all waterways.

Once introduced into aquatic ecosystems, MPs can have profound ecological impacts. MPs have been found to have major negative health effects for marine vertebrates and invertebrates, mainly by bioaccumulation in tissue. Ingestion of MPs by aquatic organisms including zooplankton, mollusks, crustaceans, aquatic worms, and vertebrates, such as fish, has been identified globally (Cole et al. 2013). Terrestrial and aquatic birds and mammals have also been found to contain MPs in their tissue and macroplastics in stomachs. MP ingestion in organisms can occur directly, through filter

feeding or misidentification of plastics as food, and indirectly, through predation of organisms already containing plastics. The bioaccumulation of MPs in tissues of

organisms allows these plastics to move through the food web and have larger effects on

II-5 species near the top. Fish, in particular, are highly affected by the toxicity of synthetic polymers which can damage or interfere with their endocrine, circulatory, immune, and muscular systems, as well as their livers, gills, gonads, and intestines (Espinosa et al.

2016). At the top of this food web, humans consume a wide variety of organisms worldwide, both aquatic and terrestrial, with the possibility of having MPs accumulated in their tissue. The blue crab (Callinectes sapidus) may be the best example of this in the

Hudson River, as they are they are one of the most commonly caught and eaten species, found anywhere between the Troy Dam to New York Harbor (NYSDEC 2019). Striped bass (Morone saxatilis) is another example, while less commonly consumed when caught in the Hudson River, they spawn in the Hudson River each year and offspring can live up to two years in the river before entering the Atlantic Ocean (Hill et al. 1989).

The accumulation of the synthetic polymers is not the only point of concern for human health and the health of all other organisms. Many plastics are composed of toxic derivatives and are manufactured using potentially harmful chemicals. Certain polymers also have the ability to absorb ambient chemicals from the environment such as PAHs,

DDT, chlordanes, and PCBs, which are known to persist in the Hudson River (Van et al.

2012). A large portion of these chemical groups have known toxicity to humans and other organisms, and MPs provide access into tissues of organisms causing potentially major health concerns (Espinosa et al. 2016). Mammals such as mice have been the focus of several new studies concerning MP polymer toxicity. Polystyrene microbeads of varying size have been identified to accumulate in the liver, kidneys, and gut of mice. This accumulation and exposure has led to impairment of energy transformation, disturbance of lipid metabolism, reduction of neurotransmission efficiency, and imbalances in

II-6

antioxidant defense systems (Deng et al. 2017). As the effects of MPs on human health are relatively unknown, studies using animal models can give a clearer perspective into

the possible effects of MP accumulation in humans.

Buoyant MPs can flow more easily throughout waterways and denser MPs that

have a greater ability to sink and accumulate in benthic sediment. Many current studies

on MP accumulation in aquatic environments focus on types of polymers that are less

dense and are easier to sample and identify; however, a study conducted by Hidalgo-Ruz

et al. (2012) shows that the majority of researched MP polymers have average densities

greater than 1.0 g/cm3, meaning they will sink in fresh water. This shows that there is a

need for investigation of certain plastic polymers below the surface water. Many of these

polymers have densities that range just above 1.0 g/cm3 which would easily allow them

to be transported by water current. Sediment analysis will allow for the identification of

the importance of marshes and tributaries on MP accumulation and transport in the

Hudson River.

This project investigated MP content differences in surface water and sediment samples from tributaries, marshes, and adjacent open water of four target sites. MP content was compared between sites (four target marshes) or location types (tributary, marsh, open water) of the same sample type (surface water, sediment). Marshes are hypothesized to be a sink for MP pollution and have the highest MP contents in surface water and sediment. High density polymers are also hypothesized to be identified in marsh sediment, but not in tributary sediment. The hypothesis was tested that the largest

watersheds and highest number of pollution discharge sites within the watershed would

II-7

have the highest MP content in both surface water and sediment regardless of location

type.

METHODS

Site Selection

The following four marshes with corresponding tributaries were chosen for

sampling based on location, ease of access, differences in land usage, ecological importance, and size of watershed. Figure 1 displays marsh locations of each site investigated in this study.

Tivoli Bays: Located in Dutchess County, NY, with Saw Kill Creek as the target

tributary, Tivoli Bays was the largest marsh investigated in this study, spanning two

miles along the east side of the river. It is also one of the more rural sites as there is little

immediate urbanized land surrounding the marsh; however, the town of Tivoli is located

northeast of the marsh and Bard College is southeast of the marsh. This site has a very

high ecological importance and is home to many bird species including the bald eagle

(Haliaeetus leucocephalus). This led to the site’s designation as a New York State

Important Bird Area and a New York Bird Conservation Area (NYSDEC 2018b).

Sampling in Tivoli Bays marsh took place in Tivoli North Bay, due to inadequate

sampling conditions in Tivoli South Bay, in which Saw Kill Creek discharges. Tivoli

North Bay marsh sampling took place at the end of Cruger Island Road in Tivoli, NY

(40o 00.527 N, 73o 54.527 W). The mouth of Saw Kill Creek is located on the southern

end of the marsh and stretches through the towns and villages of Red Hook, Milan,

Tivoli, Rhinebeck, and Annandale-on-Hudson. The Saw Kill Creek watershed includes

II-8

22 square miles of land which is primarily agricultural and residential (Saw Kill

Watershed Community 2016). The sampling location for Saw Kill Creek was behind the

parking lot of Bard College’s Shafer House (42o 00.832 N, 73o 54.527 W). Open water

sampling took place by boat about 20 m west of the train tracks that run north-south through the marsh, adjacent to the southern portion of the marsh (42o 01.125 N, 73o

55.689 W).

Fishkill Marsh: Fishkill Marsh, located in Beacon, NY, was the smallest marsh in

this study; however, Fishkill Creek, Fishkill Marsh’s main tributary source, has a very

large watershed, encompassing about 193 square miles of land in both Dutchess and

Putnam Counties. Although the city of Beacon, located directly north of this marsh, is

subjecting the marsh to industrial and commercial land use consequences, very little

urbanization in the form of residential neighborhoods has occurred to the south and east

of the marsh, which is dominated by forests. Other major land uses within the watershed

include recreational and agricultural use, making it the most diverse site (Burns et al.

2005). Sampling in Fishkill Marsh took place in front of a small wooden dock off of a

trail to the west of Madam Brett Park in Beacon, NY (41o 29.215 N, 73o 58.683 W).

Fishkill Creek Sampling took place on the north side of a small division of the creek within 100 meters of the parking lot of Madam Brett Park (41o 29.339 N, 73o 58.430 W).

Due to shallow depths surrounding the mouth of the marsh, open water sampling took

place approximately 300 meters southwest of the mouth in river accessed by boat (41o

28.810 N, 73o 59.473 W).

Sleightsburg Marsh: Located in Ulster County, NY, Sleightsburg Marsh is found

on the southern side of Rondout Creek’s mouth. Sampling at Sleightsburg Marsh took

II-9

place between the eastern end of Sleightsburg Park and a small island to its east (41o

55.216 N, 73o 58.257 W). The area around Sleightsburg Marsh is the most urbanized of

the sites chosen with the City of Kingston adjacent to the marsh and mouth of the tributary. The major land uses are commercial and industrial making it the most heavily human impacted site. Rondout Creek is the largest tributary that was analyzed in this study and its watershed includes the Wallkill River which joins Rondout Creek before entering the Hudson River (Melendez et al. 2010). The combined Rondout-Wallkill watershed accounts for about 1,190 square miles of land in Ulster, Sullivan, Orange, and

Sussex Counties. In order to obtain ample water flow and depth, sampling in Rondout

Creek occurred upstream of the connection to Wallkill River, on the northwest side of the creek, about 200 meters southwest of the route 213 bridge in Rosendale, NY (41o 50.790

N, 74o 4.397 W). Open water sampling was performed by boat approximately 100 meters southeast of the small inlet utilized for Sleightsburg Marsh sampling (41o 55.169 N, 73o

57.808 W).

Constitution Marsh: Constitution Marsh in Putnam County is over 270 acres in

area with high ecological importance similar to that of Tivoli Bays mainly due to the rich

bird habitat it provides. It is home to the Constitution Marsh Audubon Center and

Sanctuary including an education center. Sampling in Constitution Marsh took place on

the northeast end of the boardwalk located northwest of the Audubon Center in Cold

Spring, NY (41o 24.210 N, 73o 56.419 W). The surroundings of the marsh to the south

and east are similarly rural to the brook, while the small town of Cold Spring to the north

is the nearest urban area along with railroad tracks that pass through the marsh (NYSDEC

2018a). Indian Brook, the corresponding tributary, is the smallest of the tributaries that

II-10

was investigated and is surrounded mainly by forests with some residential homes and

agricultural fields. Sampling in Indian Brook occurred off of Indian Brook Road about 50 meters east of the intersection with the Bear Mountain-Beacon Highway bridge, adjacent to a small abandoned brick building and a small walkway overpass in Garrison, NY (41o

50.770 N 74o 4.401 W). Open water samples were taken by boat about 20 meters west of

the train tracks that run north-south through the marsh and directly west of the mouth to

Indian Brook (41o 23.808 N, 73o 56.688 W).

II-11

Figure 1. Map of marsh locations of target sites: Marsh locations of four major target sites displaying nearby towns and cities with populations greater than 18,000.

II-12

Sampling Set-up

Each of the tributary and marsh sampling sites were visited prior to sampling to ensure proper accessibility and water velocity. Tributary sampling took place far enough upstream in regions where the tides had no effect on the water velocity. Marsh sampling occurred in accessible areas between one and three hours after the most recent high tide to ensure ample water velocity downstream towards the river. All marsh and tributary sampling was performed twice on separate dates between 6/8/18 – 6/26/18 to allow for variations in weather conditions and water velocity.

Open water sampling took place aboard the Marist School of Science research vessel, which is a 28-foot pontoon boat with a square hatch in the bottom. Due to limitations with boat access, open water samples were only collected on one occasion for each site between 5/29/18 – 5/31/18 between the most recent high tide and the following low tide.

Sediment samples for each location were extracted within one meter of the surface water samples. For sampling consistency, all sediment samples in tributaries were also taken within one meter of the nearest stream bank. All sediment samples in marshes were also extracted at an exact location in which no water is present at low tide in order to investigate accumulation of MPs in the sediment. Open water sediment samples were taken directly below the location of the surface water sampling.

Site Characteristics

GPS coordinates of each sampling location were measured and recorded using a

Garmin Oregon 550 portable GPS system. Each sampling location was characterized by surrounding land usage, major sources of pollution within each watershed, and visible

II-13

pollution in relation to the other target sites. Each site was given a ranking of 1-4 based

on location type (tributary, marsh, open water) for visible pollution with 1 being most

visibly polluted and 4 being the least polluted. Due to time limitations in relation to tides,

water quality data could not be obtained as tributary and marsh sampling for the two

northern most sites and the two southern most sites occurred on the same days to limit the

use of fossil fuels in travel. As this study focuses on MP content at sampling locations,

these factors were not believed to have a significant impact on the variance of findings.

Surface Water Sample Collection

For tributary and marsh sampling, an Aquatic Research Instruments Inc. 

Standard Stream Drift Net (100 cm x 45 cm x 25 cm) with 333 µm mesh and detachable cod end was placed in the stream or marsh with the mouth of the net facing upstream.

The net was secured by two metal poles and situated so that the top of the net was 2 cm or more out of the water and the bottom of the net was not in contact with the stream or marsh bed. The distance from the surface of the water to the top of the net was measured for each sample for later calculations of discharge. Open water sampling took place through a metal hatch in the bottom of the anchored vessel. The same drift net was deployed through the hatch and secured with two ropes so that the top of the net was 3 cm out of the water. The vessel was positioned so that the opening of the net was facing toward the marsh and surface water was flowing into the net.

For each sample the net was set up for 15 minutes of sampling before being removed from the water. Once removed from the water, the net was rinsed with deionized water towards the cod end and the resulting debris in the cod end was scraped out with a spatula and rinsed out into a labelled glass container for later analysis. Water velocity

II-14

was determined using a Swoffer Research Instruments Current Velocity Meter (Model

2100) placed directly in front of the middle of the opening of the net. Current velocity

readings were taken four times during sampling at minutes 0, 5, 10, and 15. These values

were recorded in cm/sec and were averaged to account for abnormal flow conditions at

any point during sampling and later used to calculate discharge of water through the net

along with the net dimensions. A reading was recorded once the value had stabilized for

at least 10 seconds. For open water sampling, waves and wakes created by other vessels

may have introduced errors to the water velocity readings and discharge through the net.

Sediment Sample Collection

Sediment samples were extracted during the 15-minute surface water sampling.

Open water sediment samples were obtained using an AMS  Ekman Dredge (15cm x

15cm x 20cm). The dredge was deployed off the side of the vessel and the collected

sample was placed into a metal tray and transferred into a labelled 1 L glass container

using a spatula and deionized water for later analysis. The dredge was deployed multiple

times for certain samples that were not large enough to fill the container, ensuring a large

sample volume. For tributary and marsh sediment samples, the dredge was also utilized;

however, due to the amount of larger pebbles and stones in many samples, the dredge

was ineffective and a shovel was used instead following the same protocol.

Sediment Density Separation

Once all sample collection was completed, benthic samples were wet sieved through a stacked arrangement of sieves with openings between 4.75 mm and 0.025 mm using deionized water. Sieved material larger than 4.75 mm and smaller than 0.025 mm was removed and the collected material within the desired size range was placed in a 600

II-15

mL beaker which was then placed in a drying oven at 80oC for at least 24 hours. Sieving

and drying were performed prior to weighing to normalize the samples to contain only

dry particles of between 0.025-4.75 mm. Once dried, samples were weighed and those which contained large quantities of organic material were broken down to a smaller weight, noting the weight proportion to the original dried and sieved sample. These final samples then underwent density separation using a sediment-microplastic isolation (SMI) unit (Coppock et al. 2017). This was constructed using two 1’ long pieces of 3” diameter

PVC tubing connected by a PVC ball valve all secured using PVC glue and sealed on one end using a 3” diameter PVC coupling. The SMI unit was filled with a zinc chloride

3 solution (515g ZnCl2 per 1L H20) with a density of 1.5 g/cm to about 3-5 cm over the

ball valve (approximately 1.5 L). A magnetic stirring bar was added to the bottom of the

SMI unit and each sample was added prior to the addition of the zinc chloride solution.

The unit was then placed on a magnetic stirring plate at a constant speed and was left for

five minutes. The apparatus was then taken off the plate and left to settle for two minutes

before being placed back on the plate for three consecutive 10 second stirs with 10

second pauses. After the final stir, the unit was left to settle for another 2 minutes. Once

completed, the ball valve was closed and the resulting material and solution above the

valve was poured and rinsed with the same zinc chloride solution into a 0.025 – 4.75 mm

sieve with a 1 L beaker beneath it. The material within the sieve was transported into a

labelled 600 ml beaker and placed in a drying oven at 80oC for at least 24 hours. The

remaining debris and solution in the SMI unit was poured through the cleaned 0.025 mm

sieve into the same 1 L beaker where the solution could be reused as the process repeated

for each benthic sample.

II-16

Wet Peroxide Oxidation

A modified NOAA Marine Debris Procedure for MP separation was followed to

analyze water samples and sediment samples after density separation had been performed

(Masura et al. 2015). All surface water samples were wet sieved with deionized water through the same 4.75 mm and 0.025 mm sieves as the benthic samples, discarding all material larger and smaller than this intended range. Samples were then placed in a drying oven at 80oC for at least 24 hours. Once dried, certain samples containing large amounts of organics were weighed and only a portion of the sample was further analyzed with the weight proportion recorded. After this point all surface water and benthic

samples were processed using the same methodology. All dried samples then underwent

wet peroxide oxidation. A magnetic stirring bar was added to each 600mL beaker

containing a specific sample. 20 mL of an aqueous 0.05M Fe(II) solution (7.5 g FeSO4 ∗

7H2O and 3 mL concentrated H2SO4 per 500 mL of deionized water) and 20 mL of 30%

 hydrogen peroxide (H2O2) (Thermo Fisher Scientific ) were then added to each beaker

at room temperature and left for 5 minutes. The beaker was then placed on a hot plate,

heated to 75oC and stirred at medium speed for 30 minutes. If gas bubbles appeared, the

beaker was removed from the hot plate until boiling had stopped and deionized water was

added if boiling was vigorous. If organic material was still visible after 30 minutes, the

process was repeated for a total of 80 mL of solution per sample. Once little or no

organics were visible, the final mixture was sieved, rinsed with deionized water, and

transported to a labelled 50 mL beaker. These final samples were dried for another 24

hours in a drying oven at 80oC.

II-17

Microscope Separation and Quantification

Final samples were scraped out the 50mL beakers using a spatula onto petri

dishes that were examined under a dissecting microscope at 10.5x – 40x magnification.

Samples were spread out across the petri dishes using narrow-tipped forceps and

methodically examined for MPs. Plastics were counted using a hand tally counter for the

entire dish and the leftover material in the 50 mL beaker. Each sample was counted twice

and the final counts were averaged. Larger plastics and unknown pieces of particulate

matter suitable in size for use in infrared spectrometry were removed and placed in

labelled glass vials.

Fourier Transform Infrared Spectrometry

Certain particulate matter was chosen for examination using Fourier Transform

Infrared Spectroscopy (FTIR) to verify that the particle is composed of one or more

synthetic polymers and to identify the polymer types. FTIR spectrums were created using a NicoletTM iSTM 5-FT-IR Spectroscope that utilizes a high attenuated transverse

reflection (ATR) unit with a ZnSe crystal. Each particle was placed on the plate and

scanned over the range of 500 cm-1 – 4000 cm-1, producing unique spectra. These were automatically compared to spectra of known compounds using Nicolet Software and the

Marist library database which included the Hummel Polymer Samples library and the

Nicolet Sampler library along with others, supplying the highest percentage matches to known compounds. Most synthetic fibers found in this study were not large enough to be identified using FTIR; however, certain fiber bundles were examined.

II-18

Data Analysis

MP content in surface water samples were calculated and represented in terms of

number per liter of water discharged through the net for each sample. Discharge was

calculated using the following equation:

( ) ( ) ( ) ( ) = ( ) 𝑐𝑐𝑐𝑐 𝑁𝑁𝑁𝑁𝑁𝑁 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ 𝑐𝑐𝑐𝑐 ∗ 𝑁𝑁𝑁𝑁𝑁𝑁 𝐻𝐻𝐻𝐻𝐻𝐻𝐻𝐻ℎ𝑡𝑡 𝑐𝑐𝑐𝑐 ∗ 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑚𝑚𝑚𝑚 𝑠𝑠𝑠𝑠𝑠𝑠 Discharge was then converted from milliliters𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 to liters𝑆𝑆𝑆𝑆𝑆𝑆 of𝑆𝑆 𝑆𝑆𝑆𝑆𝑆𝑆water.𝑠𝑠𝑠𝑠 The𝑐𝑐 average number of

MPs was then divided by the calculated discharge in liters. If a sample volume was

decreased due to removal of abundant organics, the corresponding discharge value was

adjusted by the same ratio. MP content in benthic samples was calculated and represented

in terms of number of MPs per gram of dried sediment between 0.025 – 4.75 mm in

length, or microsediment (MS). Average MP counts for each sample were divided by the

mass of the original dried and sieved sample.

Statistical Analysis

One-way analysis of variance (ANOVA) was performed using the SPSS v25.0

Statistical Package (2017) to test for differences in all comparable variables. Statistical

analysis could not be performed for open water samples as only a single value of MP

content was determined for surface water and sediment samples. A Student-Newman-

Keuls (SNK) multiple-comparison test was used to determine differences among means at a probability level of p < 0.05.

II-19

RESULTS

Microplastic Content by Location Type

Average MP concentrations in surface water were not found to be significantly

different based on location type; however, marshes contained the highest MP content

(0.30 MPs/ L) while open water samples exhibited lower average concentrations (0.03

MPs/ L) (Figure 2).

Figure 2. Surface water microplastics by sample type: Content of MPs (MPs/L) in surface water samples. The bar graphs represent MP content means ± SD of three measurements. Bar graphs with different letters (a and b) are significantly different at α ≤ 0.05 determined by SNK multiple comparison test.

Marsh samples were found to have significantly higher average MP concentrations in sediment (2.28 MPs/g MS) than tributaries or open water samples

(p=0.05). Tributary and open water sediment samples contained average MP concentrations of 0.42 MPs/ g MS and 0.28 MPs/ g MS, respectively (Figure 3).

II-20

Figure 3. Sediment microplastic contents by sample type: Content of MPs (MPs/g MS) in sediment samples. The bar graphs represent MP content means ± SD of three measurements. Bar graphs with different letters (a and b) are significantly different at α ≤ 0.05 determined by SNK multiple comparison test.

Of the three major MP classifications, fibers were more abundant than either fragments or microbeads. Microbeads were the least abundant MP class examined in this study with only three microbeads identified from all samples. While less abundant than fibers, isolated fragments appeared in a variety of shapes, colors, and sizes.

Site Characterization Site characterization performed during multiple visits to each site resulted in

expectations that Rondout Creek and Sleightsburg Marsh would have the highest MP

concentrations in each sampling category. The Rondout-Wallkill watershed encompasses

56 State Pollutant Discharge Elimination System (SPDES) outfalls, the greatest amount

of all watersheds investigated. Both tributary and marsh location types for the Rondout

watershed had the highest level of visible pollution of all sites. The Fishkill Creek II-21 watershed was the only other watershed to contain SPDES outfalls, with 15 within the entirety of the watershed. Neither Saw Kill Creek nor Indian Brook watersheds contained

SPDES outfalls; they had the least visible pollution in both tributary and marsh sites

(Tables 1 and 2).

Tributary Surrounding Land Visible Number of Watershed Use Type Pollution (1- SPDES Size (mi2) 4) Outfalls in Watershed Saw Kill Residential/Commercial 1 0 22 Creek Indian Brook Residential/Recreational 2 0 5 Rondout Commercial/Residential 4 56 1190 Creek Fishkill Creek Commercial/Industrial 3 15 193

Table 1. Tributary site characteristics: Characteristics of surrounding land area and watershed of tributary sampling sites.

Marsh Surrounding Land Visible Number of Watershed Use Type Pollution (1-4) SPDES Size (mi2) Outfalls in Watershed Tivoli Bays Preserved/Recreational 2 0 22 Constitution Preserved/Recreational 1 0 5 Marsh Sleightsburg Commercial/Industrial 4 56 1190 Marsh Fishkill Industrial/Recreational 3 15 193 Marsh

Table 2. Marsh site characteristics: Characteristics of surrounding land area and watershed of marsh sample sites.

II-22

Figure 4. Map of target watersheds and SPDES locations: Target watersheds and the State Pollutant Discharge Elimination System (SPDES) outfalls within each watershed.

II-23

Tributary Microplastic Content

Fishkill Marsh was found to have the highest average MP concentrations in both

sediment (0.45 MPs/g MS) and surface water (0.51 MPs/L) for tributary samples among

all other sites. Surface water MP content in Fishkill Creek was significantly greater than all other sites, while Rondout Creek samples (0.34 MPs/L) had a significantly higher average concentration than Saw Kill Creek and Indian Brook. Tributary sediment sample concentrations were highly variable within each site, especially Indian Brook which produced a standard error of 0.21 MPs/g MS (Figure 5).

Figure 5. Tributary microplastic contents by location: Content of MPs (MPs/L and MPs/g MS) in tributary samples. The bar graphs represent MP content means ± SD of three measurements. Bar graphs with different letters (a and b) are significantly different at α ≤ 0.05 determined by SNK multiple comparison test.

Marsh Microplastic Content

Sleightsburg Marsh samples contained significantly greater average MP

concentrations in surface water (0.66 MP/L) and sediment (4.13 MPs/g MS) than all

other sites. No significant differences were found between surface water or sediment contents between the remaining three sites (Figure 6).

II-24

Figure 6. Marsh microplastic contents by location: Content of MPs (MPs/L and MPs/g MS) in marsh samples. The bar graphs represent MP content means ± SD of three measurements. Bar graphs with different letters (a and b) are significantly different at α ≤ 0.05 determined by SNK multiple comparison test.

Open Water Microplastic Content

Low variability was seen among open water surface water and sediment samples between sites. The highest MP content discovered in surface water samples was 0.054

MPs/L (Fishkill Marsh) and 0.56 MPs/g MS in sediment samples (Rondout Marsh)

(Figure 7). Standard error could not be calculated due to low sample size (n=1).

Figure 7. Open water microplastic contents by location: Content of MPs (MPs/L and MPs/g MS) in open water samples based on adjacent marsh.

II-25

Infrared Spectroscopy Of the 17 particles designated for identification through infrared spectroscopy, 16 particles resulted in matches greater than 50% to known polymers in the library database.

11 of these matches were found to be greater than 70% to their respective polymer. Of the 16 matches, nine were found in sediment and seven were found in surface water. Six different polymers were identified in this study with varying densities from 0.95-1.44 g/cm3 (Table 3). Two spectra derived from examination of a particle from Rondout Creek tributary surface water and Constitution Marsh sediment are shown in Figure 8.

Polymer Density (g/cm3) Number Identified Number Identified in Surface Water in Sediment Polyvinyl Chloride 1.38 1 (M) 2 (M) Polyethylene 0.95 2 (T, T) 3 (M, M, OW) Poly(vinyl stearate) <1.0 1 (T) 1 (M) Polyethylene 1.38 0 1 (M) Terephthalate Polystyrene 1.04 2 (T,M) 2 (M,T) Cellophane 1.44 1 (OW) 0

Table 3. Polymers identified, abundance and densities: Abundance of different polymers types identified using infrared spectroscopy and their relative densities (M, T, and OW represent whether the particle was found in a marsh, tributary, or open water sample, respectively).

II-26

a

Fourier Fourier 76.43% match to 76.43% polystyrene. match

FTIR spectra for two identified polymer particles: particles: polymer identified for two spectra FTIR particle a A) of spectra Spectrometry Infrared Transform from surface 6/16/18sampled in water Rondout Creek that on and polyethylene B) to produced 86.17%oxidized match an sediment inMarshparticle Constitutionsampled from on displayed6/26/18 a that

. Figure 8

A. B.

II-27

DISCUSSION

Microplastic Content by Location Type

The findings of this study support the hypothesis that marshes may act as a sink

for MPs. The marshes in this study contained significantly more MPs in sediment, with

an average of 2.28 MPs/g MS, than either tributary or open water samples (Figure 3).

Marshes were also found to have the highest average MP content across surface water samples; however, not significantly greater than tributary or open water MP contents.

Surface water samples taken in open water were hypothesized to have the smallest MP

concentration which is supported by this study as surface water content in open water was

significantly lower than both marshes and tributary concentrations.

As only a small proportion of larger MPs were analyzed for polymer composition,

the proportion of high-density plastics in marshes cannot be determined; however, both

low-density polymers (polyethylene and poly(vinyl stearate)) and high-density polymers

(polyvinyl chloride, polystyrene, and polyethylene terephthalate) were identified in marsh

sediment. These results along with low MP concentrations in tributary sediment and the

presence of high-density polymers in tributary surface water indicate a likely mechanism

for the accumulation of high-density MPs in Hudson River marshes. As high-density

plastics are discharged into tributaries, they are less likely to settle in streams with a

constant velocity. Consequently, once these plastics travel downstream and reach the

corresponding marsh, the water becomes stagnant or near stagnant and may allow plastics

with a density greater than that of freshwater (1.0 g/cm3) to sink and accumulate in the benthic zone. While the mechanism for low density polymer deposition is also unknown, the tidal influence on the marshes may play a key role by allowing these plastics to settle

II-28

during low tide and potentially become trapped by natural organic and inorganic material.

Accumulation of MPs in salt marsh sediment has been identified in previous studies

(Khan and Prezant 2018). As freshwater has a lower density than saltwater, the problem

of accumulating MPs in sediment is amplified in freshwater mashes such as those in this

study.

Microplastic Content by Site

The results of this study support the hypothesis that Sleightsburg Marsh would

have the highest MP content in surface water and sediment. Sleightsburg Marsh

contained significantly higher average MP concentrations in both surface water and

sediment compared to the other three marshes. This can be explained by the large size of

the watershed (1190 mi2), the high number of SPDES outfalls found in the watershed,

and the large amount of visible pollution at the site.

The findings fail to support the hypothesis that Rondout Creek would have the

highest tributary MP concentrations in sediment or surface water. Fishkill Creek was

found to have significantly higher concentrations than all other sites in surface water and

the largest average MP concentration in sediment, although not significantly greater than

any other site. Rondout Creek tributary samples were expected to contain the greatest MP

content, largely due to the size of the Rondout-Wallkill watershed in comparison to the other target watersheds. A reason for the relatively low concentrations in comparison with Fishkill Creek could be the sampling location in Rondout Creek. Sampling in

Rondout Creek took place upstream of its confluence with the Wallkill River in order to

obtain proper water velocity, stream depth, and accessibility. By removing Wallkill

River’s contribution, the watershed sampled would only represent 405 mi2 of the total

II-29

1190mi2 watershed. This may account for the lower than expected MP concentrations in

Rondout Creek tributary sampling.

Open water MP concentrations displayed very low variability between sites. This

is likely due to the low sample size and the dilution of high microplastic concentrations in

low discharge tributaries entering the high discharge Hudson River.

Putting the Hudson River in Perspective

Increases in the number of MP quantification studies worldwide and in the United

States is allowing for comparisons of MP pollution between major watersheds around the

world. The findings of this study show that the main stem of the Hudson River has an

average surface water MP content of 0.03 MPs/L (Figure 2). A study which investigated

regions of the Ottawa River in both Quebec and Ontario, Canada with varying land uses

found an average of 0.10 MPs/L in surface samples near the shore, slightly higher than

the average of this study (Vermaire et al. 2017). A different study in the United States

examined MP content in surface water from 29 Great Lake tributaries which resulted in

average concentrations of 4.2 MPs/m3 or 0.0042 MPs/L in surface water (Baldwin et al.

2016). This finding is much lower than the average surface water MP content in the

Hudson River tributaries investigated in this study. As more MP quantification studies continue to emerge, researchers will be able to create a more cohesive picture of the watersheds that are most subjected to MP pollution.

II-30

ACKNOWLEDGEMENTS

I would like to thank the Tibor T. Polgar Fellowship Committee and the Hudson

River Foundation for funding and continued support throughout this project. A special thank you goes out to my advisors Dr. Zofia Gagnon and Chris Bowser for their unbelievable mentorship, guidance, and time commitment to myself and the project. I would like to extend my gratitude to Ted Fink for assistance in obtaining open water samples. I would also like to thank Matthew Badia, Miguel Madeira, Callan McLoughlin, and Carter Schuh for assistance during sampling and analysis. Lastly, I would like to acknowledge the Marist School of Science for use of resources and support.

II-31

REFERENCES

Baldwin, A.K., S.R. Corsi, and S.A. Mason. 2016. Plastic debris in 29 Great Lake tributaries: Relations to watershed attributes and hydrology. Environmental Science & Technology 50:10377-10385.

Burns, D., L Vasilakos, and R. Oestrike. 2005. Natural resources management plan for the Fishkill Creek Watershed: a Natural Resources Inventory and Conservation Strategy. Fishkill Creek Watershed Committee. FishkillCreekWatershed.org (Accessed July 3, 2018).

Cole, M., P. Lindeque, E. Fileman, C. Halsband, R. Goodhead, J. Moger, and T.S. Galloway. 2013. Microplastic ingestion by zooplankton. Environmental Science & Technology 47:6646-6655.

Coppock R.L., M. Cole, P.K. Lindeque, A.M. Queirós, and T.S. Galloway. 2017. A small-scale, portable method for extracting microplastics from marine sediments. Environmental Pollution 230:829-837.

Deng, Y., Y. Zhang, B. Lemos, and H. Ren. 2017. Tissue accumulation of microplastics in mice and biomarker responses suggest widespread health risks of exposure. Scientific Reports 7:46687.

Espinosa, C., M.A. Esteban, and A. Cuesta. 2016. Microplastics in aquatic environments and their toxicological implications for fish. Toxicology- New Aspects to This Scientific Conundrum 1:113-145.

Hidalgo-Ruz, V., L. Gutow, R.C. Thompson, M. Thiel. 2012. Microplastics in the marine environment: A review of the methods used for identification and quantification. Environmental Science & Technology 46:3060–3075.

Hill, J., J.W. Evans, and M.J. Van Den Avyle. 1989. Species profiles: life histories and environmental requirements of coastal fishes and invertebrates (South Atlantic)- striped bass. U.S. Fish and Wildlife Service. Biological Report 82 (11.118).

Khan, M.B., and R.S. Prezant. 2018. Microplastic abundance in a mussel bed and ingestion by the ribbed marsh mussel Geukensia demissa. Marine Pollution Bulletin 130:67-75.

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. https://marinedebris.noaa.gov/sites/default/files/publications- files/noaa_microplastics_methods_manual.pdf (Accessed September 3, 2018).

II-32

Melendez, V.P., J. Rubbo, and M.J. Greene. 2010. “An interim watershed management plan for the lower, non-tidal portion of the Rondout Creek, Ulster County, New York.” Hudson River Sloop Clearwater Incorporation. http://www.clearwater.org/green-cities/watershed-management/rondout-creek- watershed-council/ (Accessed January 5, 2018).

New York Office of the Attorney General. 2015. Unseen threat: how microbeads harm New York waters, wildlife, health and environment. Office of the New York State Attorney General 15:1-14.

New York State Department of Environmental Conservation (NYSDEC). 2018a. “Constitution Marsh Audubon Center and Sanctuary.”. http://www.dec.ny.gov/outdoor/63624.html (Accessed July 5, 2018).

New York State Department of Environmental Conservation (NYSDEC). 2018b. “Tivoli Bays National Heritage Area”. http://www.dec.ny.gov/lands/92370.html (Accessed June 5, 2018).

New York State Department of Environmental Conservation (NYSDEC) 2019. “Blue Crab in the Hudson River” https://www.dec.ny.gov/animals/37185.html (Accessed March 12, 2019).

Saw Kill Watershed Community. 2016. “About the Saw Kill and its watershed”. https://sawkillwatershed.wordpress.com/about-the-saw-kill-and-its-watershed/ (Accessed January 2, 2018).

Van, A., C.M. Rochman, E.M. Flores, K.L. Hill, E. Vargas, S.A. Vargas, and E. Hoh. 2012. Persistent organic pollutants in plastic marine debris found on beaches in San Diego, California. Chemosphere 86:258-256.

Vermaire, J.C., C Pomeroy, S.M. Herczegh, O. Haggart, and M. Murphy. 2017. Microplastic abundance and distribution in the open water and sediment of the Ottawa River, Canada, and its tributaries. FACETS 2:301–314.

II-33

SIZE DOES MATTER: EXPOSURE AND EFFECTS OF MICROPLASTICS ON THE EASTERN OYSTER (CRASSOSTREA VIRGINICA)

A Final Report of the Tibor T. Polgar Fellowship Program

Erika Bernal

Polgar Fellow

Program Marine Biology and Coastal Sciences Montclair State University 1 Normal Avenue Montclair, NJ, 07043

Project Advisors:

Paul A. X. Bologna Director Marine Biology and Coastal Sciences Department of Biology Montclair State University Montclair, NJ 07043

Beth Sharack NOAA Fisheries James J. Howard Marine Sciences Laboratory Northeast Fisheries Science Center 74 Magruder Road Sandy Hook, NJ 07732

Bernal, E., B. Sharack, and P.A.X. Bologna. 2020. Size Does Matter: Exposure and Effects of Microplastics on the Eastern Oyster (Crassostrea virginica). Section III: 1-30 pp. In S.H. Fernald, D.J. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

III-1 ABSTRACT

The global impact plastic pollution has on aquatic ecosystems is rapidly

increasing, and there are numerous studies highlighting the negative impacts from

microplastic exposure. While the general effects of microplastics are becoming clearer, less is known about the specific impacts of the various polymers that make up plastic.

Moreover, many studies show the effects of exposing organisms to microplastics of the same shape and size, which is an inaccurate representation of what organisms are exposed to in the wild. Here, the Eastern Oyster (Crassostrea virginica) was exposed to four types of polymers, and its feces, pseduofeces, and internal tissues were analyzed for microplastics. The results from this study showed plastic particles were present in two main organs - the digestive system, and the gills and mantle. Polystyrene was present in nearly all individuals analyzed, suggesting this type of polymer can increase exposure which may be harmful to filter feeders. Despite the use of their rejection mechanism, oysters did not distinguish polystyrene and polyvinyl chloride from food. Polyethylene was absent in tissues, but was detected in feces, suggesting that C. virginica can reject this polymer. Toward the end of the experiment, an accumulation of polyethylene and polyvinyl chloride was documented, suggested that a longer-term exposure to weathered particles may have a greater impact via biofilm development. Due to their complexity it is necessary for microplastic studies to expose organisms to polymers of various types as well as irregular shapes and sizes. Understanding the potential impacts from diverse polymers is critical for management of waste and can provide important information on which types of plastic may be harmful to organisms inhabiting the environment.

III-2 TABLE OF CONTENTS

Abstract ...... III-2

Table of Contents ...... III-3

List of Figures and Tables...... III-4

Introduction ...... III-5

Methods...... III-9

Collection and Maintenance ...... III-9

Experimental Design ...... III-10

Plastic Fragments ...... III-10

Ingestion by Crassostrea virginica ...... III-11

Microplastics in Feces and Pseduofeces ...... III-11

Dissection and Digestion ...... III-12

Visual Inspection of Microplastics ...... III-12

Reducing Contamination ...... III-13

Statistical Analysis ...... III-13

Results ...... III-13

Discussion ...... III-20

Conclusion ...... III-23

Acknowledgements ...... III-26

References ...... III-27

III-3

LIST OF FIGURES AND TABLES

Figure 1 – The number (Mean + SE) of microplastic fragments found with the digestive system of Crassostrea virginica ...... III-14

Figure 2 – The number (Mean + SE) of microplastic fragments found within the gills and mantle of Crassostrea virginica ...... III-15

Figure 3 – The size range (mm) of microplastic fragments found within Crassostrea virginica tissues ...... III-16

Figure 4 – The number of PVC and PS microplastics found within oysters digestive system ...... III-17

Figure 5 – The number (Mean + SE) of microplastic fragments found within the feces of Crassostrea virginica ...... III-18

Figure 6 – The number (Mean + SE) of microplastic fragments found within the pseudofeces of Crassostrea virginica ...... III-19

Figure 7 – The number of Polyethylene microplastics found within the feces, and pseudofeces of Crassostrea virginica ...... III-19

III-4 INTRODUCTION

Plastic pollution is a major environmental concern for both aquatic and terrestrial

ecosystems. Currently, global plastic production is over 300 million tons annually, and

more than half of this is manufactured for single use (Lehtiniemi et al. 2018). As a result

of mass production, plastics are the largest source of marine debris, and they continue to

accumulate in the oceans and potentially threaten organisms at all trophic levels

(Lehtiniemi et al. 2018). In general, plastic is a term used to describe synthetic materials

that can be easily shaped or molded; however, it is important to recognize that the single- chain organic molecules that make up plastic are complex and vary based on the polymer construction and finished product. The polymers that are used to construct plastic make it virtually impossible to biodegrade on a short time scale, and, as a result, these objects can persist in the environment for many centuries (Green et al. 2018). Many of these synthetic polymers end up in the water as fragments of larger particles, or can arise from cosmetics and synthetic fibers from clothing (Vandermeersch et al. 2015).

Microplastics are defined as small pieces of plastic that range from 0.1 µm - 5mm

(Sussarellu et al. 2016; Green et al. 2018) and have been known to have a profound impact on marine biota from all trophic levels (Cauwenberghe et al. 2014). Although microplastics are found in many different forms, they are divided into two categories: primary and secondary. Primary microplastics are manufactured particles added to personal-care products and fillers for industrial applications. Secondary microplastics are fragments of larger plastic pieces physically broken apart through environmental exposure to sunlight and processes in the environment like wind and wave action.

Secondary microplastics also include fibers that originate from the degradation of fishing

III-5 gear or clothing (e.g., nylon, spandex, etc.) and can enter the environment through

wastewater from industries and households.

Once microplastics enter aquatic ecosystems, they can absorb persistent organic

pollutants (POPs), such as polychlorinated biphenyls (PBCs) and polycyclic aromatic

hydrocarbons (PAH) (Endo et al. 2005), and accumulate in the water column. Moreover,

the plastic themselves can release ‘additives,’ which are chemical compounds added

during manufacture to improve the performance, functionality, and aging properties of

the polymer (Lehtiniemi et al. 2018). Today, some of the most commonly used additives

in various plastic materials include plasticizers, flame retardants, and heat stabilizers,

which are known to cause endocrine disruption, cancer, and birth defects (Smith and

Bertola 2010). Furthermore, because the density of microplastics may vary based on

chemical composition of the polymers (Lusher 2015), higher density particles may sink

and accumulate in the sediment, whereas low density particles may float at the sea surface (Cauwenberghe et al. 2014). However, it should be noted that weathering processes, as well as turbulence, freshwater input, and mixing may also contribute to the relative distribution of these particles (Lusher 2015).

Today, the presence of microplastics have been reported worldwide and include samples from several marine organisms as well as sediments (Shim and Thompson

2015). Microplastics can be ingested directly by filter feeding and deposit feeding organisms, or indirectly by consumption of prey containing them (Zhang et al. 2019). It is believed that filter feeding species are especially vulnerable to microplastics because of their ability to filter large volumes of water (Sussarellu et al. 2016), thereby increasing their exposure. Several studies have tested the effects of microplastic ingestion in

III-6 bivalves (Browne et al. 2008; Von Moos et al. 2012; Cauwenberghe et al. 2014;

Sussareullu et al. 2016; Khan and Prezant 2018), as well as the abundance of microplastics within aquatic systems (Eerkes-Medrano et al. 2015; Shim and Thompson

2015). River systems in particular may show specific microplastic characteristics based

on the waste sources near the river (Eerkes-Medrano et al. 2015). As a result, estuaries

are at a higher risk to microplastics exposure, because of their close proximity to point sources and the small relative size of estuarine systems (Eerkes-Medrano et al. 2015).

Aquacultured organisms such as finfish and shellfish are also typically grown in open systems with natural seawater, which makes them extremely vulnerable and more likely to be exposed to any pollutant present in the water column (Cauwenberghe et al. 2014).

Bivalves are suitable model organisms for microplastic studies because they are good indicators of water quality, can be easily sampled, and are highly resistant to stress.

In general, suspension feeders process relatively large amounts of water during feeding, which allows them to be exposed to high amounts of harmful materials, ultimately leading to an accumulation of chemical pollutants. Bivalves, like oysters, have a unique mechanism for particle selection in which they sort particles based on their size, shape, palatability or chemical composition (Ward and Shumway 2004; Xu et al. 2017).

Typically, particle selection occurs on the gills and labial palps where particles selected are moved from the gills into the bivalve’s mouth, where they are eaten, digested and expelled as feces. (Xu et al. 2017). Unwanted particles, which include materials that are

too large and dense, small grains of sand, and detritus, are selected by the labial palps and

transferred to the mantle cavity as a mucus-bound mass. This mucus-bound material is known as pseduofeces, and although it may resemble actual feces, this unwanted material

III-7 is ejected without having passed through the digestive tract (Beninger et al. 1999).

Pseudofeces production is an effective mechanism bivalves have that allows the rejection

of inorganic particles upon encountering them in their feeding stream; however, an

accumulation of microplastics may inhibit their ability to sort and reject unwanted

particles (Xu et al. 2017), leading to their accumulation in their tissues (Browne et al.

2008).

Bivalves, such as oysters, can also exhibit immunological problems including

neurotoxic and genotoxic physiological responses to microplastic exposure (Avio et al.

2015), which could have a cascading ecological impact to a population. A recent study on

microplastics exposure demonstrated strong impacts on the feeding activity, absorption

rates, fecundity and offspring growth in oysters (Sussarellu et al. 2016). Another potential

impact from the accumulation of microplastics during filter feeding can occur by reduced

ciliary movement on the gills, which impacts the oyster’s ability to pump water (Xu et al.

2017). Previous studies indicate that there is no accumulation of microplastics in the gut

(e.g., Sussarellu et al. 2016); however, in these cases exposure focused on the use of a

single type of microbead. The use of one type of polymer is not an accurate

representation of what filter feeders are exposed to in nature. More importantly, the shape

of a plastic particle can influence the biological effects (Choi et al. 2018) and further

studies are necessary to understand the potential impacts from various types, sizes, and

shapes of polymers. Overall, the accumulation of plastics may vary and ultimately

depend on the chemical characteristics of each polymer. For example, polymers such as

polyethylene and polystyrene are believed to absorb a much higher concentration of

pollutants than compared to polyvinyl chloride (Rochman et al. 2017). Consequently,

III-8 plastic composition may shape exposure levels for pollutants and therefore, understanding the response of filter feeders to different polymers is critical to assess how microplastics may impact species and communities.

In this study, the aim was to improve the understanding of the ingestion of microplastics of varying composition, shapes, and sizes. Experiments were conducted using four common polymers found in the environment including: Polyethylene,

Polyvinyl chloride, Polystyrene, and Polypropylene. These polymers are associated with many commonly used plastic products such as bags, bottles, straws, and food containers, as well as many household and automotive products. The objective was to evaluate the ingestion rate of oysters exposed to these four types of polymers, as well as analyze their tissues, pseudofeces, and feces for microplastics.

The null hypotheses tested were 1) all four types of polymers will not be ingested despite differences in chemical composition and 2) polymers will not be present in oyster tissues, feces, or pseudofeces. While it was expected to find microplastic particles present in both pseudofeces and feces, it is predicted that more particles would be found in the feces, which would suggest that oysters are not able to discriminate between plastic and their natural food sources.

METHODS

Collection and Maintenance

Fifty (50) adult Eastern Oysters (Crassostrea virginica) were obtained from

Sweet Amalia Oyster Farm in Newfield, New Jersey during June 2018 and transported to

the National Oceanic and Atmospheric Administration (NOAA) James J. Howard

Marine Sciences Laboratory, where they were placed into two 10 gallon aquarium tanks.

III-9

Before initiating experiments, oysters were fed daily for one week with a commercial shellfish diet containing six marine microalgae species – Isochyrsis, Pavlova,

Tetraselmis, Chaetoceros calcitrans, Thalassiosira weissflogii, and Thalassiosira

pseudonana (Reedmariculture Inc.).

Experimental Design

After acclimatization, oysters were randomly selected and placed into 2-Liter glass

beakers and assigned to one of the following five treatments (N=10/treatment): Control,

Polyethylene (PE), Polystyrene (PS), Polyvinyl chloride (PVC), and Polypropylene (PP).

To standardize hunger levels, oysters were starved two days prior to the start of feeding

experiments. Air circulation and mixing was achieved by attaching glass pipettes to air

bubbler tubing. Oysters were fed daily using the commercial shellfish diet either with or

without their designated microplastic.

Plastic Fragments

Plastic polymers for this experiment were generated and identified using methods

from Fries et al. (2013). Plastic fragments made of PP were produced from a blue used

children’s toy. Fragments were generated by grinding plastic in a commercial mill and

storing the material in a glass vial. PE fragments were produced from a yellow mesh

produce bag. Fragments were generated by using scissors to cut the mesh in various sizes

and storing the material in a glass vial. PS fragments were obtained from a foam packaging

material and a scalpel was used to scrape material into various sizes for feeding. PVC

fragments were produced from a grey PVC pipe, and similarly to PS, fragments of PVC

were obtained by using a scalpel to scrape material.

III-10 In order to standardize the amount of plastic introduced into experimental

treatments, microplastic polymers were weighed and allocated into 0.0051g samples, labeled, and stored into a separate glass vial. For experimental exposure trials, the daily feeding allotment of commercial shellfish diet was added to each glass vial with their assigned polymer to a total volume of 5ml prior to exposure feeding. Each vial was thoroughly mixed by hand for one minute and then fed to the experimental oysters.

Ingestion by Crassostrea virginica

For experimental groups, oysters were fed a 5ml mixture of commercial shellfish

diet along with their assigned polymer. Similarly, oysters in the control group were fed the

same concentration of shellfish diet without any microplastics. Twenty four hours post

feeding/exposure, five random oysters (one from each group) was chosen for dissection

and analyzed for microplastics. The remaining oysters were left in beakers and fed daily

the commercial diet with the assigned microplastic group (5 ml) accordingly for the

following ten days. Each subsequent day, an oyster from each group was removed,

dissected, and analyzed for the presence of microplastics until all oysters were dissected.

There were no water changes. The last group of oysters was sacrificed on day eleven (11).

Microplastics in Feces and Pseudofeces

Collection of feces and pseudofeces occurred after feeding during observation

hours 1, 8, and 24. In order to avoid mischaracterization during analysis, feces were

collected with a dropper only when they were seen actively being released from oysters.

No collection occurred when feces were observed at the bottom of the glass container,

but release was not seen to minimize any bias associated with mixing of feces and

pseudofeces not directly observed. The same method was applied for the collection of

III-11 pseudofeces. Identification and recognition of pseudofeces was performed by using a visual confirmation; pseudofeces were released in the form of a mucus ball, whereas feces were loose and had a string-like texture.

Dissection and Digestion

Before dissection, each oyster was wiped with ethanol using a cotton cloth to

clean the external surface of any plastics and then each individual was weighed and

measured. Oysters were dissected by tissue types into Gills and Mantle, Labial Palps,

Stomach (Digestive system), and Adductor Muscle. Once tissues were dissected out, they

were placed into individual glass beakers and wet weight was recorded. Tissue weights

were compared among treatments to determine if oyster size or tissue weight differed

among treatments potentially biasing the results. Tissue samples were then digested in a

10% Potassium Hydroxide (KOH) solution. To avoid cross contamination among

samples, all tools and glassware were rinsed three times between samples and each

sample jar was immediately covered with aluminum foil once KOH was added. Finally,

each sample jar was incubated for three days at 60 degrees Celsius to allow organic

materials to digest. After digestion, solutions were filtered over a 0.45µm glass

membrane filter (Whatman) and placed into glass petri dishes to dry for 24 hours.

Visual Identification of Microplastics

Microplastic particles were visually identified using a compound microscope.

Each filter was scanned at 10x magnification and each polymer found was measured

along its longest dimension. After measuring each known polymer, a photo was taken to

ensure polymers were not counted twice.

III-12

Reducing Contamination

Preventing contamination in microplastic research is a challenge due to the airborne

fibers (Cauwenberghe et al. 2014) and possible cross contamination between samples. To

prevent contamination in various forms, several strategies were used in this experiment.

First, the preparation of microplastics was handled in a separate room using a new 100%

cotton lab coat. For processing samples, another new 100% cotton lab coat was worn at

all times and before processing, all counters were wiped using deionized water followed

by ethanol with a cotton cloth. All equipment used was rinsed three times before use and

all sample processing was performed in a closed, isolated, plastic-free room with tacky

mats. Procedural blanks were included in every KOH digestion to account for any possible

contamination. Blanks were processed in the same manner as oyster tissues.

Statistical analysis:

A one-way ANOVA was performed using the PROC GLM procedure in SAS®,

with type of plastic as the independent variable and the number of plastic fragments

collected from feces, pseudofeces, and tissue type as the dependent variables.

Significance was attributed to comparisons between means with an alpha value set at 0.05

using the LSMEANS Procedure.

RESULTS

Experimental microplastic fragments were only detected in two of the organs dissected: Digestive System (Figure 1) and the Gills and Mantle (Figure 2) tissues of the oysters, with the digestive system retaining the highest number of microplastics

III-13

(Figure 1). It should be noted that organ systems were dissected a whole. As such, it was not possible to differentiate whether plastics in the digestive system were in the lumen or had in fact passed into the digestive tract.

Figure 1: The number (Mean + SE) of Microplastic fragments found within the digestive system of Crassostrea virginica. Bar graphs with different letters represent statistically significant differences between treatment groups.

III-14

Figure 2: The number (Mean + SE) of Microplastic fragments found within the Gills and Mantle of Crassostrea virginica. Bar graphs with different letters represent statistically significant differences between treatment groups.

No polymer fragments were found within oyster adductor muscles or labial palps for all treatment groups. PS fragments were detected in more than 90 percent of oysters assigned to the PS treatment group, while PE was not found in any of the assigned oyster tissues. Oysters assigned to the PP treatment group ingested the fewest number of total particles (n=14), whereas oysters in the PS treatment group ingested the greatest number of particles (n=47). The size of microplastics documented from oyster tissues ranged from 0.05 to 2mm (Figure 3).

III-15

Figure 3: The size range (mm) of microplastic fragments found within Crassostrea virginica tissues.

An analysis of the relationship between oyster size and individual tissue weights

against the number of plastic particles present was conducted. Results showed that no significant differences were present between the oyster size and tissue weights with the

number of plastics within oysters’ tissues; however, significant differences in the presence

of microplastics among oyster treatment groups did occur for the digestive system (Figure

1; F4,42 = 9.20, P < 0.0001) and the gills and mantle tissues (Figure 2; F4,42 = 5.00, P <

0.0022). Oysters fed PS microplastics had significantly more fragments inside the

digestive system compared to the other treatment groups (Figure 1). Similarly, there was

a significantly greater number of PS fragments in their gills and mantle than any other

treatment group (Figure 2). The number of PVC fragments was significantly higher within

the digestive system (Figure 1), but was not significantly different from the control group

within the gills and mantle (Figure 2). In contrast, PP fragments were significantly higher

III-16 within oyster gills and mantle (Figure 2), but were not significantly different from the control within the digestive system (Figure 1). One interesting change in the consumption and retention of plastic particles, which developed during the experiment, was an increase in the presence of PVC and PS particles within the digestive system as the experiment progressed from Day 1 to Day 11 (Figure 4). This increase over time might reflect ‘aging’ in microplastics where they develop a microbial film once they are introduced to the

environment.

Figure 4: The number of PVC and PS particles found within oysters digestive system during the progression of the experiment.

Feces and pseudofeces were collected only during three observation hours 1, 8

and 24. No feces were produced during the 1-hour mark, therefore no collection occurred.

By the end of 24 hours, due to decomposition it was impossible to distinguish feces

between actual feces or pseudofeces; therefore, these data were removed from the

III-17 analysis. However, for the 8-hour observation period, the number of particles in feces

varied significantly among oysters assigned to PS, PVC and PE groups (Figure 5; F4,32 =

12.79, P < 0.001).

Figure 5: The number (Mean + SE) of microplastics fragments found within the Feces of Crassostrea virginica. Bar graphs with different letters represent statistically significant differences between treatment groups.

As in the tissues, the number of PS fragments were significantly higher than any

other polymer group. Experimental polymer fragments were found as early as day 1 in

pseudofeces. The number of polymer fragments in pseudofeces varied significantly

among the treatment groups (Figure 6; F4,17 = 21.32, P < 0.001). Oysters assigned to PS

and PVC group had a significantly higher number of particles present than oysters in the

PE and PP groups. Initially, PE fragments were only detected in oyster pseudofeces;

however, towards the end of the experiment (Day 11) PE fragments were also found in

the feces (Figure 7). There were no microplastics found in the feces or pseudofeces of the

control group.

III-18

Figure 6: The number (Mean + SE) of microplastic fragments found within the Pseudofeces of Crassostrea virginica. Bar graphs with different letters represent statistically significant differences between treatment groups.

Figure 7: The number of Polyethylene microplastics found within the feces, and pseudofeces of Crassostrea virginica.

III-19

DISCUSSION

One of the primary goals for this study was to expose oysters to the various types of polymers that are commonly found in the environment. In terms of management, understanding the impacts from various polymers can provide insight on the overall environmental impact a material or process may have. In particular, oysters assigned to the PS group had a significantly higher number of particles than any other treatment group. These results would suggest that PS has the potential to impact oysters and other filter feeding organisms to a greater extent. Moreover, PS was present in feces and pseudofeces, which suggests that oysters were unable to distinguish PS from food and therefore, could not utilize their rejection mechanism effectively. Similarly, an exposure study of polystyrene beads on mussels showed spheres (5 um) were found throughout the

stomach and intestine of all mussels within the experiment (Khan and Prezant 2018). PS

is one of the most commonly used and recycled plastics, with a global production of more

than 14 million US tons every year (Chandra et al. 2016). Although it is accepted by

some recycling facilities, the shredding processes of recycled polystyrene yields high

amounts of secondary microplastics that can ultimately be redistributed into the

surrounding environment. Larger pieces of polystyrene can also be easily degraded to

smaller pieces due to the combination of many environmental factors such as sun

exposure and weathering within the marine environment and thus pose a great impact to

the health of the ecosystem.

In contrast, PE was not present in any of the oyster’s tissues; however, it was found in

both types of feces. It should be noted that PE was only present in feces collected toward

the end of the experiment (Figure 7). This could be due to the formation of a biofilm on

III-20 accumulated particles in the experimental chambers, which would inhibit the oyster’s ability to reject PE. In the environment, once microplastics enter waterways they are quickly conditioned with a layer or film of organic and inorganic substances by adsorption (Rummel et al. 2017). It is through this initial conditioning layer that microorganisms begin to interact with microplastics and ultimately lead to the development of a biofilm (Rummel et al. 2017). Biofilms may contain similar taxa to those which filter feeders may ingest, as well as secrete chemicals that increase the likelihood of the microplastic being mistaken for a food source. A study on copepods showed greater ingestion of aged microplastic beads than pristine microbeads (Vroom et al. 2017), suggesting that organisms are extremely vulnerable to microplastics in the environment due to the aging processes of weathering and biofouling. Moreover, biofilm formation may differ among polymers due to the difference in polymer composition as well as the amount of supplemental chemicals that were added into the polymer during manufacturing (Rummel et al. 2017). For example, Rogers et al. (1994) suggested the higher bacterial count on PE and PVC, as compared to stainless steel, was due to the

leaching of additives contributing to biofilm development. These findings concur with the

results, which showed that PE was not initially consumed, but as the experiment

progressed and a biofilm likely formed on particles, oyster feces showed traces of PE

(Figure 7).

Today, much of single use plastic packaging is composed of PE (Plastics Europe

2016) and it is considered the most abundant form of coastal litter; therefore, the greatest

source of microplastics in the environment. A major reason PE is favored by

manufacturing companies is because it is a “thermoplastic” which means it can be heated

III-21 to its melting point, cooled and reheated again without significant damage to the material

(GESAMP 2010). However, the same characteristics that make PE and other thermoset materials (PP, PVC, and PS) versatile is what also makes them difficult to dispose of or recycle. Once heated, the chemical composition of the thermoplastic is completely changed and if heated a second time the material will simply burn (GESAMP 2010).

Over time, discarded thermoset plastics will end up in landfills or become contributors to marine litter.

Longer-term and short-term exposures of specific polymers may also result in adverse biological impacts. A study on the polychaete worm, Arenicola marina, showed a

significant reduction in its feeding activity and the gut passage time of sediments due to

the chronic exposure of sediments containing PVC (Wright et al. 2013). Similarly, a 52-

day exposure to PE showed a significant reduction in the attachment strength and

production of byssal threads in the blue mussel, Mytilus edulis (Green et al. 2018). In the

current study, the number of PVC particles began to increase toward the end of the

experiment (Figure 7), suggesting a longer-term exposure may have a larger impact on

oysters.

Exposing organisms to various size particles may help us understand the different

pathways microplastics may undergo within an organism. Smaller size particles can

translocate within an organism by passing through the cell membranes and once taken up,

they can be retained for long periods of time (Browne et al. 2007). In this study, polymers

from each treatment group with the exception of PE were confined to the stomach, gills

and mantle. Although biological impacts from microplastic ingestion were not

investigated in the present study, a study on Crassostrea gigas showed significantly

III-22 higher energy usage due to digestive interference from polystyrene microplastics in the gut (Sussarellu et al. 2016). While the present study was unable to discriminate plastics

in the different segments of the oysters’ digestive tract to confirm ingestion, Khan and

Prezant (2018) found that the mussel species Geukensia demissa did ingest similar sized

and shaped microplastics, and they were present in digestive tubules, suggesting active retention of PS. Previous studies discovered that oysters can ingest particles between 5 to

30µm (Baldwin and Newell 1995); however the results from laboratory studies like this one, as well as field experiments (Cauwenberghe et al. 2014), clearly show they can ingest larger particles. The results of the current study are consistent with those of previous studies (Browne et al. 2007; Koehler et al. 2008; Xu et al. 2017), which suggest that smaller particles are more likely to be ingested by filter feeders (Figure 3). These results also demonstrate a clear difference in the inability of oysters to discriminate PS from natural food resources, and that aging of microplastics reduces their ability to reject plastic particles. Consequently, aged microplastic presence in the environment could have substantial impacts on oyster survival and growth.

CONCLUSION

Oysters are ecosystem engineers and provide a wide range of ecosystem services such as water filtration, food, and habitat for many organisms (Beck et al. 2011). Oyster reefs also provide shoreline stabilization and coastal protection from natural disasters, and yet despite their importance, oyster habitats are continuously declining due to human-induced threats (Beck et al. 2011). Given that plastic waste is expected to continue to increase (Andrady 2011) and the amount of macroplastic fragmentation is

III-23 already happening, concentrations of microplastics will continue to heavily pollute the environment. Although oysters were observed rejecting and ejecting polymers like PE,

expelled waste and pseudofeces will ultimately draw down microplastics from the water

column, introducing them to the sediments. Subsequently, this exposes benthic epifauna

and infauna to these microplastics and the contaminants that they may carry (Galloway et

al. 2017). Microplastic bioavailability to marine animals is clear and, since they are so

ubiquitous and present throughout the environment, it would be wise to include them in

monitoring and models to predicate how their transport and accumulation may change

over time.

The results from this study clearly indicate that filter feeders such as oysters are

extremely vulnerable to ingestion of microplastics. Although oysters are more likely to

ingest smaller particles, this study shows that they are capable of ingesting larger

particles as well, which could result in digestive blockages within the oyster. Larger

pieces may also inhibit their ability to further reject polymers and as a result,

microplastics may have cascading impacts to individuals and the population. It is clear

that microplastics are a continuing and accelerating threat to the environment, but still so

much is yet to be understood.

Although there are many studies surrounding microplastics and the environmental

impact they pose, there is still so much unknown and room for more studies with

improved methodologies. There is a need for research to standardize methods using

similar terminology to reduce confusion and to aid comparison among studies. Preventing

and reducing contamination is also a major challenge due to the presence of fibers in the

atmosphere. More importantly, laboratory studies need to represent environmental

III-24 conditions by using similar concentrations and irregular shapes and sizes of particles to accurately understand the biological effects of microplastics. Studies using chemical analyses such as Raman spectroscopy are also needed to accurately identify fibers as plastic or nonsynthetic materials (Remy et al. 2015).

There is a need for exposure studies to evaluate differences between longer-term exposure and short-term exposure on feeding and health of organisms. In this study, PVC began to accumulate toward the end of the experiment, and PE, which was not consumed initially, was observed in both feces and pseudofeces toward the end of the experiment indicating aged plastics may harbor biofilms which make plastic particles indistinguishable from food items. This suggests that longer-term exposure to PVC and

PE would have a greater impact on bivalves than ‘fresh’ particles. Moreover, there is

little information on the fate of microplastics and whether particles are deposited in deep-

sea sediments (Choy et al. 2019) or limited to the shelf and coastline regions. Vertical

movement of various types and sizes of microplastics is also unknown, yet an important

research topic since microplastics with biofilms may sink, but once the biofilm is

removed through processes like digestion, the particles may become buoyant again

(GESAMP 2010). Overall, there is a need for further studies to evaluate the absorption and desorption rates between pollutants and microplastics, and whether this process is reversible (GESAMP 2010). For example, in regions where persistent, bioaccumulative and toxic substances concentrations are high, microplastics can readily become vectors and transport these toxic substances into cleaner remote regions. Notably, the majority of microplastic studies frequently concentrate on particles of the same size, because there is a significant amount of time, effort, and expense needed to process and analyze samples

III-25

(Lehtiniemi et al. 2018), but this does not mirror conditions in the real world. This study demonstrates several key findings and highlights the importance of investigating various polymers, including polymers with irregular shapes and sizes to ensure results are as accurate and unbiased as possible and to mimic real world conditions. Smaller particles have the potential to impact different organisms including individuals at lower trophic levels, whereas larger particles can cause blockages and inhibit biological processes like swimming and feeding behaviors (Choi et al. 2018). Also, this study showed how oysters

preferentially selected against some type of plastics, but as plastics age, they lose the

ability to distinguish them as non-food resources. This means that research into certain

polymers, like PE and PVC, require longer-term exposures and biofilm development to

fully understand the potential impacts these substances have on organisms and the

environment.

ACKNOWLEDGMENTS

I would like to thank Lisa Calvo from Sweet Amalia Oyster Farm for providing the oysters for this experiment. I would also like to thank Sandy Hook NOAA Fisheries for their facilities that were used in the housing and containment of oysters. Thank you to Nadia Sergis, for assisting in processing samples, and Ashok Deshpande for providing plastic material for feeding. A special thank you to Beth Sharack for assisting in oyster dissection and providing guidance during experiment. Thank you to my committee members Dr. Wu and Dr. Kight for their support and aid in my research. Finally, I would like to thank my advisor Dr. Paul Bologna for his guidance, mentoring and assistance throughout my research. Research was supported by the Tibor T. Polgar Fellowship program through the Hudson River Foundation.

III-26

REFERENCES

Andrady L.A. 2011. Microplastics in the marine environment. Marine Pollution Bulletin 62: 1596 – 1605.

Avio, G. C., S. Gorbi, M. Milian, M. Benedetti, D. Fattorini, G. d’Errico, M. Pauletto, L. Bargelloni, and F. Regoli. 2015. Pollutants bioavailability and toxicological risk from microplastics to marine mussels. Environmental Pollution 198: 211-222.

Baldwin, S.B., and R.E.I. Newell. 1995. Relative importance of different size food particles in the natural diet of oyster larvae (Crassostrea virginica). Marine Ecology Progress Series 120: 135 – 145.

Beck, W.M., R.D. Brumbaugh, L. Airoldi, A. Carranza, L. D. Coen, C. Crawford, O. Defeo, G.J. Edgar, B. Hancock, M.C. Kay, H.S. Lenihan, M.W. Luckenbach, C.L. Toropova, G. Zhang, and X. Guo. 2011. Oyster reefs at risk and recommendations for conservation, restoration, and management. BioScience. 61: 107 – 116.

Beninger, G.P., A. Veniot, and Y. Poussart. 1999. Principles of pseudofeces rejection on the bivalve mantle; integration in particle processing. Marine Ecology Progress Series 178: 259-269.

Browne, M.A., A. Dissanayake, T. Galloway, D.M. Lowe, and R.C. Thompson. 2008. Microscopic plastic translocates to the circulatory system of the mussel, Mytilus edulis (L). Environmental Science and Technology. 45: 9175 – 9179.

Browne, M.A., T.S. Galloway, and R.C. Thompson. 2007. Microplastic – An emerging contaminant of potential concern? Integrated Environmental Assessment and Management, 3: 559 – 561.

Cauwenberghe, L., and C. Janssen. 2014. Microplastics in bivalves cultured for human consumption. Environmental Pollution. 193: 65 – 70.

Chandra, M., C. Kohn, J. Pawlitz, and G. Powell. 2016. Real Cost of Styrofoam Experiential Learning Project. Saint Louis University.

Choi, J.S., Y. Jung., N. Hong., and J. Park. 2018. Toxicological effects of irregularly shaped and spherical microplastics in a marine teleost, the sheepshead minnow (Cyprinodon variegatus). Marine Pollution Bulletin. 129: 231 – 240.

Choy, C. A., B.H. Robison., T.O. Gagne., B. Erwin., E. Firl., R.U. Halden., J.A. Hamilton., K. Katija., S.E. Lisin., C. Rolsky., and K.S. Van Houtan. 2019. The vertical distribution and biological transport of marine microplastics across the epipelagic and mesopelagic water column. Scientific Reports 9: 7843. https://doi.org/10.1038/s41598-019-44117-2.

III-27

Eerkes-Medrano, D., R.C. Thompson., and D.C. Aldridge. 2015. Microplastics in freshwater systems: a review on the emerging threats, identification of knowledge gaps and prioritization of research needs. Water Research. 75: 63-82.

Endo, S., R. Takizawa, K. Okunda, H. Takada, K. Chiba, H. Kanehiro, H. Ogi, R. Yamashita, and T. Date. 2005. Concentration of polychlorinated biphenyls (PCBs) in beached resin pellets: Variability among individual particles and regional differences. Marine Pollution Bulletin 50: 1103 – 1114.

Fries, E., J.H. Dekiff, J. Willmeyer, M. Nuelle, M. Ebert, and D. Remy. 2013. Identification of polymer types and additives in marine microplastic particles using pyrolysis- GC/MS and scanning electron microscopy. Environmental Science: Processes and Impacts 15: 1949 – 1956.

Galloway, T., M. Cole, and C. Lewis. 2017. Interactions of microplastic debris throughout the marine ecosystem. Nature Ecology & Evolution 1: 0116.

GESAMP. 2010. IMO/FAO/UNESCO-IOC/UNIDO/WMO/IAEA/UN/UNEP Joint Group of Experts on the Scientific Aspects of Marine Environmental Protection); Bowmer, T. and Kershaw, P.J., 2010 (Eds.), Proceedings of the GESAMP international Workshop on plastic particles as a vector in transporting persistent, bio- accumulating and toxic substances in the oceans. GESAMP Rep. Stud. No. 82: 68.

Green, S.D., T.J. Colgan., R.C. Thompson., and J.C. Carolan. 2018. Exposure to microplastics reduces attachment strength and alters the haemolymph proteome of blue mussels (Mytilus edulis). Environmental Pollution 246: 423-434.

Khan, B.M., and R. Prezant. 2018. Microplastic abundances in a mussel bed and ingestion by the ribbed marsh mussel Geukensia demissa. Marine Pollution Bulletin. 130: 67- 75.

Koehler, A., U. Marx, K. Broeg, S. Bahns, and J. Bressling. 2008. Effects of nanoparticles in Mytilus edulis gills and hepatopancreas – A new threat to marine life? Marine Environ Research. 66: 12-14.

Lehtiniemi M., S. Hartikainen, P. Näkki, J. Engstrom-Öst, A. Koistinen, and O. Setälä 2018. Size matters more than shape: Ingestion of primary and secondary microplastics by small predators. Food Webs 16: 2352 – 2496.

Lusher, A. 2015. Microplastics in the marine environment: distribution, interactions and effects. Pages 245-307 in Bergmann M., L. Gutow, and M. Klages, editors Marine Anthropogenic Litter. Springer..

Plastics Europe. 2016. Plastics – the facts 2016. An analysis of European plastics production, demand and waste data. Last accessed January 2020: www.plasticseurope.org/application/files/4315/1310/4805/plastic-the-fact-2016.pdf

III-28

Remy, F., F. Collard, B. Gilbert, P. Compère, G. Eppe, and G. Lepoint. 2015. When microplastic is not plastic: the ingestion of artificial cellulose fibers by macrofauna living in seagrass macrophytodetritus. Environmental Science and Technology, 49: 11158 – 11166.

Rochman, C., J.M. Parnis., M. Browne., S. Serrato., E.J. Reiner., M. Robson., T. Yong., M.L. Diamond., and S.J. Teh. 2017. Direct and indirect effects of different types of microplastics on freshwater prey (Corbicula fluminea) and their predator (Acipenser transmontanus). Plos One 12, e0187664.

Rogers, J., A.B. Dowsett, P.J. Dennis, J.V. Lee, and C.W. Keevil. 1994. Influence of plumbing materials on biofilm formation and growth of Legionella pneumophilia in potable water systems. Applied and Environmental Microbiology 60: 1842 – 1851.

Rummel D.C., A. Jahnke, E. Gorokhoba, D. Kühnel, and M. Schmitt-Jansen. 2017. Impacts of biofilm formation on the fate and potential effects of microplastic in the aquatic environment. Environmental Science and Technology Letters 4: 258 – 267.

Smith, M.I., and V. Bertola. 2010. Effect of polymer additives on the wetting of impacting droplets. Physical Review Letters, 104: 154502.

Shim, W.J., and R.C. Thompson. 2015. Microplastics in the ocean. Archives of Environmental Contamination and Toxicology, 69: 265-268.

Sussarellu, R., M. Suquet., Y. Thomas., C. Lambert., C. Fabioux., M.E. Pernet., N. Goic., V. Quillen., C. Mingant., Y. Epelboin., C. Corporeau., J. Guyomarch., J. Robbens., I. Paul-Pont., P. Soudant., and A. Huvet. 2016. Oyster reproduction is affected by exposure to polystyrene microplastics. Proceedings of the National Academy of Science. 113: 2430-2435.

Vandermeersch G., L. Van Cauwenberghe, J. Colin, A. Marques, K. Granby, G. Fait, M. Kotterman, J. Diogene, K. Bekaert, J. Robbens, and L. Deviese. 2015. A critical view on microplastic quantification in aquatic organisms. Environmental Research 143: 46-55.

Von Moos, N., P. Burkhardt-Holm, and A. Köhler. 2012. Uptake and effects of microplastics on cells and tissue of the blue mussel Mytilus edulis L. after an experimental exposure. Environmental Science and Technology 46: 11327 – 11335.

Vroom, E.J.R., A.A. Koelmans, E. Besseling, and C. Halsband. 2017. Aging of microplastics promote their ingestion by marine zooplankton. Environmental Pollution 231: 987 – 996.

III-29

Ward, J.E., and S.E. Shumway. 2004. Separating the grain from the chaff: particle selection in suspension-and deposit-feeding bivalves. Journal of Experimental Marine Biology and Ecology 300: 83-130.

Wright, S.L., D. Rowe, R.C. Thompson, and T.S. Galloway. 2013. Microplastic ingestion decreases energy reserves in marine worms. Current Biology 23: 1031 – 1033.

Xu, X-Y., W.T. Lee, A.K.Y. Chan, H.S. Lo, P.K.S. Shin, and S.G. Cheung. 2017. Microplastic ingestion reduces energy intake in the clam Atactodea striata. Marine Pollution Bulletin. 124: 798 – 802.

Zhang, F., Y.B. Man, W.Y. Mo, K.Y. Man, and M.H. Wong. 2019. Direct and indirect effects of microplastics on bivalves, with a focus on edible species: A mini-review. Critical Reviews in Environmental Science and Technology, 1-35. DOI: 10.1080/10643389.2019.1700752.

III-30

THE EFFECT OF SALINITY ON EASTERN OYSTER REPRODUCTION IN THE HUDSON RIVER ESTUARY

A Final Report of the Tibor T. Polgar Fellowship Program

Kaili M. Gregory

Polgar Fellow

Environmental and Sustainability Sciences Cornell University Ithaca, NY 14853

Project Advisor:

Matthew Hare Department of Natural Resources Cornell University Ithaca, NY 14853

Gregory, M. K. and M. Hare. 2020. The Effect of Salinity on Eastern Oyster Reproduction in the Hudson River Estuary. Section IV:1-43 pp. In S.H. Fernald, D.J. Yozzo, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

IV-1

ABSTRACT

Oysters have largely been missing from the Hudson River Estuary (HRE) since the early 1900s due to human activity, pollution and overharvesting. Water quality has been steadily improving in the estuary and coupled with the discovery of a wild remnant population, restoration is looking more feasible. Water quality factors such as dissolved oxygen, pH, and salinity can have effects on the survival and effectiveness of oysters.

This research focuses on the effect of salinity on the timing of gametogenesis in eastern oysters (Crassostrea virginica) between June and August 2018 in the Hudson River

Estuary (HRE). Gametogenesis was monitored in 2-year old oysters from experimental cage populations in the HRE. Adult oysters from a remnant population were transplanted to diverse salinities so growth rate and survivorship could be measured and compared.

Gametogenesis was evaluated using a histological gonad index, ranking the amount and maturity of gametes in the oyster gonad on a scale of 0-10. A GI value of 5 represents peak maturation, and a value of 6 represents active spawning. Based on literature, it was hypothesized that oyster gametogenesis would be delayed in lower salinity water relative to higher salinities in the Hudson River Estuary. The reproductive pattern was not a simple shift of the whole reproductive cycle, as predicted. A more dramatically altered reproductive phenology within the seasonal period studied was observed. The northern, low salinity sites started spawning earlier and spawned for longer through the reproductive season than the high salinity oysters. The results of this study demonstrate that adult oysters from the remnant wild population are flexible and tolerant of higher salinity conditions, and regardless of genetic strain, oysters change the timing of their reproductive output depending on the salinity where they live.

IV-2 TABLE OF CONTENTS

Abstract ...... IV-2

Table of Contents ...... IV-3

Lists of Figures and Tables ...... IV-4

Introduction ...... IV-5

Methods...... IV-11

Tappan Zee wild adult collection and transplant ...... IV-13

Monitoring of oysters and sampling for histology ...... IV-14

Histology preparation and Condition Index ...... IV-15

Histology slide analysis ...... IV-17

Statistical analyses ...... IV-18

Results ...... IV-19

Discussion ...... IV-27

Acknowledgments...... IV-33

References ...... IV-34

Appendix ...... IV-37

IV-3 LIST OF FIGURES AND TABLES

Figure 1 – Map of Hare Lab study sites ...... IV-12

Figure 2 – Shucked oyster anatomy...... IV-16

Figure 3 – Averaged weekly temperature data ...... IV-19

Figure 4 – Averaged weekly salinity data ...... IV-20

Figure 5 – Gonad index means and 95% confidence intervals - June ...... IV-22

Figure 6 – Gonad index means and 95% confidence intervals - July ...... IV-23

Figure 7 – Gonad index means and 95% confidence intervals - August ...... IV-24

Figure 8 –June GI vs. shell height ...... IV-25

Figure 9 – July GI vs. shell height ...... IV-25

Figure 10 – August GI vs. shell height ...... IV-25

Figure 11 – Gonad Index vs. Condition Index ...... IV-26

Figure 12 – Shell height over time (growth) for wild dredged TZ-HB adults ..... IV-27

Table 1 – Counts of oysters sampled for histology...... IV-14

Table 2 – Description of Gonad Index stages ...... IV-17

Table 3 – ANOVA results for effect on GI by all tested variables ...... IV-20

IV-4 INTRODUCTION

Crassostrea virginica, commonly known as the eastern oyster, was historically

abundant in the Hudson River Estuary. In the nineteenth and early twentieth century,

oysters were one of New York City’s top exports (Kurlansky 2006). Hudson River

oysters made it as far as Chicago, St. Louis, San Francisco, and even to Great Britain. In

1842, around $6 million worth of oysters were being sold in New York City. Adjusting

for inflation, that would be over $25 million in today’s economy. Eating oysters was

common and desired by people around the country (Kurlansky 2006). New York City

might be known today as the Big Apple, but back then the Big Oyster would have been

more appropriate.

As New York’s population grew, so did its pollution levels. Sewage and other types of waste were dumped into the estuary, harming the millions of oysters that lived on the bottom. Due to a synergistic mix of water pollution, eutrophication, disease and overharvesting, the once-abundant oyster beds near New York City were mostly dead by the beginning of the 20th century. Harvesting from the HRE beds was outlawed in 1920s

(Kurlansky 2006; MacKenzie 2007). Oyster populations are currently absent from most

of the Hudson River Estuary, but new data provides some hope for restoring this

keystone species to a more ecologically functional population despite the highly altered

conditions now existing in this urban estuary.

The water quality of the Hudson River Estuary has been steadily improving for

the past thirty years (Stinnette et al. 2018). Legislation such as the Clean Water Act of

1972 and the New York State Pure Waters Bond Act of 1965 prompted key initiatives to

clean up the Hudson. New sewage treatment plants have since been built along the

IV-5 estuary, and efforts by the New York State Department of Environmental Conservation and New York City Department of Environmental Protection have made positive strides towards cleaning up the HRE. The water is now generally safe to swim in, and more fish

species are entering the estuary (Stinnette et al. 2018). This long-term trend towards a

cleaner estuary system creates potential for restoration of a self-sustaining oyster population.

A successful restoration is characterized by a population large enough to be self- sustaining with measurable positive effects on ecosystem function. The eastern oyster is a keystone species, serving as a bioengineer that shapes its environment in ways that increase biodiversity and can indicate ecosystem health (Raj and Reson 2008). The population’s decline in the twentieth century left the estuary without this important biological and ecological keystone species.

The oyster’s reputation as a keystone species is based on several attributes.

Oysters are filter feeders, drawing water through their gills in order to obtain food. A single oyster can filter 50-55 gallons of water per day, removing particulates, detritus and dissolved pollutants (Raj and Reson 2008). Their effectiveness as filter feeders makes them reliable indicators of ecosystem health. Large populations of oysters in the

Chesapeake Bay have been shown to improve water quality in more shallow, mesohaline parts of the estuary (Coen et al. 2007). The water filtration process creates a positive feedback loop such that cleaner water promotes oyster survival and more diverse benthic communities. Additionally, the physical structure of an oyster bed provides habitat to numerous estuarine invertebrates and fish stocks that are economically important to humans (Peterson et al. 2003).

IV-6 A wild remnant population of Crassostrea virginica was discovered just south of

the Tappan Zee bridge in the Haverstraw Bay portion of the HRE (Figure 1) where

average salinity is low. Near the remnant population there has been consistently high

recruitment since observations began in 2012, but elsewhere in the HRE, only few and

scattered recruits are found in a typical year (McFarland and Hare, 2018). Until the

discovery of this wild remnant population, there was no known persistent reproductive

population in the HRE.

Discovery of the Tappan Zee-Haverstraw Bay (TZ-HB) remnant oyster

population gives greater hope to efforts focused on restoring oysters to the lower estuary

(closer to the ocean) because they need not start from nothing. As indicated by The

Nature Conservancy in their practitioner’s guide to restoration practice (zu Ermgassen

2016), wild oyster recruitment or transplants are a preferred method for restoration

compared to planting aquaculture or hatchery-produced ‘seed’ oysters (Brumbaugh et al.

2006). Hatchery propagation, even when done carefully from wild parents, typically results in genetic bottlenecks compared to wild populations (Appleyard and Ward 2006;

Hornick and Plough 2018). Oyster strains artificially selected for commercially valuable traits, such as fast growth, disease resistance, and attractive shells, are sometimes used in a restoration context (Baggett et al. 2014) but have unknown reproductive fitness in natural habitats. Restoration outcomes depend on fitness related performance over the entire life cycle. Thus, reproduction is important to measure and understand in order to evaluate the efficacy of different oyster seeding strategies in restored habitat.

Recruitment of wild-set oysters (newly-settled “spat”) has been monitored in the

TZ-HB region since 2008 and robust spat recruitment was observed in six of seven

IV-7 monitored years (Carthan and Levinton 2013; McFarland and Hare 2018; Starke 2010).

Additionally, ongoing research has shown that spat collected proximal to the wild TZ-

HB, when transplanted along the HRE salinity gradient and experimentally compared to

hatchery-produced oysters from mesohaline parents, show relative performance that

depends on location (McFarland and Hare, 2018). Preliminary findings suggest that the wild TZ-HB have superior growth and survival at the upper estuary low-salinity sites, but

equal or worse performance at harbor and Jamaica Bay (lower estuary) sites relative to an

aquaculture strain (McFarland and Hare 2018). These results suggest that the TZ-HB

oyster population may be locally adapted to their low salinity habitat as a result of long-

term isolation there. The research described here expands on these ongoing experiments

by quantifying and comparing one aspect of reproduction, the timing and extent of gonad

maturation.

The objective of this study is to understand how salinity and oyster strain (genetic

ancestry) affects reproductive timing for 2-year old oysters in the Hudson River Estuary.

Most restoration-related eastern oyster research has studied aspects of performance in the early parts of oyster life history (e.g., spat growth and survivorship), whereas reproduction has rarely been measured in natural eastern oyster populations. Studies have shown that salinities anywhere from 5-10 practical salinity units (psu) significantly slow- down or even inhibit the process of gametogenesis (Kennedy et al. 1996; Butler 1949); however, the same authors showed how tolerance to low salinity depends on the salinity at which broodstock were conditioned (at which gametogenesis occurred). It is challenging to study reproduction, especially in temperate waters where oysters need two

years to reach reproductive age. Histological measurement of gametogenesis stage is

IV-8 expensive, and fecundity is technically challenging to measure accurately (Kennedy et al.

1996; Loosanoff and Davis 1952; Mroch et al. 2012). The lack of studies on reproduction leaves great uncertainty about variation in critical fitness components: year of first reproduction, seasonal timing of spawn, age- and size-specific fecundity, and larval survivorship. Thus, sustainable oyster restoration in the HRE requires understanding of oyster response to a broad range of environmental conditions at every life stage: larva, juvenile, and adult.

Oyster reproduction is cyclic, separated into three stages generating the next generation of sessile juveniles: gametogenesis, spawning and external fertilization, and larval growth and settlement (Kennedy and Battle 1964). Gametes are synchronously released by adults so that zygotes are formed by external fertilization and then develop into early-stage larvae. The larvae spend 2-3 weeks in the water column, dispersing by tidal currents and advection. Available food resources (phytoplankton) will determine the duration of the larval period before settlement competency is developmentally accomplished. The eventual settlement location of oyster larvae is hard to predict because environmental factors such as water currents and predators vary among sites. After oyster

larvae become competent to settle, they ‘swim’ to the bottom using ciliary propulsion,

where they find a hard substrate and metamorphose into sessile spat (Kennedy et

al.1996).

Eastern oysters are protandrous hermaphrodites, maturing first as males and later

becoming females (Kennedy et al. 1996). In temperate waters C. virginica spends most of

the winter as an undifferentiated sex (Kennedy et al.1996). Sex differentiation, spawning,

and reproduction follow a seasonal pattern. Gametogenesis builds gonad tissue through

IV-9 the Spring, and at a temperature between 20-30°C, the oysters begin to release gametes

into the water column (Kennedy et al. 1996; Barber et al. 1991). The oysters spawn

synchronously, so when one releases its gametes, nearby oysters are triggered to release

their own. (Mann et al. 2014; Loosanoff and Davis 1952). After the oysters have

spawned, they resorb their follicles and re-enter the sexually undifferentiated stage in the fall. At this time follicles become very small and separated from each other by lots of connective tissue. At this stage, no differentiated sex cells are visible; however, some slight gametogenic activities might still be occurring. As it gets colder, the oysters enter a torpor stage in which gonads are dormant until the next Spring (Kennedy et al.1996;

Loosanoff 1942).

Because of its importance for aquaculture, the thermal induction of oyster spawning is well studied. This study will examine the second-order effect of salinity in a natural system (Volety 2008; Loosanoff and Davis 1952). Studies in a natural system are vulnerable to potentially confounding variables such as nutrient levels and pollution that can make it challenging to identify specific environmental triggers or thresholds. The effects of salinity, in particular, are important to study in the HRE because of the potential utility of transplanting oysters from the wild remnant low-salinity TZ-HB population to southern, moderate to high salinity portions of the estuary for restoration.

Oysters are tolerant of a broad range of salinities (~ 5 – 40 psu), so they typically occur across much of the salinity gradient within estuaries. Results from the HRE may provide observations that are more generally true, at least in temperate estuaries.

There is agreement among previous research that lower salinities limit or even inhibit reproduction in oysters. A 1949 study found that oyster reproduction is entirely inhibited

IV-10 under 6 psu (Butler 1949).) A study by Volety found that oysters at a salinity below 14 psu demonstrated poor spat recruitment, excessive valve closure and slower growth. The smaller oysters were less fecund, producing significantly fewer eggs than their larger, higher salinity (14-28 psu) counterparts.

Relative to the limits suggested by these early studies, it seems contradictory that the wild TZ-HB population currently exists at salinities ranging from 0-12 psu, depending on season. The TZ-HB population’s existence suggests that C. virginica reproduction might be more phenotypically plastic than originally thought, or that the

TZ-HB population may be uniquely adapted to the low salinity environment.

Based on this literature and the observation that peak spat recruitment is often in

September for the TZ-HB wild population (McFarland and Hare 2018), the hypothesis tested in this study was that oyster gametogenesis is delayed in lower salinity water relative to higher salinities in the Hudson River Estuary. The delay might also be related to the timing of Spring water warming, which is delayed in the upper estuary relative to the lower estuary. This study evaluates gonad maturation relative to three main factors: sex, location (i.e., salinity), month, and oyster strain in the HRE.

METHODS

As stated above, temperate latitude eastern oysters require two years of growth before they become reproductive. Therefore, the objective of this project was to study the eastern oyster cohorts out-planted as one-month old juveniles in August 2016 by the Hare

Lab. Over the past two years, the Hare Lab has monitored survivorship and growth for three oyster strains (3 experimental cohorts with different parentage) maintained in

IV-11 experimental cages hung from docks. These 2016 cohorts are independent replicates of

initial experiments run from 2015 to 2017 at a subset of the sites (McFarland and Hare

2018).

Four outplant sites were used for

sampling 2 year old experimental oysters in

this study (Figure 1). They included “river”

sites at Hastings-on-the-Hudson (HH) and

Yonkers Science Barge (SB), a “harbor” site

located at Red Hook, Brooklyn (RED), and a

Jamaica Bay site at the Sebago Canoe Club in

Paerdegat Basin (PGB). Temperature and

salinity were recorded every two hours near

each of these sites using YSI 600 sondes

deployed by the Hare Lab or using a nearby

Hudson River Environmental Conditions Figure 1. Map of Hare Lab oyster sampling study Operating System (HRECOS) site. Both the sites along the HRE salinity gradient. sondes and the experimental cages were hung

1m below a dock or were on a rocky bottom in the low intertidal zone (Hastings). Data from Piermont Pier, which is near SB and HH was obtained from the HRECOS (http://hudson.dl.stevens-

tech.edu/hrecos/d/index.shtml). Sonde data from Kingsborough Community College

(KCC) in Jamaica Bay was used to represent environmental conditions at the nearby

Jamaica Bay (PGB) site.

IV-12 There are three different cohorts being compared in this study: wild, hatchery and

aquaculture, all the same age (within 1 month). The wild oyster sample was obtained as

August or September wild-set recruits on bivalve shell in the vicinity of the wild TZ-HB

remnant population. The hatchery cohort was produced by strip-spawning wild Martha’s

Vineyard oysters from a moderate salinity lagoon (similar to methods in McFarland and

Hare 2018). The larvae were then cultured by the Martha’s Vineyard Shellfish Group and

sent to the Cornell Cooperative Extension hatchery in Southold, New York to set on shell

in an upwelling system for three weeks during August 2016 before out planting to the

HRE cages. Seed oysters from a domesticated aquaculture strain also were acquired in

August 2016. Aquaculture seed and hatchery-produced oysters were kept in small-mesh bags until large enough to be contained in poly mesh netting bags.

Tappan Zee Wild Adult Collection and Transplant

Wild TZ-HB adults were desired in order to transplant them to different salinities and measure resulting variation in their gonad maturation. The oysters were dredged in cooperation with Dr. Tiffany Medley, using Monmouth University’s R/V Seahawk. A total of 276 live, wild adult TZ-HB oysters were dredged on June 15, 2018 for this study.

The oysters were stored in a cooler with water from the dredging location overnight. On

June 16, the oysters were randomly distributed in groups of 50-65 to the following sites:

HH, SB, LM, RED, and PGB (Fig 1). Unfortunately, logistical constraints prevented collections before June 15. It was decided that June 15 was too far into the reproductive maturation season to cleanly interpret reproduction after transplant, so the TZ-HB wild

IV-13 adults were monitored for variation in growth rate and survivorship and held for

experiments in 2019.

Monitoring of Oysters and Sampling for Histology

Samples for histology were collected three times during the summer of 2018: June

15-21, July 8-13, and August 8-13. At each site visit, growth and survivorship were measured for each cohort. Shell height was measured to the nearest millimeter using

Mitutoyo Calipers (Mitutoyo America Co., Aurora, IL, USA). Fouling organisms were cleaned off cages at each visit, but not from oysters unless they interfered with measurements. For histology, a maximum of 20 oysters were collected from each cohort at each site (Table 1). The number collected was limited by cohort abundance and budget.

Planned sample size variation prioritized July and August samples, and poor survivorship

at some sites prevented complete sampling.

Table 1. Counts of oysters sampled for histology from each cohort at each site during each month. Site HH SB RED PGB

Jun Jul Aug Jun Jul Aug Jun Jul Aug Jun Jul Aug

AQ 13 20 20 13 20 20 11 18 20 13 19 20

Hatchery 13 12 12 13 20 20 0 0 0 13 18 20

Wild 13 20 19 13 20 20 0 8 6 13 20 20

IV-14 Histology Preparation and Condition Index

Oysters were transported in coolers with ambient HRE water to the Fort Totten

Urban Field Station lab (US Forest Service) where dissection and preservation was done.

Lab work was performed as close to sampling date and time as possible in order to ensure that the histology results reflected the oyster’s response to their respective site’s environment, not the environment of the cooler used to transport them. Most processing was same-day, or at most the morning after collection.

The oysters were cleaned with a wire brush in order to remove any barnacles, dirt or plant tissue. Shell height was measured (again, repeating field measures) and then the oyster was shucked carefully so as to not mangle the oyster’s tissue. Using a scalpel, the oyster was removed from the shell, taking care to not damage the soft tissue. For each oyster, a new razor blade was used to make a single clean, diagonal cut across the body of the oyster, and then another cut along the same axis, but 4mm from the initial cut

(Figure 2). The oyster slice was placed in a labeled plastic cassette and preserved in

Davidson’s Solution for 7 days (Volety 2008). As much as possible, the position of the cross section was standardized among all individuals. After fixation and a 70% ethanol wash, the samples were sent to the Cornell University Animal Health Diagnostic Center for histology slide preparation with hematoxylin and eosin staining.

In bivalves a useful indicator of overall physiological condition is the condition index, a measure of the ratio of dry tissue weight to dry shell weight. Excess energy that is not spent on reproduction or daily metabolism is invested into the oyster’s biomass.

IV-15

Figure 2. Shucked oyster anatomy showing orientation (parallel lines) of the section taken for histology preparation (Howard et. al, 2004). Whereas shell growth is cumulative with age, soft body tissue shows seasonal mass

variation on top of cumulative growth; spring-summer gonad maturation often generates a peak in soft tissue mass and in condition index (Ruiz et al. 1992). Environmental stressors can cause oysters to invest less in growth, thus affecting condition index (Volety

2008).

Condition index was measured using all soft oyster tissue except the histology slice. Thus, CI in this study may have inflated variance due to variation in the proportional mass of histology slices, and it is not comparable to CI in other studies.

Oyster soft tissue was placed in a labeled, pre-weighed aluminum dish and along with both shells was placed in a drying oven at 70 °C for approximately 36 hours. Condition index was determined from the dry weights, using the following formula (Lucas and

Beninger 1985; McFarland and Hare 2018).

dry tissue weight (g) CI = × 100 dry shell weight (g)

IV-16 Histology Slide Analysis

The histology slides were examined with a compound microscope under 10X and

40X magnification to determine sex and gametogenic stage. Gonad index (GI), a method for representing gametogenic stage, was scored using the criteria described in Table 2.

Pictures of each gonad stage for each gender were taken using Leica LAS Microscope

Software. The pictures can be found in the Appendix.

Table 2. Description of Gonad Index stages (derived from Volety 2008). Stage Description 0 Neuter or indeterminate sex despite good histology preparation; no presence of follicle or connective tissue 1 Determinate sex indicating beginnings of gametogenesis; no mature gametes visible 2 Females: no more than one-third mature eggs relative to developing eggs Males: fringe follicles starting to accumulate mature gametes; low density of sperm 3 Females: no more than one-half mature eggs relative to developing eggs Males: visible connective tissue; visible sperm tails in center of some follicles 4 Females: mostly mature polygonal eggs and distended follicles with some developing eggs still visible Males: small amount of visible connective tissue; compact follicles; sperm tails visible in most follicles 5 Females: only mature polygonal eggs; no empty spaces from spawning gametes Males: sperm have visible tails; uniformly high sperm density in packed follicles; no visible connective tissue 6 Active spawning is occurring; follicle structure is disrupted Females: general rounding of eggs; a few empty spaces from released eggs Males: reduction in sperm density; sperm tails less visible 7 Follicles one-half depleted of gametes Males: sperm area reduced to one-half of gonadal area 8 Follicles two-thirds depleted of mature gametes; further reduction of gonadal and connective tissue area 9 Only residual gametes remain; determinate sex; further reduction of gonadal and connective tissue area 10 Gonads devoid of most or all gametes; connective tissue is still visible but very minimal; sex not always determinate Females: one or two visible eggs remain Males: connective tissue has a few remaining sperm to allow for sex identification

IV-17

Statistical Analyses

A random linear model was created based on the gonad index results to predict multiple interaction effects of sex, location, strain, and month on GI. For this model, the

GI values from 6-10 were relabeled to 5-1 (6 = 5, 7 = 4, 8 = 3, 9 = 2, 10 = 1) to allow for averages that reflect the true gonad condition. For example, two oysters with gonad indexes of 2 and 9, respectively, would have a meaningless “unfolded” average GI of 5.5.

Neither of the oysters are at peak reproductive spawning, yet that is what the “unfolded” average would suggest. For the folded index with 9=2, the “folded” average in this example would be 2, reflecting the oyster’s true relation to peak reproduction (Volety

2008). A dummy variable [0,1] was assigned to the GI values before “folding”, called ‘GI trend’ to represent the gametogenesis and spawning GI distributions. GI values 1-5 were assigned GI trend=0 to represent oysters that are still actively growing gonads, or gametogenesis. GI values 6-10 were assigned GI trend=1 to represent oysters that are spawning out or resorbing their gametes. GI values of 0 were ignored for the GI trend variable because it was impossible to confirm if an oyster with a GI of 0 had not undergone gametogenesis or had completely resorbed gametes. A one-way Analysis of

Variance (ANOVA) was performed on the model that included the effects of sex, location, strain, month, and GI trend on GI.

Results were deemed significant at P < 0.05. Predicted means and confidence intervals were generated using the random linear model and plotted for select combinations of variables. A Spearman’s test of correlation was used to determine the

IV-18 relationship between GI and shell height. Statistical analyses were conducted using

RStudio (RStudio Team 2016).

RESULTS

Temperatures ranged from 16-28 °C in the summer months. Surprisingly, HH had the highest temperatures, even though it was the northernmost site. A study by Loosanof identified 20 °C as the temperature at which oysters spawn, which sets up the prediction that when oysters cross this threshold, they will begin to spawn (Loosanof and Davis

1952). HH is the first site to cross the threshold, followed by KCC and RED. SB temperature data was not available in the earlier part of the study period. Environmental data was not available for SB until the beginning of July.

Figure 3. Averaged weekly temperature data with standard error bars. The dashed horizontal line represents the temperature threshold for spawning in temperate oysters (Loosanof and Davis 1952)

IV-19 Predictably, salinity was lowest at the two southernmost sites: RED and KCC.

Their salinities ranged from 16-27 psu. HH had the lowest salinity, followed by RED.

The salinity data matches the known salinity gradient in the estuary (Figure 1).

Figure 4. Averaged weekly salinity data with standard error bars.

Table 3. ANOVA results for effect on GI by all tested variables. ** denotates a significant P value. Variable P values Sex 0.2227975 Strain 0.0842607 Month 7.725e-10 ** Location 0.9357032 GI trend 3.508e-05 ** Strain*GI trend 0.9443797 Month* GI trend 5.245e-07 ** Location*GI trend 0.0464173 ** Month*Location 4.044e-09 ** Month*Location* GI trend 0.0008605 **

IV-20

In the one-way Type III ANOVA test, strain did not have a significant effect, even in interaction with GI trend. As expected, due to temperate latitude seasonality, GI varied significantly between months, and there was also an interaction between month and GI trend. Location had a significant effect on the GI means only in the context of an interaction with GI trend or GI trend and month.

Graphs of means and 95% confidence intervals were created in order to observe

GI distribution differences. If the confidence intervals did not overlap, it was determined that the difference between the compared groups was significant. The ANOVA results

(Table 3) reveal that there is a significant difference between the GI trend values 0 and 1, or gametogenesis and spawning GI patterns, respectively. Therefore, for visualization, the

GI data was ‘unfolded’ by separating gametogenesis (0-5) and spawning (6-10) means in

Figures 5, 6 and 7. The ANOVA test also demonstrated that strain had no significant effect, so it was not included in the confidence intervals.

The June gametogenesis data demonstrates a significant difference in gonad maturation between the highest salinity site (PGB) and the two lowest salinity sites (HH and SB). The lower salinity sites, which are closer to the location of the wild remnant population, had a higher gametogenesis GI, indicating that the oysters were farther along in the gametogenesis than the oysters at the lower salinity sites. The site with most advanced reproduction was Science Barge (SB), with almost half the sample in the spawning stage. The other sites had very few individuals in the spawning stage. The RED data for this and other months show large confidence intervals due to especially high variance in GI values observed in oysters at this location.

IV-21

Figure 5. Gonad index means and 95% confidence intervals plotted for each site for the month of June. Labels of the number of gametogenesis and spawning oysters for each site are to the right of their corresponding interval and mean. A vertical dashed line separates the two sides of the ‘folded’ GI, approximately corresponding to gametogenesis and spawning stages.

Red Hook was the outlier in July, even after accounting for large confidence

intervals around the GI means. In the gametogenesis GI stages, there was no difference

between the lower and higher salinity sites. Compared the June, the oyster sample at PGB

is much more advanced in terms of gametogenesis GI. PGB in Jamaica Bay had a higher

number of spawning oysters than river sites and a significantly higher GI spawning mean

(GI = 7.0) than HH. The spawning GI average for HH and SB were approximately the same in June and July. Overall, there are more oysters in the spawning stage in July across all sites.

IV-22

Figure 6. Gonad index means and 95% confidence intervals plotted for each site for the month of July. Labels of the number of gametogenesis and spawning oysters for each site are to the right of their corresponding interval and mean. A vertical dashed line separates the two sides of the ‘folded’ GI, approximately corresponding to gametogenesis and spawning stages.

In August, the river sites (HH and SB) continued to show evidence of spawning in less than half the sample, whereas more than 2/3 of the oysters at RED and PGB had spawned. There is no difference between sites in the gametogenesis stage. As seen in

July, spawning PGB oysters were significantly more advanced than at river sites with an average GI (8.64) closer to 10 than HH (7.42) and SB (7.69).

IV-23

Figure 7. Gonad index means and 95% confidence intervals plotted for each site for the month of August. Labels of the number of gametogenesis and spawning oysters for each site are to the right of their corresponding interval and mean. A vertical dashed line separates the two sides of the ‘folded’ GI, approximately corresponding to gametogenesis and spawning stages.

To explore whether reproductive timing might be influenced by oyster size, shell height was plotted against gonad index and separated by month. A Spearman’s test of correlation was performed for each relationship with GI for June, July and August. In

June and August there was no significant monotonic relationship between oyster size and gonad maturation stage (Figures 9,11). For July, the P value was 0.025 and the rho value was 0.160 for a positive correlation (Figure 10).

IV-24

Figure 8. June GI vs. shell height. Figure 9. July GI vs. shell height.

Figure 10. August GI vs. shell height.

IV-25 To observe the relationship between oyster condition and gonad maturation, GI was plotted against condition index (Figure 11). Shucking procedures occasionally resulted in broken shell that was unrecoverable for condition index measurement. As expected, Figure 11 shows that condition index peaks when the gonad index is around 5 or 6, representing the ripest reproductive stages with the greatest soft tissue mass.

Figure 11. Gonad Index vs. Condition Index.

Growth rate and survivorship of dredged TZ-HB oysters was measured in October as well as June - August. The wild TZ-HB oysters continued to grow once they had been transplanted (Figure 13). At RED, there was a 45% mortality rate between August and

October. Mortality was very low at all other transplant sites (< 7.4%).

IV-26

Figure 12. Shell height over time (growth) for wild dredged TZ-HB

adults transplanted to five sites in the Hudson River

Estuary.

DISCUSSION

The results of the random linear model and confidence intervals do not lend support to the simple hypothesis that gametogenesis is delayed in lower salinities along the Hudson River Estuary; however, there is an observable difference between the reproductive timing of oysters at different locations along the salinity gradient. In June, few oysters were spawning and the locations with lower salinity had more oysters closer to peak spawning (GI = 5) on average than the oysters at higher salinity locations. Later in the reproduction season, in July and August, the post-spawn oysters at high salinity sites had significantly higher GI values (more depleted and reabsorbed gonads) relative to those at low salinity sites. Together, these observations suggest that oysters at higher salinities had a delayed, but more rapid spawning period relative to oysters at low salinity. This was also apparent in August by the proportion of oysters that had not yet

IV-27 spawned. River sites at low salinity had more than half pre-spawned individuals whereas the harbor (RED) and Jamaica Bay (PGB) sites were mostly spawned-out. In other words, the northern low salinity sites seemed to start spawning earlier and extend reproduction later, compared with the harbor and Jamaica Bay sites. These findings for

River sites are consistent with the September-October spat set reported for the TZ-HB area (McFarland and Hare 2018).

The four sites were arranged along a salinity gradient, but spatially positioned such that the two northern sites (HH and SB) had similar and significantly lower salinity values (5 – 15 psu) than the southern sites (18 – 26 psu; Figure 4). There was some temperature variation among sites, but at the first collection in June the range of temperatures was 18 – 21 °C across sites, close to or above the threshold temperatures that have been reported from temperate waters. It is possible that during April and May, when both average temperature and salinity gradients are steeper than in June, early gametogenesis may have been more advanced at the mesohaline sites, which crossed the temperature threshold earlier than the southern sites (Figure 3; Loosanof and Davis

1952).

Overall, the observed temporal pattern of reproduction documented over the summer months in the Hudson River Estuary supports the expectation that oysters in temperate regions will spawn with a major peak during the summer months and enter a non-reproductive stage in the winter months (Cox and Mann 1992; Loosanof and Davis

1952). The sampling constraints in this study made it impossible to evaluate whether some oysters redeveloped gametes for a Fall spawn. This is one of the first reports that shows the degree of temporal discreetness of a seasonal peak in reproduction to be

IV-28 dependent on salinity. This finding implies that hatcheries will be more successful getting oysters to condition uniformly if done at moderate to high salinity.

In the ANOVA test, Location did not show significance on its own, but this is likely because not all locations were different from each other. Some of the locations had very similar GI data (HH and SB, RED and PGB). The ANOVA does not test for differences between individual locations, which is why means and confidence intervals were calculated. When Month and GI trend were taken into account with Location, represented by Month*Location* GI trend, and there was significance in the linear model.

Data from September and October would have been valuable to observe how long it takes the oysters to completely spawn and enter their torpor period. A longer sampling period would also be interesting to see if any locations supported a Fall resurgence in gametogenesis and spawning, as reported in the mid-Atlantic and northern Gulf of

Mexico (Southworth and Mann 2004; Hayes and Menzel 1981).

There is potential for measurement error in this study related to the GI scoring.

Even with clearly defined categories, there is some subjectivity in scoring intermediate specimens. GI values of 0 and 10 do not contain any gametes and can be confused with each other. GI= 10 is slightly different in that some follicle tissue remains. In addition to multiple GI evaluation trials, the oysters with GI values of 0 and 10 were closely examined to determine the correct scoring. Incorrect scoring between the 0 and 10 is possible, which would cause the average GI values and confidence intervals to be skewed. GI values of 0 were not included in the ANOVA test or confidence intervals, but values of 10 were because of the remaining follicle tissue. Small and uneven sample sizes

IV-29 limit the ability to precisely quantify and compare gametogenesis state. Most of the broad

confidence limits in Figures 4, 5 and 6 are due to especially small sample sizes.

The results of this study offer a unique examination of strain, month and location

effects on gametogenesis timing for C. virginica along a temperate estuarine salinity

gradient. It is not clear how to reconcile results from this study with the findings of Butler

(1949), who found that gametogenesis was inhibited in 90% of his study population when

salinity was below 6 psu. At the northern most site included in this study, HH, average

weekly salinity reached briefly down to 5 psu or slightly below. Previous studies have

found that the TZ-HB region regularly experiences Spring salinities below 5 psu,

sometimes for multiple consecutive weeks during the spring snow melt (McFarland and

Hare 2018). Thus, oysters at the low salinity River sites in this study may have

experienced inhibition of gametogenesis earlier in the year before sampling began. If that

is true, it implies that oysters can phenologically compensate for environmental

conditions that delay gametogenesis (cold temperature and/or low salinity).

The aquaculture, hatchery and wild strains were all able to spawn gametes and

showed similar gametogenic phenology. This result suggests that the variation in

reproductive timing observed in this study is not genetic, but a result of acclimation and phenotypic plasticity. Therefore, with respect to reproduction, transplanted oysters can be expected to adjust the timing of their gametogenesis to acclimate to new salinity and temperature conditions. The experimental oysters studied developed from an age of ~4 weeks at the site from which they were sampled. Uncertainty remains about how long acclimation would take after a transplant and whether there are key developmental periods that facilitate acclimation. These findings are based on a comparison of three

IV-30 temperate oyster strains, and therefore do not contradict the previous conclusion, based on common garden studies, that genetically-based physiological races occur between more latitudinally disparate source populations (Barber et al. 1991).

The significant but weak monotonic relationship between gonad index and shell height suggests that larger oysters progress through the reproductive cycle sooner than smaller oysters. The power to detect a correlation was highest in July, with a weak monotonic relationship was observed in (rho = 0.160). This positive correlation implies that larger same-age oysters are slightly more advanced in gametogenesis (and resorption) than smaller oysters at a given time, on average. Mostly studies have measured the positive correlation between fecundity (egg count) and shell height (Mann et al. 2014; Mroch et al. 2012), making these findings unique.

Condition index is a ratio of tissue weight to shell weight, varying with seasonal changes in physiological status, especially reproduction (Volety 2008). When oysters begin to produce gametes, their soft tissue weight increases, thus increasing the value of the condition index. When plotted against GI, condition index increased as GI increased from 1-5. Interestingly, condition index did not increase much between GI 5 and 6, even though by definition, oysters only began to release gametes at GI 6 (Fig. 11). This implies that gonad mass is not increased during the final stage of gametogenesis.

Although GI was not measured for the transplanted TZ-HB wilds, growth and survivorship provide valuable information about how these wild adult oysters performed in salinities different than where they settled and matured. There does not appear to be any evidence for a dredging or transplant affect depressing growth, even when transplanted to higher salinity at PGB. Growth and survivorship were not significantly

IV-31 different among locations during the summer months, but examination in October

revealed performance differences. The oysters at RED had significantly lower growth and

survivorship in October. Over 45% of oysters died at RED whereas most oysters survived

at the other 4 sites. The pattern of poor survivorship at RED for TZ-HB wild adult oysters

also was observed in other cohorts at this site (McFarland and Hare, 2018). Moderate

salinity at RED is ideal for eastern oysters, but there are many factors that could account

for this trend such as harbor pollution, low nutrient availability, high sediment load and

disease. More in-depth experimentation and environmental monitoring at this site would

be necessary to pinpoint the cause of poor survivorship.

Besides mortality at RED, the growth and survivorship of TZ-HB oysters at the

other transplant sites demonstrates potential for TZ-HB oysters to be a source for

successful transplantations. Although wild local oysters are preferred for restoration

efforts to capitalize on any local adaptations, comparisons among strains here indicate

that gametogenic phenology may be suitable in any temperate oyster strain, whether wild

or hatchery produced, if allowed a suitable period for acclimation. Further study would

need to be done to compare other fitness components and to examine the population

genetic effects of using hatchery-produced or domesticated oysters in restoration. Firm

conclusions will require a more complete evaluation of fitness for transplanted oysters

over the entire life cycle. These experimental transplants have created the opportunity for

further studies to determine how transplanted adults from a low-salinity area survive and grow along the HRE salinity gradient, and by spawning them in the future it will be possible to test relative performance of their larvae. This study is only the beginning of

IV-32 the necessary amount of information to inform conservation and repopulation efforts in the Hudson River Estuary.

ACKNOWLEDGEMENTS

We would first like to thank the Hudson River Foundation for the opportunity and funding of the Tibor T. Polgar fellowship. Katie McFarland, Sarah Gisler, and Tiffany

Medley supported this research by offering advice, help with field work, and data analysis consults. We are grateful to Captain Jim Nickels for leading a successful oyster dredging trip. Thanks also to the Cornell Statistical Consulting Unit for advice on the data analysis and graphical representation for this research. We thank the U.S. Forest

Service Fort Totten Urban Field Station for providing housing and a lab for this field work. Finally, this work would not be possible without the generous site hosts that keep

Hare Lab experimental oyster cages safe.

IV-33 REFERENCES

Appleyard, S.A., and R.D. Ward. 2006. Genetic diversity and effective population size in mass selection lines of pacific oyster (Crassostrea gigas). Aquaculture 254:148– 159.

Baggett, L.P., S.P. Powers, R. Brumbaugh, L.D. Coen, B. DeAngelis, J. Greene, B. Hancock and S. Morlock. 2014. Oyster habitat restoration monitoring and assessment handbook. The Nature Conservancy. Arlington, VA.

Barber, B. J., S.E. Ford, and R.N. Wargo. 1991. Genetic variation in the timing of gonadal maturation and spawning of the eastern oyster, Crassostrea virginica (Gmelin). The Biological Bulletin 181:216–22.

Brumbaugh, R.D., M.W. Beck, L.D. Coen, L. Craig and P. Hicks. 2006. A Practitioner’s Guide to the Design and Monitoring of Shellfish Restoration Projects: An Ecosystems Approach. The Nature Conservancy. Arlington, VA.

Butler, P.A. 1949. Gametogenesis in the oyster under conditions of depressed salinity. The Biological Bulletin 96(3):263-269

Carthan, R., and J.S. Levinton. 2013. Recruitment of oysters within the Hudson River Estuary. Section III: 1-28 pp. In S.H. Fernald, D. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2012. Hudson River Foundation, NY.

Coen, L.D., R.D. Brumbaugh, D. Bushek, R. Grizzle, M.W. Luckenback, M.H. Posey M, S.P. Powers, and S.G. Tolley. 2007. Ecosystem services related to oyster restoration. Marine Ecology Progress Series 341:303-307.

Cox. C., and R. Mann 1992. Temporal and spatial changes in fecundity of Eastern oysters, Crassostrea Virginia (Gmelin, 1791) in the lower James River, Virginia. Journal of Shellfish Research 11(1):49-54.

Hayes, P.F., and R.W. Menzel. 1981. The reproductive cycle of early setting Crassostrea virginica (Gmelin) in the northern Gulf of Mexico, and its implications for population recruitment. The Biological Bulletin 160(1):80-88.

Hornick, K., and L. Plough. 2018. Genetic impact of a large-scale eastern oyster supplementation program using native broodstock. Abstract from American Fisheries Society meeting.

Howard D.W., E.J. Lewis, B.J. Keller and C.S. Smith. 2004. Histological techniques for marine bivalve mollusks and crustaceans. NOAA Technical Memorandum NOS NCCOS 5, 218 pp.

IV-34 Kennedy, A.V., and H.I. Battle. 1964. Cyclic changes in the gonad of the American oyster Crassostrea virginica (gmelin). Canadian Journal of Zoology 42(2):305- 321

Kennedy, V.S., R.I.E. Newell and A.F. Eble (Eds.) 1996. The Easter Oyster Crassostrea virginica. Maryland Sea Grant College, College Park.

Kurlansky, M. 2006. The Big Oyster: History on the Half Shell. Ballantine Books, New York.

Loosanoff, V.L. 1942. Seasonal gonadal changes in the adult oysters, Ostrea virginica, of Long Island Sound. The Biological Bulletin 82(2):195–206.

Loosanoff, V., and H.C. Davis. 1952. Temperature requirements for maturation of gonads of northern oysters. The Biological Bulletin, 103(1): 80-96.

Lucas A., and G. Beninger. 1985. The use of physiological condition indices in marine bivalve aquaculture. Aquaculture 44:187–200

MacKenzie, C.L., Jr. 2007. Causes underlying historical decline in eastern oyster (Crassostrea virginica Gmelin, 1791) landings. Journal of Shellfish Research 26(4)

Mann, R., M. Southworth, R.B. Carnegie, and R.K. Crockett. 2014. Temporal variation in fecundity and spawning in the eastern oyster, Crassostrea virginica, in the Piankatank River, Virginia. Journal of Shellfish Research 33(1):167-176.

McFarland, K and M.P. Hare. 2018. Restoring oysters to urban estuaries: Redefining habitat quality for eastern oyster performance near New York City. PLoS ONE 13(11): e0207368.

Mroch, R. M., D.B. Eggleston, and B.J. Puckett. 2012. Spatiotemporal variation in oyster fecundity and reproductive output in a network of no-take reserves. Journal of Shellfish Research 31(4):1091-1101.

Stinnette, I., M. Taylor, L. Kerr, R. Pirani, S. Lipuma and J. Lodge. 2018. State of the Estuary 2018. Hudson River Foundation. New York, NY.

Peterson, C.H., J.H. Grabowski, and S.P. Powers. 2003. Estimated enhancement of fish production resulting from restoring oyster reef habitat: Quantitative valuation. Marine Ecology Progress Series 264:249–264.

Raj, S., and P.J. Reson. 2008. Oysters in a new classification of keystone species. Resonance 13:648-654.

IV-35 RStudio Team. 2016. RStudio: Integrated Development for R. RStudio, Inc., Boston, MA URL http://www.rstudio.com/.

Ruiz, C., D. Martinez, G. Mosquera, M. Abad, and J.L. Sànchez. 1992. Seasonal variations in condition, reproductive activity and biochemical composition of the flat oyster, Ostrea edulis, from San Cibran (Galicia, Spain). Marine Biology 112(1):67–74.

Southworth, M., and R. Mann. 2004. Decadal scale changes in seasonal patterns of oyster recruitment in the Virginia sub estuaries of the Chesapeake Bay. Journal of Shellfish Research 23(2):391+.

Starke, A.F. 2010. Restoration of the Hudson River oyster: A Physiological and Spatial Assessment of Crassostrea virginica’s Restoration in the Hudson River, NY. Master’s Thesis. Stony Brook University (Stony Brook, NY).

Volety, A.K. 2008. Effects of salinity, heavy metals and pesticides on health and physiology of oysters in the Caloosahatchee Estuary, Florida. Ecotoxicology 17(7). zu Ermgassen, P., B. Hancock, B. DeAngelis, J. Greene, E. Schuster, M. Spalding and R. Brumbaugh. 2016. Setting objectives for oyster habitat restoration using ecosystem services: A Manager’s Guide. TNC, Arlington VA. 76pp.

IV-36 APPENDIX

Photographs of oyster gonads taken on Leica LAS Microscope Software. “1F 25x” is a female oyster with a GI of 1 taken at 25x.

1F 25x 1F 100x

1M 25x 1M 100x

2F 16x 2F 80x

IV-37

2M 12.5x 2M 50x

3F 10x 3F 90x

3M 16x 3M 40x

IV-38

4F 10x 4F 63x

4M 12.5x 4M 40x

5F 10x 5F 90x

IV-39

5M 12.5x 5M 25x

6F 12.5x 6F 80x

6M 10x 6M 40x

IV-40

7F 12x 7F 80x

7M 16x 7M 50x

8F 12.5x 8F 80x

IV-41

8M 9x 8M 32x

9F 12.5x 9F 100x

9M 8x 9M 32x

IV-42

Lower resolution 10F photo unavailable 10F 63x

10M 12.5x 10M 55x

0 25x 0 80x

IV-43

BUGS ON DRUGS: THE INFLUENCE OF REDOX ENVIRONMENTS ON THE MICROBIAL DEGRADATION OF PHARMACEUTICALS IN THE HUDSON RIVER WATERSHED

A Final Report of the Tibor T. Polgar Fellowship Program

Michelle L. Zeliph

Polgar Fellow

Ph.D. Candidate, Microbial Biology School of Environmental and Biological Sciences Department of Biochemistry and Microbiology Rutgers, The State University of New Jersey New Brunswick, NJ 08901

Project Advisor:

Max M. Häggblom School of Environmental and Biological Sciences Department of Biochemistry and Microbiology Rutgers, The State University of New Jersey New Brunswick, NJ 08901

Zeliph, M.L. and M.M. Häggblom. 2020. Bugs on Drugs: The Influence of Redox Environments on the Microbial Degradation of Pharmaceuticals in the Hudson River Watershed. Section V: 1-48 pp. In S.H. Fernald, D.J. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

V-1

ABSTRACT

Pharmaceuticals and personal care products (PPCPs) are emerging contaminants

in aquatic ecosystems throughout the world. The central goal of this study was to

determine how the redox environment impacts the anaerobic biodegradability of PPCPs

by native microorganisms in anoxic estuarine sediments, using the Hudson River

watershed as a model. Long-term PPCP contamination in the environment was hypothesized to result in the enrichment of microbial communities capable of utilizing

PPCPs as carbon and energy sources. It was also hypothesized that microorganisms from

different points in the watershed and active under different redox conditions would have

different metabolic potentials to degrade PPCPs. Anaerobic microcosms were established

using site water and sediment collected from five sites within the Lower Hudson River

watershed to determine rates and extent of degradation. Microcosms were amended with

excess nitrate, sulfate, or no additional terminal electron acceptor and spiked with four

pharmaceutical compounds as model PPCPs and sampled periodically to monitor

degradation and transformation by high performance liquid chromatography. Anaerobic

sediment and site water microcosms amended with untreated sewage wastewater were

used to simulate a combined sewer overflow event, introducing a complex suite of PPCP

compounds. PPCPs were extracted by solid phase extraction and analyzed by liquid

chromatography–tandem mass spectrometry. Microbial communities capable of

degrading or transforming PPCPs were found in sediments from the Hudson River,

supporting the hypothesis that long-term contamination results in the enrichment of

microbial communities capable of utilizing PPCPs as carbon and energy sources. The

redox environment appeared to play a lesser role in the biodegradability of PPCPs.

V-2

TABLE OF CONTENTS

Abstract ...... V-2

Table of Contents ...... V-3

Lists of Figures and Tables ...... V-5

Introduction ...... V-8

PPCPs in wastewater and watersheds ...... V-8

Microbial degradation and biotransformation ...... V-11

Prominent PPCPs: analgesic and antipyretic compounds ...... V-11

Ecological relevance ...... V-14

Methods...... V-15

Site selection and sample collection ...... V-15

Establishment of microcosms for degradation assays ...... V-18

Analysis of degradation assays ...... V-19

Establishment of simulated combined sewer overflow (CSO) influent

incubations ...... V-21

Analysis of influent incubations ...... V-22

Results ...... V-24

Acetaminophen ...... V-24

Aspirin...... V-27

Diclofenac ...... V-27

Ibuprofen ...... V-31

Pharmaceuticals in site water ...... V-34

V-3

Simulated CSO influent incubations ...... V-34

Discussion ...... V-37

Future directions ...... V-41

Acknowledgements ...... V-42

References ...... V-43

V-4

LIST OF FIGURES AND TABLES

Figure 1 – Anaerobic degradation of select PPCPs in Arthur Kill sediment

cultures ...... V-13

Figure 2 – Anaerobic degradation of select PPCP compounds in Hudson River sediment

cultures collected at the 79th Street Boat Basin, Manhattan, NY ...... V-14

Figure 3 – Map of sampling sites ...... V-17

Figure 4 – Mean concentration of acetaminophen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of acetaminophen was observed in the sterile control (n = 1)...... V-25

Figure 5 – Mean concentration of acetaminophen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site.

Microcosms were amended with excess sulfate (10mM Na2SO4). No loss of

acetaminophen was observed in the sterile controls (n = 2)...... V-26

Figure 6 – Mean concentration of acetaminophen ± 1 SD in microcosms established with

sediment and site water from all five sampling sites. n = 3 for each sampling site.

Microcosms were amended with excess nitrate (10mM KNO3). No loss of acetaminophen was observed in the sterile controls (n = 2)...... V-26

Figure 7 – Mean concentration of aspirin ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of aspirin was observed in the sterile control (n = 1)...... V-27

V-5

Figure 8 – Mean concentration of diclofenac ± 1 SD in microcosms established with

sediment and site water from all five sampling sites. n = 3 for each sampling site. No

additional electron acceptor was provided. No loss of diclofenac was observed in the

sterile controls (n = 4)...... V-28

Figure 9 – Mean concentration of diclofenac ± 1 SD in microcosms established with

sediment and site water from all five sampling sites. n = 3 for each sampling site.

Microcosms were amended with excess sulfate (10mM Na2SO4). No loss of diclofenac

was observed in the sterile controls (n = 4)...... V-29

Figure 10 – Mean concentration of diclofenac ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site.

Microcosms were amended with excess nitrate (10mM KNO3). No loss of diclofenac was observed in the sterile controls (n = 4)...... V-30

Figure 11 – HPLC chromatograms depicting the loss of diclofenac (DCF) and rise of unknown metabolite (UNK) in STA 2 microcosm (site water with no amendment) over the 100 day experiment (A. Day 1; B. Day 14; C. Day 43; D. Day 100). Aspirin (ASP) and ibuprofen (IBU) loss was also observed. Upward and downward pointing arrows represent increasing or decreasing concentration, respectively...... V-31

Figure 12 – Mean concentration of ibuprofen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of ibuprofen was observed in the sterile controls (n = 4)...... V-32

V-6 Figure 13 – Mean concentration of ibuprofen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site.

Microcosms were amended with excess sulfate (10mM Na2SO4). No loss of ibuprofen

was observed in the sterile controls (n = 4)...... V-33

Figure 14 – Mean concentration of ibuprofen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site.

Microcosms were amended with excess nitrate (10mM KNO3). No loss of ibuprofen was observed in the sterile controls (n = 4)...... V-33

Figure 15 – Mean concentration of PPCPs ± 1 SD in simulated combined sewer overflow

(CSO) influent incubations established with sediment and site water from three sampling sites over a 40 day period. n = 3 for each sampling site/time point. n = 2 for sterile controls...... V-36

Table 1 – High performance liquid chromatography parameters for analysis of four pharmaceutical compounds...... V-20

Table 2 – LC-MS/MS data showing the suite of PPCPs present in the site water from the

MAN, BUC, and REN sites. Values represent mean concentration ± 1 SD. n = 3 for

MAN, n = 4 for BUC and REN sites...... V-34

V-7

INTRODUCTION

Pharmaceuticals and personal care products (PPCPs) include a diverse array of

thousands of chemical substances, including prescription and over-the-counter therapeutic drugs, veterinary drugs, fragrances, and cosmetics. Pharmaceuticals are used to prevent and treat health problems in humans or in agriculture to enhance growth or health of livestock. Personal care products, such as shampoos, moisturizers, perfumes, and deodorants, are generally cosmetic or thought to improve quality of life and contain a variety of compounds as active ingredients, preservatives, stabilizers, etc. These compounds are released directly or indirectly to the environment, but their long-term fate is poorly understood. They are biologically active and can thus pose adverse effects to aquatic biota. Because PPCPs have been detected in river and estuarine environments throughout the world, it is imperative to understand the ultimate fate of these compounds.

PPCPs in wastewater and watersheds

There are several ways in which PPCPs enter the environment, but the primary pathway is through incomplete removal during the wastewater treatment process.

Pharmaceuticals that are administered orally or through injection are excreted through urine and feces in both their unmetabolized form and as metabolized products. PPCPs that are applied topically and not fully absorbed, such as moisturizers and facial cleansers, enter wastewater when they are washed from the skin during bathing

(Daughton and Ruhoy 2009). In addition, unused pharmaceuticals are often disposed of down the drain or toilet, directly adding unmetabolized pharmaceuticals to wastewater.

Other significant sources of PPCPs in wastewater include waste from the pharmaceutical

V-8 industry, as well as healthcare facilities, such as hospitals and nursing homes (Phillips et al. 2010; Nagarnaik et al. 2012).

Wastewater treatment plants (WWTPs) are primarily designed to remove readily degradable organic material and pathogenic microorganisms. Although they are not specifically designed to remove PPCPs, some removal with varying efficiencies occurs depending on the chemical. Though many of these compounds are degraded or transformed during the wastewater treatment process, they can still be detected in effluent

(Ternes 1998; Miѐge et al. 2009; Gros et al. 2010) and in the receiving surface waters

(Ternes 1998; Gros et al. 2010; MacGillivray 2013). PPCP removal occurs via sorption to sludge, degradation, transformation, etc., but exact mechanisms are poorly delineated.

Many of the pharmaceuticals are excreted as various metabolites which can then be transformed back into the parent compound during the treatment process (Deo and

Halden 2013).

In addition to WWTPs, combined sewer overflows (CSOs) may be as relevant as

WWTPs as sources of PPCPs in riverine environments (Kay et al. 2017). CSOs are expected to become increasingly important sources of PPCPs in future years. In the northeastern United States, climate change is predicted to cause increased extreme precipitation events and earlier snow melt. Heavier rain and earlier snow melts will likely lead to more frequent flooding and combined sewer overflow events, with the consequence of more untreated sewage being discharged directly into receiving waters.

An increase in combined sewer overflow events will lead to an increase in PPCPs entering these receiving waters. New York State is especially vulnerable to increasing amounts of PPCPs being introduced into riverine ecosystems by CSOs as there are

V-9

currently about 800 CSO outfalls in the state, with many CSO outfalls located on the

Hudson River (NYSDEC 2018).

Kolpin et al. (2002) compiled an extensive survey of pharmaceuticals, hormones,

and other organic contaminants in U.S. water sources, showing that a broad range of

chemicals commonly occur at low concentrations downstream from intensive agriculture

or urbanization. A more recent global-scale analysis (Hughes et al. 2012) indicated that

pharmaceutical contamination is extensive due to widespread consumption and

subsequent disposal to rivers. With improved analytical methodology, the presence of

PPCPs in different environments is now well established and these compounds are widely

detected in surface, ground, and coastal waters, and even drinking water (see for

example, Ebele et al. 2017; Hollender et al. 2009; Yu et al. 2011; Li et al. 2013; Santos et

al. 2013; Moschet et al. 2013; Beretta et al. 2014; Lv et al. 2014; Sengupta et al. 2014;

Sun, Q. et al. 2014, 2016; Subedi et al. 2015; Thomaidi et al. 2015, 2016). This represents

a major concern in terms of their potential impact on the environment and human health.

PPCPs are biologically active and thus could adversely affect organisms within river and

estuarine ecosystems. A recent study by Cantwell et al. (2018) examined the presence of

sixteen pharmaceuticals in the Hudson River Estuary, in both the river transect and New

York Harbor. They found that at some sampling sites, especially those nearby WWTP outfalls, several pharmaceuticals, including sulfamethoxazole, carbamazepine, propranolol, and acetaminophen, were measured at concentrations reported to have chronic effects on aquatic biota. At their Kingston Sewage Treatment Plant outfall sampling site (RK 148.2), five additional compounds, including trimethoprim, ranitidine,

furosemide, gemfibrozil, and metoprolol, were measured at concentrations reported to

V-10

cause chronic effects. Further research must be carried out to better understand the

ultimate fate of these compounds in the Hudson River Estuary and other aquatic

environments in order to determine the risk they pose to the organisms that inhabit them.

Microbial degradation and biotransformation

It is widely known that microorganisms play an important role in the

biodegradation of pollutant chemicals in the environment (for reviews see Häggblom

1992; Häggblom et al. 2003; Fennell et al. 2011; Sun, W. et al. 2016); however, to date,

PPCP biodegradation experiments have mainly focused on aerobic conditions relevant to

activated sludge in wastewater treatment plants (Onesios et al. 2009; Onesios-Barry et al.

2014; Cydzik-Kwiatkowska and Zielińska 2016) and there is limited information on

PPCP biodegradability in anoxic environments (e.g., aquatic sediments or WWTP sludge digesters). Microbially mediated transformation products generated in these environments must be identified, and their ecotoxicity must be evaluated in order to gain a more comprehensive idea of the threat that PPCPs may pose to aquatic organisms.

Identifying what microorganisms are responsible for the degradation and transformation

of these compounds will also provide valuable information regarding the fate of PPCPs in

the environment.

Prominent PPCPs: analgesic and antipyretic compounds

Analgesic pharmaceuticals provide pain relief, while antipyretic pharmaceuticals

help to reduce fever. Over-the-counter nonsteroidal anti-inflammatory drugs (NSAIDs)

such as aspirin and ibuprofen, the prescription NSAID diclofenac, and the over-the-

V-11 counter pharmaceutical acetaminophen all fall into this class of compounds. Because of their widespread, frequent use, understanding the environmental fate and ecotoxicity of these compounds and their transformation products is essential. Several studies have reported on bacterial pure cultures isolated from activated sludge, wastewater, or sediment that can remove these frequently detected PPCPs. Bacterial strains have been found that are capable of degrading or transforming aspirin (Parales et al. 2017), acetaminophen (De Gusseme et al. 2011; Zhang et al. 2012), ibuprofen (Murdoch and

Hay 2005; Almeida et al. 2012; Marchlewicz et al. 2017), and diclofenac (Bessa et al.

2017). Wang and Wang (2016) provide a summary of isolated bacterial pure cultures capable of degrading specific PPCPs. Though extensive research has been done to determine the fate of these compounds in aerobic environments and the microorganisms responsible for their degradation and transformation, the understanding of their biodegradability under anoxic conditions is limited.

Preliminary work evaluated the anaerobic biodegradability of the analgesic and antipyretic pharmaceuticals aspirin, acetaminophen, ibuprofen, and diclofenac, which are among the most prevalent PPCPs in WWTP influent and are often detected in effluent and surface waters. The extent of anaerobic biodegradation or transformation observed in

Arthur Kill sediment microcosms, prepared using estuarine sediment from the heavily contaminated tidal strait in the Hudson-Raritan Estuary located between Staten Island,

New York and New Jersey, was dependent on both the PPCP and redox condition.

Aspirin and acetaminophen were biodegraded under the three redox conditions tested

(denitrification, sulfidogenesis and methanogenesis), but degradation rates were influenced by the electron accepting condition (Figure 1).

V-12

Figure 1: Anaerobic degradation of select PPCPs in Arthur Kill sediment cultures (Zeliph et al. 2016).

Anaerobic biodegradation of ibuprofen and diclofenac was not observed in these microcosms established with Arthur Kill sediment. Initial data from the Hudson River

(Figure 2) and the Jiulong River (Zeliph et al. unpublished) suggest that these same patterns of degradation may hold true for other sites. Thus, some PPCPs are biodegraded to various degrees in river and estuarine sediment by native microorganisms, but depending on the dominant redox condition of the sediments, PPCP compounds may be recalcitrant in the natural environment.

V-13

Figure 2: Anaerobic degradation of select PPCP compounds in Hudson River sediment cultures collected at the 79th Street Boat Basin, Manhattan, NY (Zeliph et al. unpublished).

Ecological relevance

Because of their widespread and frequent use, it is important to understand the ultimate fate of PPCPs in the environment. Much is known about the potential for degradation, strains capable of degradation, and metabolites associated with these compounds in aerobic environments; however, there is still much to be learned about the fate of these compounds in anoxic environments, such as aquatic sediments. If they are recalcitrant, several of these parent compounds pose a threat to aquatic organisms. For example, ibuprofen can inhibit the growth of phototrophs such as duckweed and phytoplankton, even at low concentrations (Pomati et al. 2004; Brausch et al. 2012).

Ibuprofen can also affect egg laying and hatching in higher organisms such as mollusks, V-14

as well as egg hatching and survival rates in fish (Pounds et al. 2008; Han et al. 2010). If

these compounds are being transformed, it is important to know what they are being

transformed into and how these transformation products may affect aquatic organisms.

For example, hydroxy-ibuprofens, metabolites of ibuprofen, and p-aminophenol, a

metabolite of acetaminophen, are believed to be more toxic than their parent compounds

(Marco-Urrea et al. 2009; Song and Chen 2001). A recent study also found that a

metabolite of diclofenac, diclofenac methyl ester, had significantly higher

bioconcentration than its parent compound in two keystone aquatic invertebrate species

and exhibited greater acute toxicity (Fu et al. 2020). The fate of these compounds in

anoxic environments must be investigated further.

The central goal of this study was to determine how the redox environment

impacts the anaerobic biodegradability of select PPCP compounds and the bacterial

communities responsible for their degradation in freshwater-estuarine sediments using

the Hudson River watershed as a model. Long-term PPCP contamination in the environment was hypothesized to result in the enrichment of microbial communities capable of utilizing PPCPs as carbon and energy sources. It was also hypothesized that microorganisms from different points in the watershed and active under different redox conditions would have different metabolic potentials to degrade PPCPs.

METHODS

Site selection and sample collection

The initial experiment proposed that sediment and site water would be collected

within the saline, brackish, and freshwater portions of the Hudson River (Figure 3), downstream of effluent discharge sites of sewage treatment plants in Manhattan, Buchanan, V-15

Poughkeepsie, Hudson, and Rensselaer; however, at two of the five sites (Poughkeepsie and

Hudson), there was little to no access to fine sediment, which was required to set up the

proposed experiments. Two alternative sites within the freshwater portion of the Hudson

River were chosen. In place of the Poughkeepsie site, sediment and site water were collected

at the Norrie Point Environmental Center in Staatsburg. Though there is not a WWTP outfall directly upstream of this site, Rondout Creek, which has been cited as a major contributor of sewage treatment derived micropollutants in the Hudson River (Carpenter

and Helbling 2018), is about 10 km upstream. Though the mouth of Rondout Creek is on

the western shore of the river and Staatsburg on the eastern shore, this stretch of the

Hudson River is relatively narrow (only 1 - 3 kilometers at points) and Rondout Creek may influence the concentration of various pharmaceuticals there. In place of the Hudson site, sediment and site water were collected in Germantown, downstream of their WWTP outfall. Though the community of Germantown is small and likely does not discharge high concentrations of pharmaceuticals, especially in comparison to the Hudson WWTP, this site has not been evaluated in previously published work studying pharmaceuticals in the Hudson River Estuary. At each site, three independent sediment and site water samples were collected 10 – 30 m apart. Surface sediment samples were taken from below the sediment-water interface and site water samples were taken at a depth of 0.5 m.

Samples were then transported to Rutgers, The State University of New Jersey in New

Brunswick, NJ and stored at 4°C for 1 – 3 days.

V-16

Figure 3: Map of sampling sites (adapted from NYSDEC 2020). V-17

Establishment of microcosms for degradation assays

Microcosms were established using the sediment and water collected from each of

the sampling sites. Each sediment sample was added to the corresponding site water

sample to make a 10% sediment (v/v) slurry. To prepare anaerobic cultures, the slurries

were sparged with nitrogen gas for 20 minutes before the addition of 1 ml/L of a 1 g/L

resazurin stock solution (redox indicator). 25 ml of each slurry was then transferred to three 60 ml serum bottles (with N2 gas streaming into them). One sediment slurry

microcosm was amended with 10 mM KNO3 (excess nitrate), one was amended with

10 mM Na2SO4 (excess sulfate), and the final microcosm received no additional terminal

electron acceptor. The serum bottles were capped with 20 mm blue butyl rubber stoppers

and aluminum rings and crimped closed to maintain an anoxic environment. Using sterile

anaerobic technique, the microcosms were then spiked with an aqueous solution of

acetaminophen (ACT), aspirin (ASP), and ibuprofen (IBU) to achieve a final

concentration of 200 μM of each pharmaceutical. Because diclofenac (DCF) is not very

soluble in water, it was dissolved in methanol and added to the microcosms using a

different technique. Before the sediment slurry was added to each serum bottle, a small

amount of silica was added. Enough diclofenac solution was then applied to the silica to

achieve a final concentration of 200 μM and the methanol was allowed to evaporate in a

fume hood for one hour. The sediment slurry was then added. After spiking with the

pharmaceuticals, microcosms designated sterile controls were autoclaved for 30 minutes.

The established cultures were incubated at room temperature in the dark.

V-18

Analysis of degradation assays

Samples were collected from the microcosms on Days 1, 8, 14, 22, 29, 43, and

100 of the experiment. A sample of 0.7 ml was removed using sterile syringes and 18G needles that had been purged with nitrogen gas. To extract any pharmaceuticals that might be sorbed to the sediment, 0.7 ml methanol was added to the sample. The samples were then shaken for 30 minutes. After extraction, each sample was centrifuged at 16,300 g for five minutes to separate the sediment and supernatant. The supernatant was removed with a 1 ml syringe and filtered through a 0.45μm Millipore Millex-HV PVDF filter into a screw cap high performance liquid chromatography (HPLC) vial. The samples were stored at -20°C until analysis.

Analysis was performed using an Agilent 1100 Series high performance liquid chromatograph (HPLC) with a Phenomenex SphereClone™ C18 LC column (5 µm particle size, ODS(2) phase, 80 Å pore size, 250 x 4.6 mm). The HPLC used a diode array detector with two lamps capable of producing wavelengths in both the UV and visible range. A flow rate of 1 ml/min was chosen. Parameters specific to each compound are summarized in Table 1.

V-19

Table 1: High performance liquid chromatography parameters for analysis of four pharmaceutical compounds.

Compound Eluent Injection Wavelength Length (% methanol/ Volume (nm) of water/acetic acid) (μl) Analysis (min) Acetaminophen 30/69/1 10 240 6

Aspirin 69/30/1 10 240 6

Diclofenac 69/30/1 30 230 12

Ibuprofen

Because of the different parameters, acetaminophen and aspirin were analyzed separately while diclofenac and ibuprofen were analyzed together. A water blank was analyzed first to ensure a flat baseline (required for accurate integration of the chromatograms), followed by standards of a known concentration (25 μM, 50 μM, 100

μM, 250 μM, 500 μM and 1000 μM) of the pharmaceutical being analyzed. Following the standards was another water blank, followed by the extracted microcosm samples.

V-20

Between samples from each of the five sites were water blanks to ensure that the baseline

remained constant.

The HPLC was connected to a desktop computer which was used to analyze and

store data. ChemStation for LC Rev. A 08.03 [847] software was used to calculate peak

area. Using the data generated by the standards, standard curves were created for each

analysis by plotting the known concentration of the standards versus the peak area of

each standard. Using this standard curve, the concentration of the pharmaceuticals in each

sample could be calculated. Once the unknown concentrations had been determined, the

values for the three site samples were averaged together and one standard deviation was

calculated. These values were then plotted versus time to monitor degradation.

Establishment of simulated combined sewer overflow (CSO) influent incubations

Cultures were established using the sediment and water from each of the sampling sites, as well as influent from a wastewater treatment plant in Central New Jersey to simulate a CSO event, thus introducing a complex suite of PPCP compounds to the microbial community. Influent was first added to each of the site water samples to make a

10% influent (v/v) mixture. Each sediment sample was then added to the corresponding site water/influent mixture to make a 10% sediment (v/v) slurry with a final volume of

1200 ml. To prepare anaerobic cultures the slurries were sparged with nitrogen gas for 30 minutes. The first (t0) samples were collected and then the cultures were sealed with a stopper and cap. After sealing the bottle, the headspace was removed and replaced with nitrogen gas using sterile anaerobic technique to ensure that no oxygen remained in the cultures. Then, 0.5 ml of a 1 g/L resazurin stock solution was added to serve as a redox

V-21

indicator. Cultures designated sterile controls were autoclaved for 30 minutes after

removing the gas in the headspace. Nitrogen gas was added back to the headspace using

sterile anaerobic technique after autoclaving. The established cultures were incubated at

room temperature in the dark.

Analysis of influent incubations

Samples were taken from the cultures on Days 0, 20, 40, and 80 of the

experiment. A 250 ml sample was removed from each culture under a constant stream of

nitrogen gas. Sediment was removed and the liquid portion was extracted by a solid

phase extraction technique adapted from Kwon and Rodríguez (2014). To prepare for

extraction, 200 ml of the liquid portion was vacuum-filtered using a Pall Life Sciences

Supor-200 47 mm 0.2 µm membrane filter to remove any remaining particulate. The

filtrate was then spiked with 40 ng 13C-sulfamethoxazole to serve as an internal standard.

In addition to the simulated CSO culture samples, 250 ml of site water was filtered and

spiked with 50 ng 13C-sulfamethoxazole to determine the concentration of PPCPs at each

of the sampling locations.

Oasis HLB cartridges (6 cc with 150 mg sorbent) and a Waters Extraction

Manifold were used to carry out the solid phase extraction protocol (Kwon and

Rodríguez 2014). The final methanol extract (10 ml) was evaporated with a gentle stream

of nitrogen gas to approximately 1 ml, then transferred to a chromatography vial where it

was evaporated until dry. Dried samples were stored at -20°C until transported to

collaborators at the Institute for Urban Environment, Chinese Academy of Sciences, in

Xiamen, China.

V-22

Prior to analysis, extracts were re-dissolved in 1 ml of methanol-water (1:1 v/v)

and filtered using a 0.2 μm Millipore syringe-driven filter. 110 pharmaceutical

compounds were selected for analysis, including 37 antibiotics, 8 anti-

inflammatory/antipyretic drugs, 11 antiparasitic drugs, 15 cardiovascular drugs, 21

central nervous system drugs, and 18 endocrine/family-planning drugs. The selected

pharmaceutical compounds in the filtered extract were analyzed using a Shimadzu liquid chromatograph (LC) coupled with an ABI 6500 triple quadruple tandem mass

spectrometry (MS/MS). Chromatographic separation was performed using a Kinetex C18

column (100 mm × 2.1 mm × 2.6 µm). A flow rate of 0.30 ml/min was chosen and the injection volume of the samples was 10 µl. Additional details about the mobile phase and gradient elution program, as well as how the selected compounds were identified and quantified can be found in the supplementary materials of Hong et al. (2020). Once the concentrations of the pharmaceutical compounds had been determined, the values for the three site samples were averaged together and one standard deviation was calculated. A bar chart was created to compare the concentrations of the detected pharmaceutical compounds on Days 0, 20, and 40 and across three sites (MAN, BUC, and REN).

Due to issues with an extraction reagent, samples from the STA and GER sites

could not be included in the analysis. Day 80 samples from all sites also had to be

excluded.

V-23

RESULTS

Acetaminophen

By Day 8 of the experiment, acetaminophen had been degraded in all microcosms established with sediment and site water from Buchanan (BUC), Staatsburg (STA),

Germantown (GER), and Rensselaer (REN). This trend was observed in microcosms with no electron acceptor amendment to the site water (Figure 4) and in microcosms amended

with excess sulfate (Figure 5) and excess nitrate (Figure 6). The extent of acetaminophen

degradation observed in microcosms established with sediment and site water from

Manhattan (MAN), however, varied. In microcosms with no amendment to the site water,

no significant degradation was observed in the MAN microcosms within the first 22 days

of the study (Figure 4). By Day 43, acetaminophen concentrations in these microcosms were still comparable to those observed on Day 22, but all showed ≥50% loss of

acetaminophen by Day 100.

In MAN microcosms spiked with excess sulfate, near complete degradation was observed in one microcosm by Day 8 and two microcosms by Day 14. Complete degradation was not observed in the remaining MAN microcosm within the first 22 days of the study (Figure 5). By Day 43, no significant degradation was observed in this microcosm, but all MAN microcosms spiked with excess sulfate showed complete degradation of acetaminophen by Day 100.

In microcosms spiked with excess nitrate, complete degradation was observed in all MAN microcosms by Day 14 (Figure 6). Though the extent of acetaminophen degradation varied, in all cases the degradation observed in the MAN microcosms lagged

V-24 behind the degradation observed in the microcosms from the other four sites. No significant loss of acetaminophen was observed in the sterile controls.

Figure 4. Mean concentration of acetaminophen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of acetaminophen was observed in the sterile control (n = 1).

V-25

Figure 5. Mean concentration of acetaminophen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. Microcosms were amended with excess sulfate (10mM Na2SO4). No loss of acetaminophen was observed in the sterile controls (n = 2).

Figure 6. Mean concentration of acetaminophen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. Microcosms were amended with excess nitrate (10mM KNO3). No loss of acetaminophen was observed in the sterile controls (n = 2).

V-26

Aspirin

By Day 8 of the experiment, aspirin had been degraded in all microcosms

established with sediment and site water from all five sites (Figure 7). This trend was

observed in microcosms amended with excess nitrate, excess sulfate, and in microcosms

with no amendment. No significant loss of aspirin was observed in the sterile controls.

Figure 7. Mean concentration of aspirin ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of aspirin was observed in the sterile control (n = 1).

Diclofenac

By Day 100, diclofenac was still present in most of the microcosms established in

site water with no electron acceptor amendment. The loss of this compound varied both

between sites and between samples from the same site, as evidenced by the large error

bars (Figure 8). By Day 43, ≥50% loss of diclofenac was observed in one microcosm from the BUC site and the STA site. By Day 100, ≥50% loss of diclofenac was observed in at least one microcosm from all but the MAN site. Complete or near complete loss of

V-27

diclofenac was observed in one microcosm from the STA and GER sites by Day 100. No

significant loss of diclofenac was observed in the sterile controls.

Figure 8. Mean concentration of diclofenac ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of diclofenac was observed in the sterile controls (n = 4).

Similar trends were observed in microcosms amended with excess sulfate (Figure

9) and excess nitrate (Figure 10). In microcosms amended with excess sulfate, by Day 43,

≥50% loss of diclofenac was observed in at least one microcosm from the BUC, STA, and GER sites. By Day 100, ≥50% loss of diclofenac was observed in at least two microcosms from all but the MAN site. Complete or near complete loss of diclofenac was observed in all three microcosms from the STA and GER sites and one microcosm from the REN site by Day 100. In microcosms amended with excess nitrate, by Day 43, ≥50% loss of diclofenac was observed in at least one microcosm from the STA and GER sites.

By Day 100, ≥50% loss of diclofenac was observed in at least two microcosms from all

V-28 but the MAN site. Complete or near complete loss of diclofenac was observed in all three microcosms from the STA site and one microcosm from the GER site by Day 100.

Figure 9. Mean concentration of diclofenac ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. Microcosms were amended with excess sulfate (10mM Na2SO4). No loss of diclofenac was observed in the sterile controls (n = 4).

V-29

Figure 10. Mean concentration of diclofenac ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. Microcosms were amended with excess nitrate (10mM KNO3). No loss of diclofenac was observed in the sterile controls (n = 4).

Accompanied by the loss of diclofenac, with a retention time of 8.5 minutes, was the emergence of a new peak in the HPLC chromatogram with a retention time of 9.0 minutes. The loss of diclofenac and rise of this new peak appear to be proportional

(Figure 11).

V-30

A.

DCF IBU

B.

DCF IBU

C.

DCF UNK

D. DCF UNK

Figure 11. HPLC chromatograms depicting the loss of diclofenac (DCF) and rise of unknown metabolite (UNK) in STA 2 microcosm (site water with no amendment) over the 100-day experiment (A. Day 1; B. Day 14; C. Day 43; D. Day 100). Aspirin (ASP) and ibuprofen (IBU) loss was also observed. Upward and downward pointing arrows represent increasing or decreasing concentration, respectively.

Ibuprofen

By Day 43 of the experiment, complete or near complete degradation of ibuprofen was observed in all microcosms established in site water with no electron acceptor amendment (Figure 12). By Day 22, ≥50% loss of ibuprofen was observed in one

V-31

microcosm from the MAN, STA, and GER sites and in all three microcosms from the

BUC and REN sites. By Day 29, ≥50% loss of ibuprofen was observed in two

microcosms from the MAN and GER sites and in all three microcosms from the BUC,

STA, and REN sites. No significant loss of ibuprofen was observed in the sterile controls.

Figure 12. Mean concentration of ibuprofen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. No additional electron acceptor was provided. No loss of ibuprofen was observed in the sterile controls (n = 4).

Similar trends were observed in microcosms amended with excess sulfate (Figure

13) and excess nitrate (Figure 14). In microcosms amended with excess sulfate, by Day

22, ≥50% loss of ibuprofen was observed in at least two microcosms from all but the

MAN site. ≥50% loss of ibuprofen was not observed in MAN microcosms until Day 100.

In microcosms amended with excess nitrate, by Day 22, ≥50% loss of ibuprofen was

observed in all microcosms except those from the MAN site. ≥50% loss of ibuprofen was

observed in one MAN microcosm by Day 43 and complete degradation was observed in

all microcosms by Day 100.

V-32

Figure 13. Mean concentration of ibuprofen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. Microcosms were amended with excess sulfate (10mM Na2SO4). No loss of ibuprofen was observed in the sterile controls (n = 4).

Figure 14. Mean concentration of ibuprofen ± 1 SD in microcosms established with sediment and site water from all five sampling sites. n = 3 for each sampling site. Microcosms were amended with excess nitrate (10mM KNO3). No loss of ibuprofen was observed in the sterile controls (n = 4). V-33

Pharmaceuticals in Hudson River site water

Of the 110 pharmaceuticals analyzed, only 13 were found to be above the detection limit in the extracts of the site water from the MAN, BUC, and REN sites.

These compounds, including three antibiotics, two anti-inflammatory/antipyretic drugs,

four cardiovascular drugs, and four central nervous system drugs, were present in all

samples. Except for the antibiotic trimethoprim, mean PPCP concentrations were highest

in extracts from the MAN site (Table 2).

Table 2: LC-MS/MS data showing the suite of PPCPs present in the site water from the MAN, BUC, and REN sites. Values represent mean concentration ± 1 SD. n = 3 for MAN, n = 4 for BUC and REN sites. Pharmaceutical Concentration Class Name Abbreviation (ng/L) MAN BUC REN Antibiotic Ofloxacin OFC 31 ± 6 12 ± 2 6 ± 2 Sulfamethoxazole SMX 46 ± 4 24 ± 8 9 ± 3 Trimethoprim TMP 23 ± 6 44 ± 14 12 ± 3 Anti- Diclofenac DCF 205 ± 13 39 ± 17 33 ± 10 inflammatory Salicylic acid SCA 451 ± 26 259 ± 29 116 ± 20 Cardiovascular Diltiazem DTZ 3.4 ± 1.0 1.4 ± 0.2 1.2 ± 0.6 hydrochloride Gemfibrozil GFZ 39 ± 8 11 ± 2 18 ± 8 Metoprolol MPL 60 ± 12 36 ± 16 47 ± 21 tartrate Valsartan VST 250 ± 43 104 ± 35 62 ± 40 Central Nervous Caffeine CAF 641 ± 133 207 ± 34 477 ± 112 System Carbamazepine CBZ 19 ± 11 16 ± 3 16 ± 8 Citalopram CLP 18 ± 1 3.9 ± 0.8 2.0 ± 0.4 hydrobromide Venlafaxine VFX 50 ± 15 43 ± 8 22 ± 9 hydrochloride

Simulated CSO influent incubations

Of the 110 pharmaceuticals analyzed, only 16 were found to be above the

detection limit in the extracts of the influent incubations. 14 compounds, including two

V-34

antibiotics, three anti-inflammatory/antipyretic drugs, four cardiovascular drugs, and five

central nervous system drugs, were present in all samples (Figure 15). The antibiotic

lincomycin was only detected in the two sterile controls and the antibiotic sulfamethazine

was only detected in two of the three MAN cultures (concentration >100 ng/L). Though

some changes in concentration were observed, by the end of the experiment diclofenac,

ibuprofen, salicylic acid, gemfibrozil, metoprolol tartrate, valsartan, carbamazepine, and

gabapentin all remained well above the detection limit. Sulfamethoxazole concentration

was below the detection limit (BDL) by Day 20 in MAN cultures and BDL by Day 40 in

the BUC and REN cultures. Some loss of sulfamethoxazole was observed in the sterile

controls, but the concentration remained well above the detection limit. Trimethoprim concentration was BDL by Day 20 in the MAN and REN cultures and BDL by Day 40 in the BUC cultures. Loss of trimethoprim was observed in the sterile controls, but the concentration remained above the detection limit. Diltiazem hydrochloride concentration was BDL by Day 20 in BUC cultures and <1 ng/L by Day 40 in MAN and REN cultures.

No loss was observed in the sterile controls. Caffeine concentration was greatly reduced by Day 40 in all cultures, while no loss was observed in the sterile controls. Citalopram hydrobromide and venlafaxine hydrochloride concentration was BDL by Day 20 in all cultures, including sterile controls.

V-35

Days

Figure 15. Mean concentration of PPCPs ± 1 SD in simulated combined sewer overflow (CSO) influent incubations established with sediment and site water from three sampling sites over a 40-day period. n = 3 for each sampling site/time point. n = 2 for sterile controls.

V-36

DISCUSSION

The central goal of this study was to determine how redox environment impacts the biodegradability of select PPCP compounds and the bacterial communities responsible for their degradation in anoxic sediments using the Hudson River watershed as a model. Long-term PPCP contamination in the environment was hypothesized to result in the enrichment of microbial communities capable of utilizing PPCPs as carbon and energy sources. It was also hypothesized that microorganisms from different points in the watershed and active under different redox conditions would have different metabolic potentials to degrade PPCPs.

Aspirin was readily degraded by the native microorganisms in Hudson River sediment and the addition of excess sulfate or nitrate as terminal electron acceptors did not affect its degradability. This is contrary to previous microcosm studies carried out using methanogenic, sulfidogenic, and denitrifying minimal salts media (Figures 1 and

2). In these studies, the terminal electron acceptor appeared to influence the rate of

aspirin degradation. The rapid microbial degradation of aspirin is not surprising because

salicylic acid, aspirin’s active metabolite which is structurally similar, is a phenolic

phytohormone found in plants and is thought to be ubiquitous in nature.

Acetaminophen was also readily degraded by the native microorganisms, except

for those in the MAN microcosms with no amendment to the site water and one MAN

microcosm with excess sulfate. This lag in degradation has been observed previously in

sulfidogenic microcosms established with sediment collected from the Arthur Kill and

previously collected from the Manhattan site (Figures 1 and 2). Though degradation was most delayed in MAN microcosms with no additional electron acceptor added, sulfate

V-37

was likely the dominant anaerobic terminal electron acceptor in these microcosms due to

the saline, sulfate-rich environment from which the samples were collected. For the other

four sites, this experiment showed no significant difference in acetaminophen

degradability between microcosms established using site water with no additional

electron acceptor added and those spiked with excess nitrate or sulfate. Therefore,

acetaminophen degradability appears to be site-specific rather than influenced by the

terminal electron acceptor.

Unlike aspirin and acetaminophen, diclofenac was not readily degraded by the native microorganisms; however, there was loss of diclofenac that was not observed in previous experiments carried out in minimal media (Figures 1 and 2). In these previous experiments, there was no biotic diclofenac loss over more than 50 days, whereas in this study, diclofenac loss was observed in some microcosms by Day 14. The loss of diclofenac in this experiment is likely attributed to transformation rather than complete degradation. As the area of the peak associated with diclofenac in the HPLC chromatogram decreased, a new peak with a later retention time emerged and the area of this peak increased with the loss of diclofenac (Figure 11). Liquid chromatography-mass spectrometry may be used to identify this metabolite. Like acetaminophen, degradability appears to be more dependent on site than terminal electron acceptor, with degradation in

MAN microcosms lagging behind the other four sites.

Like diclofenac, the biotic reduction of ibuprofen was not observed in previous experiments carried out in minimal media (Figures 1 and 2). In these previous experiments, there was no biotic ibuprofen loss over more than 50 days, whereas in this study, complete or near complete degradation of ibuprofen was observed in some

V-38 microcosms by Day 14 (Figure 14). Amendment with terminal electron acceptors did seem to affect degradability of ibuprofen in MAN microcosms, as microcosms with no amendment degraded ibuprofen faster than those spiked with excess sulfate or nitrate.

In conclusion, microbial communities capable of degrading or transforming all four pharmaceutical compounds of interest were found in sediments from the Hudson

River, supporting the hypothesis that long-term PPCP contamination results in the enrichment of microbial communities capable of utilizing PPCPs as carbon and energy sources; however, the hypothesis that redox environment impacts the biodegradability of

PPCPs was not fully supported for microorganisms cultivated in site water. The extent of anaerobic biodegradation or transformation observed in the microcosms was instead dependent on both the PPCP and location from which the microorganisms were cultivated. The MAN microcosms exhibited a unique degradation profile compared to the other four sites and further experiments must be carried out to determine if salinity or another site characteristic adversely influences the degradability of PPCPs at this site.

The results of the degradation assay experiments focusing on the four compounds of interest were not in agreement with those of the simulated CSO influent incubation experiments. Though loss of some of the pharmaceutical compounds was observed, most of the compounds were still present at concentrations well above the limit of detection even after 40 days (Figure 15). Four of the five compounds that did show loss BDL also showed loss within the sterile controls. The loss of these compounds may be due to sorption rather than microbial degradation and the extraction of PPCPs from the solid fraction of the samples must be analyzed to confirm this. Due to the physical properties of citalopram hydrobromide and venlafaxine hydrochloride, these compounds are

V-39

expected to partition to sediment (Kwon and Armbrust 2008; Golovko et al. 2020). The

adsorption and desorption of antibiotics in aquatic environments can be much more

complex and influenced by factors such as pH, salinity, amount of organic carbon and

clay in the sediment, and phosphate and nitrate concentration. Both trimethoprim (Li and

Zhang 2017) and sulfamethoxazole (Martínez-Hernández et al. 2014) have the ability to adsorb to and desorb from sediments and soils depending on the environmental conditions and can be found in both surface water and sediment samples collected from

the environment (see for example Hu et al. 2018; Kairigo et al. 2020). This may explain

why citalopram hydrobromide and venlafaxine hydrochloride concentrations were BDL,

whereas only some loss of sulfamethoxazole and trimethoprim was observed in the sterile

controls.

Perhaps the most noteworthy observation was that both salicylic acid (the active

metabolite of aspirin) and diclofenac remained in the influent incubations despite their

degradation/transformation in the microcosm experiments. This suggests that microcosm

studies focused on just one or even a small mixture of pharmaceutical compounds may

not accurately predict the fate of PPCPs in the environment where they would be present

as part of a complex mixture. One reason for this may have been an inhibitory effect from

one or more of the other compounds present in the influent mixture. Studies of other

aromatic organic compounds have demonstrated that compounds known to be degraded

individually may be degraded at a slower rate or completely inhibited when present in a

mixture (see for example Bielefeldt and Stensel 1999; Baggi and Zangrossi 1999;

Hennessee and Li 2016). Further experiments must be carried out to investigate which of

these compounds might be inhibitory.

V-40

Future directions

One of the ultimate goals of this project is to enrich for the microbial populations responsible for the degradation and transformation of PPCPs in order to characterize and identify the active members within the community. After enriching for the degrading community members, actively degrading microcosms will be selected for DNA extraction and bacterial community analysis. rRNA phylogenetic analysis will be performed to identify the microbial community members. The relative abundances of these organisms can then be analyzed to determine which organisms are dominant within that community. After the microorganisms that mediate PPCP degradation in anoxic environments have been identified, there is potential for developing molecular monitoring tools, as well as elucidating potential strategies for enhancing biodegradation.

V-41

ACKNOWLEDGEMENTS

I would like to sincerely thank Helena Andreyko, Sarah Fernald, and Dr. David Yozzo from the Hudson River Foundation for their patience and guidance throughout the duration of this project and the Tibor T. Polgar Fellowship for the financial support. I would like to thank my advisor, Dr. Max Häggblom, for his continued support, mentorship, and encouragement. I would also like to thank our collaborators Dr. Bing

Hong and Dr. Shen Yu at the Institute for Urban Environment, Chinese Academy of

Sciences. Additionally, I would like to express gratitude to the undergraduate students who assisted with this project: Rohan Guha (University of Chicago), Daniel Hazel

(Rutgers University), and Sarah Fichot (Rutgers University). Finally, I would like to acknowledge my father, Ronald Zeliph, for his excellent company and assistance while sampling.

V-42

REFERENCES

Almeida, B., H. Kjeldal, I. Lolas, A.D. Knudsen, G. Carvalho, K.L. Nielsen, M.T.B. Crespo, A. Stensballe, and J.L. Nielsen. 2012. Quantitative proteomic analysis of ibuprofen- degrading Patulibacter sp. strain I11. Biodegradation 24:615-630.

Baggi, G., and M. Zangrossi. 1999. Degradation of chlorobenzoates in soil suspensions by indigenous populations and a specialized organism: interactions between growth and non-growth substrates. FEMS Microbiology Ecology 29:311-318.

Beretta, M., V. Britto, T.M. Tavares, S.M. Teixeria da Silva, and A.L. Pletsch. 2014. Occurrence of pharmaceutical and personal care products (PPCPs) in marine sediments in the Todos os Santos Bay and the north coast of Salvador, Bahia, Brazil. Journal of Soils and Sediments 14:1278-1286.

Bessa, V.S., I.S. Moreira, M.E. Tiritan, and P.M.L Castro. 2017. Enrichment of bacterial strains for the biodegradation of diclofenac and carbamazepine from activated sludge. International Biodeterioration and Biodegradation 120:135-142.

Bielefeldt, A.R., and H.D. Stensel. 1999. Modeling competitive inhibition effects during biodegradation of BTEX mixtures. Water Research 33:707-714.

Brausch, J. M., K.A. Connors, B.W. Brooks, and G.M. Rand. 2012. Human pharmaceuticals in the aquatic environment: a review of recent toxicological studies and consideration for toxicity testing. Reviews of Environmental Contamination and Toxicology 218:1- 99.

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. 2018. Spatial patterns of pharmaceuticals and wastewater tracers in the Hudson River Estuary. Water Research 137:335-343.

Carpenter, C.M.G., and D.E. Helbling. 2018. Widespread micropollutant monitoring in the Hudson River Estuary reveals spatiotemporal micropollutant clusters and their sources. Environmental Science and Technology 52:6187-6196

Cydzik-Kwiatkowska, A., and M. Zielińska. 2016. Bacterial communities in full-scale wastewater treatment systems. World Journal of Microbiology and Biotechnology 32:66.

Daughton, C.G., and I.S. Ruhoy. 2009. Environmental footprint of pharmaceuticals: The significance of factors beyond direct excretion to sewers. Environmental Toxicology and Chemistry 28:2495-2521.

De Gusseme, B., L. Vanhaecke, W. Verstraete, and N. Boon. 2011. Degradation of acetaminophen by Delftia tsuruhatensis and Pseudomonas aeruginosa in a membrane bioreactor. Water Research 45:1829-1837.

V-43

Deo, R.P., and R.U. Halden. 2013. Pharmaceuticals in the built and natural water environment of the United States. Water 5:1346-1365.

Ebele, A.J., M.A. Abdallah, and S. Harrad. 2017. Pharmaceuticals and personal care products (PPCPs) in the freshwater aquatic environment. Emerging Contaminants 3:1-16.

Fennell, D.E., S. Du, H. Liu, F. Liu, and M.M. Häggblom. 2011. 6.13. Dehalogenation of polychlorinated dibenzo-p-dioxins and dibenzofurans, polychlorinated biphenyls and brominated flame retardants and potential as a bioremediation strategy. Pages 135- 149 in S. Agathos, and M. Moo-Young, editors. Comprehensive Biotechnology: Principles, Applications and Regulations in Industry, Agriculture, Medicine and the Environment, Second Edition, Volume 6. Pergamon, Oxford.

Fu, Q., D. Fedrizzi, V. Kosfeld, C. Schlechtriem, V. Ganz, S. Derrer, D. Rentsch, and J. Hollender. 2020. Biotransformation changes bioaccumulation and toxicity of diclofenac in aquatic organisms. Environmental Science and Technology 54:4400- 4408.

Golovko, O., A-L. Rehrl, S. Köhler, and L. Ahrens. 2020. Organic micropollutants in water and sediment from Lake Malaren, Sweden. Chemosphere 258:127293.

Gros, M., M. Petrović, A. Ginebreda, and D. Barceló. 2010. Removal of pharmaceuticals during wastewater treatment and environmental risk assessment using hazard indexes. Environment International 36:15-26.

Häggblom, M.M. 1992. Microbial breakdown of halogenated aromatic pesticides and related compounds. FEMS Microbiology Reviews 9:29-71.

Häggblom, M.M., Y.B. Ahn, D.E. Fennell, L.J. Kerkhof, and S.K. Rhee. 2003. Anaerobic dehalogenation of organohalide contaminants in the marine environment. Advances in Applied Microbiology 53:61-84.

Han, S., K. Choi, J. Kim, K. Ji, S. Kim, B. Aho, J. Yun, K. Choi, J.S. Khim, X. Zhang, and J.P. Giesy. 2010. Endocrine disruption and consequences of chronic exposure to ibuprofen in Japanese medaka Oryzias latipes and freshwater cladocerans Daphnia magna and Moina macrocopa. Aquatic Toxicology 98:256-264.

Hennessee, C.T., and Q.X. Li. 2016. Effects of Polycyclic Aromatic Hydrocarbon Mixtures on Degradation, Gene Expression, and Metabolite Production in Four Mycobacterium Species. Applied and Environmental Microbiology 82:3357-3369.

Hollender, J., S.G. Zimmermann, S. Koepke, M. Krauss, C.S. McArdell, C. Ort, H. Singer, U. von Gunten, and H. Siegrist. 2009. Elimination of organic micropollutants in a municipal wastewater treatment plant upgraded with a full-scale post-ozonation followed by sand filtration. Environmental Science and Technology 43:7862–7869.

V-44

Hong, B., S. Yu, Y. Niu, J. Ding, Q. Lin, X. Lin, and W. Hu. 2020. Spectrum and environmental risks of residual pharmaceuticals in stream water with emphasis on its relation to epidemic infectious disease and anthropogenic activity in watershed. Journal of Hazardous Materials 385:121594.

Hu, Y., X. Yan, Y. Shen, M. Di, and J. Wang. 2018. Antibiotics in surface water and sediments from Hanjiang River, Central China: Occurrence, behavior and risk assessment. Ecotoxicology and Environmental Safety 157:150-158.

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 and Technology 47:661-677.

Kairigo, P., E. Ngumba, L-R. Sundberg, A. Gachanja, and T. Tuhkanen. 2020. Contamination of surface water and river sediments by antibiotic and antiretroviral drug cocktails in low and middle-income countries: occurrence, risk, and mitigation strategies. Water 12:1376.

Kay, P., S.R. Hughes, J.R. Ault, A.E. Ashcroft, and L.E. Brown. 2017. Widespread, routine occurrence of pharmaceuticals in sewage effluent, combined sewer overflows and receiving waters. Environmental Pollution 220B:1447-1455.

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 and Technology 36:1202-1211.

Kwon, J., and J.M. Rodríguez. 2014. Occurrence and removal of selected pharmaceuticals and personal care products in three wastewater-treatment plants. Archives of Environmental Contamination and Toxicology 66:538-548.

Kwon, J., and K.L. Armbrust. 2008. Aqueous Solubility, n-Octanol–Water Partition Coefficient, and Sorption of Five Selective Serotonin Reuptake Inhibitors to Sediments and Soils. Bulletin of Environmental Contamination and Toxicology 81: 128-135.

Li, J., and H. Zhang. 2017. Factors influencing adsorption and desorption of trimethoprim on marine sediments: mechanisms and kinetics. Environmental Science and Pollution Research 24: 21929-21937.

Li, X., W. Zheng, and W.R. Kelly. 2013. Occurrence and removal of pharmaceutical and hormone contaminants in rural wastewater treatment lagoons. Science of the Total Environment 445-446:22-28.

Lv, M., Q. Sun, A. Hu, L. Hou, J. Li, X. Cai, and C-P. Yu. 2014. Pharmaceuticals and personal care products in a mesoscale subtropical watershed and their application as sewage markers. Journal of Hazardous Materials 280:696-705.

V-45

MacGillivray, A.R. 2012, revised 2013. Contaminants of emerging concern in the tidal Delaware River: pilot monitoring survey 2007-2009. Final Report to Delaware River Basin Commission. West Trenton, NJ.

Marchlewicz, A., U. Guzik, K. Hupert-Kocurek, A. Nowak, S. Wilczyńska, and D. Wojcieszyńska. 2017. Toxicity and biodegradation of ibuprofen by Bacillus thuringiensis B1(2015b). Environmental Science and Pollution Research 24:7572- 7584.

Marco-Urrea, E., M. Pérez-Trujillo, T. Vicent, and G. Caminal. 2009. Ability of white-rot fungi to remove selected pharmaceuticals and identification of degradation products of ibuprofen by Trametes versicolor. Chemosphere 74:765-772.

Martínez-Hernández, V., R. Meffe, S. Herrera, E. Arranz, and I. de Bustamante. 2014. Sorption/desorption of non-hydrophobic and ionisable pharmaceutical and personal care products from reclaimed water onto/from a natural sediment. Science of the Total Environment 472:273-281.

Miѐge, C., J.M. Choubert, L. Ribeiro, M. Eusѐbe, and M. Coquery. 2009. Fate of pharmaceuticals and personal care products in wastewater treatment plants – conception of a database and first results. Environmental Pollution 157:1721-1726.

Moschet, C., C. Götz, P. Longrée, J. Hollender, and H. Singer. 2013. Multi-level approach for the integrated assessment of polar organic micropollutants in an international lake catchment: the example of Lake Constance. Environmental Science and Technology 47:7028–7036.

Murdoch, R.W., and A.G. Hay. 2005. Formation of catechols via removal of acid side chains from ibuprofen and related aromatic acids. Applied and Environmental Microbiology 71:6121-6125.

Nagarnaik, P.M., A.L. Batt, and B. Boulanger. 2012. Healthcare facility effluents as point sources of select pharmaceuticals to municipal wastewater. Water Environment Research 84:339-345.

New York State Department of Environmental Conservation (NYSDEC). 2018. Combined Sewer Overflow (CSO). New York Department of Environmental Conservation. http://www.dec.ny.gov/chemical/48595.html (accessed February 21, 2018).

New York State Department of Environmental Conservation (NYSDEC). 2020. Hudson River Watershed Map. New York Department of Environmental Conservation. https://www.dec.ny.gov/docs/remediation_hudson_pdf/hrewatershed.pdf (accessed July 30, 2020).

Onesios, K.M., J.T. Yu, and E.J. Bouwer. 2009. Biodegradation and removal of pharmaceuticals and personal care products in treatment systems: a review. Biodegradation 20:441–466.

V-46

Onesios-Barry, K.M., D. Berry, J.B. Proescher, I.K.A. Sivakumar, and E.J. Bouwer. 2014. Removal of pharmaceuticals and personal care products during water recycling: microbial community structure and effects of substrate concentration. Applied and Environmental Microbiology 80:2440–2450.

Parales, R.E, R. Navara, R. Gettys, and J.J. Huang. 2017. Genome sequence of Pseudomonas putida strain ASAD, an acetylsalicylic acid-degrading bacterium. Genome Announcements 5:e01169-17

Phillips, P.J., S.G. Smith, D.W. Kolpin, S.D. Zaugg, H.T. Buxton, E.T. Furlong, K. Esposito, and B. Stinson. 2010. Pharmaceutical formulation facilities as sources of opioids and other pharmaceuticals to wastewater treatment plant effluents. Environmental Science and Technology 44:4910-4916.

Pomati, F., A.G. Netting, D. Calamari, and B.A. Neilan. 2004. Effects of erythromycin, tetracyclin and ibuprofen on the growth of Synechocystis sp. and Lemna minor. Aquatic Toxicology 67:387-396.

Pounds, N., S. Maclean, M. Webley, D. Pascoe, and T. Hutchinson. 2008. Acute and chronic effects of ibuprofen in the mollusc Planorbis carinatus (Gastropoda: Planorbidae). Ecotoxicology and Environmental Safety 70:47-52.

Santos, L. H., M. Gros, S. Rodriguez-Mozaz, C. Delerue-Matos, A. Pena, D. Barceló, and M.C. Montenegro. 2013. Contribution of hospital effluents to the load of pharmaceuticals in urban wastewaters: identification of ecologically relevant pharmaceuticals. Science of the Total Environment 461-462:302-316.

Sengupta, A., J.M. Lyons, D.J. Smith, J.E. Drewes, S.A. Snyder, A. Heil, and K.A. Maruya. 2014. The occurrence and fate of chemicals of emerging concern in coastal urban rivers receiving discharge of treated municipal wastewater effluent. Environmental Toxicology and Chemistry 33:350-358.

Song, H., and T.S. Chen. 2001. p-Aminophenol-induced liver toxicity: tentative evidence of a role for acetaminophen. Journal of Biochemical and Molecular Toxicology 15:34-40.

Subedi, B., N. Codru, D.M. Dziewulski, L.R. Wilson, J. Xue, S. Yun, E. Braun-Howland, C. Minihane, and K. Kannan. 2015. A pilot study on the assessment of trace organic contaminants including pharmaceuticals and personal care products from on-site wastewater treatment systems along Skaneateles Lake in New York State, USA. Water Research 72:28-39.

Sun, Q., M. Lv, A. Hu, X. Yang, and C.P. Yu. 2014. Seasonal variation in the occurrence and removal of pharmaceuticals and personal care products in a wastewater treatment plant in Xiamen, China. Journal of Hazardous Materials 277:69-75.

Sun, Q., M. Li, C. Ma, X. Chen, X. Xie, and C-P. Yu. 2016. Seasonal and spatial variations of PPCP occurrence, removal and mass loading in three wastewater treatment plants located in different urbanization areas in Xiamen, China. Environmental Pollution 208:371–381. V-47

Sun, W., V. Krumins, D. Fennell, L. Kerkhof, and M.M. Häggblom. 2016. Anaerobic degradation of aromatic compounds, Pages 5.1.3-1 – 5.1.3-14 in Yates M, C. Nakatsu, R. Miller, and S. Pillai, editors. Manual of Environmental Microbiology, Fourth Edition. ASM Press, Washington, DC.

Ternes, T.A. 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32: 3245-3260.

Thomaidi, V.S., A.S. Stasinakis, V.L. Borova, and N.S. Thomaidis. 2015. Is there a risk for the aquatic environment due to the existence of emerging organic contaminants in treated domestic wastewater? Greece as a case-study. Journal of Hazardous Materials 283:740–747.

Thomaidi, V.S., A.S. Stasinakis, V.L. Borova, and N.S. Thomaidis. 2016. Assessing the risk associated with the presence of emerging organic contaminants in sludge-amended soil: A country-level analysis. Science of the Total Environment 548-549:280–288.

Wang, J., and S. Wang. 2016. Removal of pharmaceuticals and personal care products (PPCPs) from wastewater: a review. Journal of Environmental Management 182:620–640.

Yu, Y., Q. Huang, Z. Wang, K. Zhang, C. Tang, J. Cui, J. Feng, and X. Peng. 2011. Occurrence and behavior of pharmaceuticals, steroid hormones, and endocrine- disrupting personal care products in wastewater and the recipient river water of the Pearl River Delta, South China. Journal of Environmental Monitoring 13:871-878.

Zeliph, M., A. Rupakula, B. Hong, S. Yu, and M.M. Häggblom. 2016. Microbial degradation and transformation of pharmaceuticals and personal care products. Poster presented at: International Society for Microbial Ecology (ISME). Montreal, Quebec, Canada.

Zhang, L., J. Hu, R. Zhu, Q. Zhou, and J. Chen. 2012. Degradation of paracetamol by pure bacterial cultures and their microbial consortium. Applied Microbiology and Biotechnology 97:3687–3698.

V-48 ASSESSING MODE OF REPRODUCTION IN VALLISNERIA AMERICANA OF THE HUDSON RIVER, NY, AND THE CHESAPEAKE BAY, MD

A Final Report of the Tibor T. Polgar Fellowship Program

Carrie E. Perkins

Polgar Fellow

Plant Sciences University of Maryland-College Park College Park, MD 20742

Project Advisor:

Dr. Maile C. Neel Plant Sciences University of Maryland-College Park College Park, MD 20742

Perkins, C.E. and M.C. Neel. 2020. Assessing mode of reproduction in Vallisneria americana of the Hudson River, NY and the Chesapeake Bay, MD. Section VI:1-59 pp. In S.H. Fernald, D.J. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

VI-1 ABSTRACT

Vallisneria americana, a dioecious macrophyte native to eastern North America, is

capable of both sexual and asexual reproduction. Mode of reproduction plays a role in

determining amounts and structure of genetic diversity at multiple scales, potential for

dispersal, and resilience to disturbances. Observations from 2015 indicated that sexual

reproduction in the Hudson River was lower than in the Chesapeake Bay. Mode of reproduction was assessed at 14 sites on the Hudson River representing tidal-saline, tidal- fresh, and non-tidal environments. Three tidal Chesapeake Bay sites were sampled for comparison. Owing to previous findings that high salinity and low temperature inhibit flowering in V. americana, it was hypothesized that flowering would be positively correlated with large plant size, high water temperature, low salinity, high genotypic diversity, and high bed density. In July, male and female inflorescences were counted using shoot-level transect sampling and the size of each plant was recorded. In August, female flowers were counted and the presence of male flowers was noted using surface transect sampling. Chesapeake Bay sites had an average of 312.3 inflorescences in shoot-level sampling and 1137.3 female inflorescences in surface sampling, whereas the averages for the Hudson River were 15.64 and 431.5, respectively. The highest- producing Hudson River sites had statistically similar numbers of inflorescences to

Chesapeake Bay sites. Plants with long leaves and many ramets were more likely to flower. Chesapeake Bay plants had longer leaves and Hudson River plants had more ramets. Spatial isolation of sexes was evident at both the estuary scale and transect scale.

The results of this study give managers a new way to evaluate the resilience of V. americana beds throughout the Hudson River, informing restoration decisions.

VI-2

TABLE OF CONTENTS

Abstract ...... VI-2

Table of Contents ...... VI-3

List of Figures and Tables...... VI-4

Introduction ...... VI-5

Methods...... VI-10

Results ...... VI-24

Discussion ...... VI-41

Acknowledgments...... VI-53

References ...... VI-54

VI-3

LISTS OF FIGURES AND TABLES

Figure 1 – Map of Hudson River sampling sites ...... VI-11

Figure 2 – Map of Chesapeake Bay sampling sites ...... VI-12

Figure 3 – Distribution of the number of inflorescences per sample at each site VI-25

Figure 4 – Total number of male, female, and non-flowering shoots per site ..... VI-26

Figure 5 – Density of female inflorescences (per 1000m2) at each site ...... VI-28

Figure 6 – Maximum leaf length, leaf width, and number of ramets distribution of flowering vs. non-flowering individuals ...... VI-30

Figure 7 – Probability of flowering at varying max leaf lengths and widths ...... VI-32

Figure 8 – Probability of flowering at varying max leaf lengths and # ramets ... VI-32

Figure 9 – Maximum daily salinity at Hudson River monitoring stations, averaged across years from 2008 to 2018 ...... VI-34

Figure 10 – Mean of the minimum daily temperature at Hudson River and Chesapeake Bay monitoring stations, averaged across years from 1984 to 2018 ...... VI-34

Figure 11 – Coefficient of Variation of numbers of shoots among buckets ...... VI-36

Figure 12 – Site-level deviation from 50:50 sex ratio ...... VI-39

Figure 13 – Number of segments containing both male and female inflorescences ...... VI-40

Table 1 – Sampling dates, number of samples, transect lengths, genotypic diversity, and inflorescences and flowers seen at each site ...... VI-17

Table 2 – Number of multilocus genotypes from each site that flowered in the greenhouse at the Appalachian Laboratory...... VI-37

VI-4

INTRODUCTION

Like many native aquatic plants worldwide, Vallisneria americana Michx. in the

Hudson River Estuary is threatened by chronic poor water quality and increasingly extreme and unpredictable weather events. Because this species grows submersed in the water column, it is especially vulnerable to conditions that limit light, such as excessive suspended sediment and high nutrient levels in the water (Orth and Moore 1983: Davis

1985). This dependence is strong enough that often submersed species are biological sentinels, alerting us to declines in water quality (Orth et al. 2017). Chronic light limitation (e.g., Davis 1985) and rare but extreme weather events (Bayley et al. 1978) have caused massive, decades-long declines in V. americana across eastern North

America (Brush and Hilgartner 2000; Findlay et al. 2014). Most recently, intense flooding, scouring, and sediment deposition from Hurricane Irene and Tropical Storm

Lee in 2011 caused major losses throughout the Hudson River (Hamberg et al. 2017).

Before those storms, V. americana was the dominant submersed aquatic species in the

Hudson River (Nieder et al. 2004, 2009), typically forming single-species or low

diversity beds through a combination of vegetative and sexual reproduction.

Both long-term declines and catastrophic losses are of great concern to managers

because submersed aquatic plants such as V. americana are foundation species and

ecological engineers (Orth et al. 2006). They play critical roles in providing habitat for fish and crustaceans (Heck and Thoman 1984) and sustenance for waterfowl (Lubbers et al. 1990), oxygenating water (Findlay et al. 2006), stabilizing sediments, and removing excess nutrients from the water (Sand‐jensen 1998). As such, potential for resilience in

V. americana is of great interest to managers and scientists.

VI-5

Populations are said to be resilient in the face of disturbance if they are able to

absorb change while maintaining the same function, regain essential functions after a

loss, or acclimate in response to altered environmental conditions (Holling 1973; Sgrò et

al. 2011). Resilience can take many forms, and could include habitat patches persisting

year-to-year, patches shifting while maintaining the same core location, patches continuing to provide a constant amount of ecosystem services, and patches being lost but regained through colonization. Each of these forms of resilience fundamentally requires that plants reproduce and that propagules reach suitable habitat in which they can grow

and ultimately reproduce themselves. The ability to grow requires environmental

conditions within the range of tolerance for the species. Reproduction and dispersal

strongly affect the amounts and distribution of genetic diversity which increases the

potential for tolerance of and acclimation to a broader range of conditions, and for

adaptation as conditions change (e.g., Procaccini and Piazzi 2001; Hughes and

Stachowicz 2004; Sgrò et al. 2011). For clonal species such as V. Americana, relative amounts of asexual versus sexual reproduction are particularly important for determining the structure of genetic diversity (Eckert and Barrett 1993). Thus, without any reproduction even short-term resilience is not possible, and longer-term resilience depends on the mode of reproduction (asexual versus sexual) and dispersal.

Dramatic recovery since the 2011 storms provides good evidence for resilience in

V. americana; however, the species is still absent from some locations that supported large beds, indicating recovery is not complete. Of more concern for the future is that genetic sampling in 2015 indicates that most sites have relatively low genotypic diversity

(i.e., they consist of many copies of one or a few genetic individuals) compared to most

VI-6

of the Chesapeake Bay and tidal Potomac River. Further, more genotypes are found to

occur at multiple sites. These patterns suggest that recovery was accomplished largely

through asexual reproduction and dispersal of vegetative propagules rather than via

sexual reproduction.

Because V. americana is dioecious (each genotype is either male or female),

spatial isolation of the sexes can hinder sexual reproduction. For pollen to reach the

female flowers that float on the water’s surface, male flowers must rise to the surface

after breaking free from the inflorescence that grows from the base of shoots. They contact the female either as complete flowers or as floating pollen. Pollen movement is extremely localized, typically to 2 to 5 m (Lloyd et al. 2018). In the extreme, spatial

isolation can arise from chance Allee effects if only one sex reaches a site during founder

events or if one sex is lost due to population bottlenecks (Eckert and Barrett 1993). If

isolated beds consist of only one sex, sexual reproduction is impossible. Even when both

sexes are present, low genotypic diversity can limit reproduction through sex bias

(Engelhardt et al. 2014) and extensive clonal growth that isolates males and females.

This reduced sexual reproduction may limit resilience to future change or disturbance if

low genotypic and genetic diversity yields low potential to acclimate or adapt.

Even more fundamental than the spatial distribution of male and female

inflorescences, is flowering itself. Preliminary observations during field sampling for

genetic analysis in 2015 indicated that many V. americana beds in the Hudson River

produced fewer inflorescences than those in the Chesapeake Bay and Potomac River.

Female flowers were observed at only a few sites. Few male inflorescences (composed

of hundreds of highly reduced flowers) were noted on collected plants, and almost no

VI-7

male flowers were seen on the water surface in the field. Not surprisingly, given the lack

of male flowers, very few mature fruits were seen even where female flowers were

present.

Although intriguing, these observations were anecdotal and better understanding

of the capacity for sexual reproduction in the Hudson River was needed. In summer

2018, two field studies were conducted to quantify inflorescence production and potential

for pollination in Hudson River plants and compare them to the Chesapeake Bay. In

July, fine-scale sampling was conducted to count female and male inflorescences on

collected plants. Although extremely informative, this detailed sampling represented

only a small area of each sampled bed. To estimate flowering and reproduction over a

larger area at each site, this detailed sampling was conducted in August when flowers and

fruits visible at the surface of the water on longer transects were counted.

To test hypotheses for possible mechanisms for limited flowering, the field sampling was augmented with a mesocosm study and with genetic and greenhouse data collected previously in a related project. Five hypotheses for low levels of flowering were tested: 1) Plants that had colonized after the 2011 storms had not reached reproductive size in 2015, given known effects of plant size on flowering (Titus and

Hoover 1991; Engelhardt et al. 2014); 2) Salinity or temperature were not conducive to

flowering. Flowering is negatively affected when salinity is above 5 to 10 ppt (French

and Moore 2003). Although studies have suggested that water temperature is important

for flowering (Titus and Stephens 1983; Titus and Hoover 1991; Best and Boyd 2001),

there are no data on specific temperature requirements; 3) Genotypic diversity was too

low to yield flowering (Engelhardt et al. 2014); 4) Density is too low or beds are too

VI-8

patchy to support flowering for either of two possible reasons. First, positive feedbacks in dense beds result in heightened local water clarity (Gurbisz and Kemp 2014). Second, higher density beds may reflect more resource rich environments that support both more vegetative growth and more sexual reproduction; 5) Plants in the Hudson River have

lower capacity to flower due to an intrinsic shift towards asexual reproduction.

Beyond flowering, potential for pollination was assessed by quantifying proximity

of male and female individuals at and within sites, and whether these proximities differed

in relation to salinity and genotypic diversity was tested. The results of this assessment of

reproduction give managers an indication of both short- and long-term resilience of sites

spanning the salinity gradient of the Hudson River, and highlight the specific risks.

Quantifying the number of inflorescences and their sex ratios makes it possible to assess

the potential for sexual reproduction and for future genotypic diversity among sites. Sites

that produce many flowers of both sexes have the potential to generate new genotypes

resulting in higher genotypic diversity and the ability to acclimate and adapt to changing

environmental conditions (e.g., Procaccini and Piazzi 2001; Hughes and Stachowicz

2004; Sgrò et al. 2011). By contrast, sites that persist through vegetative reproduction

and primarily amplify existing genotypes by cloning may have limited capacity under

changing conditions (Eckert et al. 2016). Determining if, and specifically, how flowering

or pollination is limited provides direction to managers who may seek to increase

reproduction to increase diversity.

VI-9

METHODS

Sampling sites were chosen to represent the non-tidal (n=2), tidal-fresh (n=7), and

tidal-saline (n=5) environments of the Hudson River (Figure 1). Three tidal Chesapeake

Bay sites were selected to provide comparison (Figure 2). All sites were sampled twice

in 2018. During small-scale intensive sampling in July, flowering on individual shoots

was assessed and flowering was associated with characteristics of those shoots, as well as

of each site. These detailed samples represented only a small part of each site. In August,

sampling took place on longer transects to get a more comprehensive estimate of

flowering at each site, but flowering could not be related to characteristics of individuals.

All sites had previously been sampled for genetic analysis in 2015 and 2016 at the

University of Maryland College Park. Resulting unpublished data were used to estimate

genotypic diversity at each site. From those same studies, plants from 30 samples from

each of 11 of the Hudson River sites and 15 samples from each of two Chesapeake Bay

sites had been propagated at the University of Maryland Center for Environmental

Science Appalachian Laboratory. Variation in flowering was quantified among genotypes from a subset of sampled sites in a mesocosm experiment planted with turions from these propagated individuals. Further, data on flowering from the greenhouse samples were summarized to provide an additional assessment of capacity for flowering of Hudson River genotypes.

Shoot-Level Sampling

In July, flowering was estimated on all V. americana shoots collected from 20 to

25, ~0.04 m² samples taken every ~4 m along a transect line at each site (Table 1). The

path of each transect and location of each sample were recorded using a handheld Garmin

VI-10

Etrex 30 GPS unit. If a sample frame had no plants, the absence was noted and sampling continued until at least 20 samples with plants were obtained.

Figure 1. Map of Hudson River sampling sites showing tidal regime (▲= non- tidal, ♦= tidal-fresh, and ● = tidal-saline) for each site and whether the site was included (gray) in the mesocosm experiment or not (black). HRECOS monitoring station sites (×) are also noted.

VI-11

Figure 2. Map of Chesapeake Bay sampling sites showing region (●=Central Bay and ▲=Potomac River). Water quality monitoring station sites (×) are also noted.

All single ramets and ramets connected by stolons in each sample were counted, as were stolons that were connected to undeveloped ramets or that had been connected to shoots but were broken. To avoid confusing them with undeveloped ramets, the term shoots is used to refer to ramets on which leaves were expanded and green such that they

could photosynthesize and potentially support inflorescences. Each separate shoot or set of connected shoots is referred to as an individual; however, many of them could be the same genetic individual.

VI-12

All female flowers and male inflorescences were recorded on each ramet

(hereafter collectively called inflorescences for convenience). Sexual reproductive status

of each shoot was noted as male, female, or non-flowering based on presence or absence of inflorescences. Individuals were noted as male or female if any shoot possessed

inflorescences of the respective sex, and were coded as non-flowering if there were no

inflorescences. The number of inflorescences on each shoot was counted. The number of

shoots that were female, male, and non-flowering, and the number of inflorescences were

summarized at the individual, sample, and site levels.

Plant size was measured in three ways. First, the number of connected shoots and

the number of stolons that were broken or attached to undeveloped shoots (hereafter

referred to collectively as total number of ramets) were summed per individual. Second,

the length of longest leaf of each shoot was measured to the nearest centimeter. Third,

for each single shoot or connected set of shoots, the width of the widest leaf was

measured to the nearest millimeter. The maximum shoot leaf length, leaf width per

individual, and number of ramets per individual were used to analyze the effect of plant

size on flowering.

Although it was not feasible to collect detailed environmental data at each site,

sampling from three broad tidal and salinity environments in the Hudson River made it

possible to categorically assess the effects of salinity on flowering. Salinity data from

three Hudson River Environmental Conditions Observing System (HRECOS) monitoring

stations - Norrie Point, NY (lower to mid tidal-fresh environment), West Point (near the

upper end of the tidal-saline), and Piermont Pier, NY (lower tidal-saline environment)

were used to describe different sampling sites within the River (HRECOS 2018).

VI-13

Temperature data from the aforementioned Hudson River stations as well as the station at

Tivoli, NY, and from 16 monitoring stations in the Chesapeake Bay were used to

compare minimum daily water temperature in the early growing season between the

regions (Data Download | Chesapeake Bay Program 2018).

Overall bed density was estimated using the mean and coefficient of variation

(CV) of the number of shoots per sample at each site, including samples that had no

plants. Density within patches was also calculated by excluding empty buckets for the

calculations. Patchiness was estimated as the ratio of samples with plants to the total

number of samples needed to obtain ≥20 samples with plants. Dense, continuous beds

will have a high average number of shoots with a low CV, and a ratio of occupied to total

samples near 1.

Surface Sampling

In August, all female flowers at the water surface were counted and presence of

floating male flowers was noted along two transect lines that were mapped using Garmin

GPS units. Transects were a minimum of 500 m long and observed from kayaks, and the

sampling area was limited to 0.9 m on either side of the boats. The maximum length

depended on the extent of plants and amount of flowering at each site (Table 1). More

search effort was expended when no or few flowers were found. The size was

standardized for the statistical analysis described below to density of female flowers

found per 1000 m2 to allow comparison across sites. Waypoints were used to identify the start and end of V. americana patches along the transect and to identify line segments with which to quantify spatial distribution of flowers along the transects. The method of

marking line segments depended on the number of flowers at a site. When many female

VI-14 flowers were found, waypoints were recorded after counting every ~100 inflorescences.

In addition to quantifying spatial variation in presence and density, intermediate increments also helped with accuracy of counts. If single flowers were found, their precise location was marked.

Presence of V. americana and abundance of flowers associated with the marked waypoints were aligned to tracks using a full outer join of the two datasets using the base

R, version 3.5.1 (R Core Team 2018) merge function with latitude and longitude as the common joining columns. This process yielded a list of georeferenced points, appropriately coded with the features identified at specific waypoints (e.g., beginning and end of the transect, beginning and end of patches of Vallisneria, and beginning and end of segments with a specified density of inflorescences). Lists of coordinates bounded by transect, patch, and flowering segment start and end waypoints were transformed into spatial line segments in a SpatialLinesDataFrame using the sp package, version 1.3-1

(Pebesma and Bivand 2018). Finally, the length of each line segment was determined using lineLength in the SDraw Package, version 2.1.3 (McDonald 2016). From these coded line segments, the total length of each transect line, the length of the transect occupied by V. americana, and the length of the transect in which female and male flowers were found were calculated. The number of gaps between patches was also calculated.

Bed density was estimated by calculating the proportion of total distance of each transect containing plants, and patchiness was estimated by summing the number of gaps between beds at each site.

VI-15

Mesocosm Study

Propagules of 16 genotypes from 6 of the field sampling sites were planted in

May 2018 in mesocosms at the Appalachian Lab in Frostburg, MD, to determine whether flowering differs among genotypes when plants are grown in ambient light conditions.

Sites were selected based on turion availability, but at least one site from each of the three Hudson River environments was represented. A female genotype (CRV7A3) from a V. americana bed in Croton River, just upstream from the CRO sampling site, was also planted. In total, 16 genotypes (11 females and 5 males) were planted, with 8 replicate pots per genotype and 4 turions per pot. Longest leaf, number of shoots, and number and sex of inflorescences in each bucket were recorded monthly from June to September.

When plants bloomed, the longest leaf of each flowering shoot was recorded.

Flowering in Greenhouse Grown Plants

To assess more broadly whether Hudson River genotypes have reduced capacity to flower, flowering in genotypes that have been propagated at the Appalachian

Laboratory since their collection from 11 Hudson River sites in 2015 and two

Chesapeake Bay sites in 2016 was quantified. The number of male genotypes and number of female genotypes from each site were calculated and used to compute sex ratios within sites.

VI-16

Table 1. Sampling dates, number of samples, transect lengths, genotypic diversity, and inflorescences and flowers seen at each site.

# of July August Samples Total patch with Shoot- distance/ plants/ Level total Total # of Transect distance Date Samples Length # # Genotypic Date searched August total Site Code Sampled Taken (m) Shoots Inflorescences 1 Diversity Sampled (m) female flowers Hudson River Stillwater STW 07/13 22/24 48.6 110 10 0.24 08/11 1090/1141 786

Mechanicville MEC 07/13 22/67 199.1 175 22 0.42 08/11 657/1760 128 Parda Hook PHK 07/12 22/35 79.5 166 7 0.55 08/11 1246/3554 121

Stockport STK 07/19 21/26 59.3 117 9 0.12 08/09 1080/1080 0

Brandow Point BPT 07/18 22/22 117.8 153 49 0.35 08/13 761/898 2661

Roger’s Island RIS 07/18 22/23 59.6 231 39 0.43 08/13 1491/3567 554 Cheviot CHV 07/17 & 22/22 67.4 0.27 08/09 2471/2480 112 07/19 303 25 Tivoli North TNB 07/16 23/30 129.0 0.53 08/10 1146/2516 1411 Bay 148 17 Magdalen MIS 07/16 23/23 47.8 0.24 08/10 752/820 264 Island 155 13 Annsville ACR 07/14 20/25 65.8 65 1 0.79 08/10 107/5661 0

VI-17

Creek Turning Point TPT 07/15 22/31 133.2 79 12 0.29 08/15 1363/1372 83 George’s GIS 07/10 21/21 66.9 0.18 08/09 1771/2084 4 Island 119 8 Croton CRO 07/20 22/42 115.5 114 3 0.63 08/08 1468/1650 3

Nyack NYK 07/11 25/26 132.6 240 7 0.10 08/08 1240/1274 6 Chesapeake Bay Wilson Point 08/07 07/28 Park WPP 23/23 65.0 219 337 0.59 645/696 2581 Brown’s Creek BC 07/28 23/23 34.5 213 276 0.62 07/28 217/546 817 Aquia Landing AL 08/06 23/25 94.6 250 324 1.00 08/06 683/699 14 1Genotypic diversity is calculated based on the number of genotypes found at a site (G) relative to the number of samples taken (N) in 2015 in the Hudson River and 2016 in the Chesapeake Bay using the formula: (G-1)/(N-1).

VI-18

Analysis - Flowering

In shoot-level sampling, the number of inflorescences was summarized at three

main levels: site, sample, and individual. The number of inflorescences was also

summarized at the region level, but unequal sample sites in the Chesapeake Bay and

Hudson River require that comparisons be interpreted with caution. Number of

inflorescences per site estimates the total amount of flowering, whereas the number per

sample standardizes across sites with different sample numbers. Summarizing at the

individual level controls for variation in number of individuals and shoots in samples.

In addition to numbers of inflorescences, the number of flowering shoots and number of

flowering individuals at each site and in each sample were calculated. Inflorescences

were measured if they were at least 2 cm.

Significance of region-level differences in number of inflorescences, number of

flowering shoots, and number of flowering individuals per site and per sample was

assessed using nested ANOVAs, with Site as a random effect nested within Region as a

fixed effect, as implemented by the stats package in R, version 3.5.1 (R Core Team

2018).

In surface sampling, the total number of female flowers and density per 1000 m2

at each site were computed. The site-level densities were used in an ANOVA with Site

as a random effect nested within Region as a fixed effect to determine whether the

Chesapeake Bay sites had higher density of female flowers than the Hudson River. The

distance (m) of patch-only transect containing female flowers and distance not containing

female flowers were computed for each site, and the ratio of patch-only transect

containing flowers to not containing flowers was calculated. The number and length of

VI-19

transect segments containing male flowers were computed to estimate the presence of

males.

The number of flowers per genotype in the replicated mesoscosm experiment was

noted. Presence and sex of flowers were noted for the field collected samples of known

genotype that had been growing in the greenhouse. Reproductive status was summarized

to the genotype level as male, female, or non-flowering.

Testing Hypotheses for Levels of Flowering

Plant Size

Distributions of maximum leaf length, leaf width, and number of ramets for flowering versus non-flowering individuals from the July shoot-level sampling were plotted. Differences were tested for in these three size variables between the Hudson

River and Chesapeake Bay using separate nested ANOVAs, with Sample as a random

effect nested within Site, and Site as a random effect nested within Region as a fixed effect. A multivariate logistic regression with backwards stepwise variable entry was used to quantify the effect of maximum leaf length, leaf width, and total number of ramets per individual on the probability that the individual was flowering. After examining regression predictions of flowering at varying leaf lengths, leaf widths, and

numbers of ramets, the size below which each individual has a 95% chance of not

flowering was determined. Individuals with leaves that had been broken off by the

sampling frame were excluded from these analyses.

VI-20

Environmental Conditions

Salinity

The maximum daily salinity (ppt) at three HRECOS monitoring stations,

Piermont Pier, West Point, and Norrie Point, was calculated for comparing salinity levels along the gradient of the river. These maximum daily values were averaged across 2008 to 2018 for each station and plotted by day of year. The tidal-saline sites were located between the Piermont Pier and West Point stations so these stations bracket salinities at the sampled sites. Tidal-fresh sites were all upstream from Norrie Point. Differences in ratio of flowering to total individuals (from the July shoot-level sampling) and site-level density of female flowers (from the August surface sampling) among tidal salinity categories were formally tested for using separate one-way ANOVAs.

Temperature

The minimum daily temperatures (°C) at Hudson River and Chesapeake Bay monitoring stations were averaged across all stations within each region and across all years from 1984 to 2018 in the Bay, and from 2008 to 2018 in the Hudson River. The differences in the minimum values on each day were compared, and then the first day of the year each minimum temperature (based on whole °C) was observed in each region was determined.

Genotypic Diversity

To understand if sites with more genotypes flowered more, a logistic regression was used to determine the degree to which site-level genotypic diversity predicted the ratio of flowering to non-flowering individuals per site measured in the July shoot-level

VI-21 sampling. This analysis included only Hudson River sites to prevent confounding genotypic and environmental differences of the Chesapeake Bay samples.

Bed Density

The effect of shoot density and degree of patchiness at sites was estimated using logistic regression to predict the ratio of flowering to non-flowering individuals per site as a function of the mean and coefficient of variation (CV) of the number of shoots per sample and the proportion of samples with plants.

In the surface sampling, density was estimated as the proportion of total distance of each transect containing plants and as number of gaps between patches. The proportion of patch-only transect containing flowers was also computed. To analyze effects of bed density in surface sampling, a logistic regression was used to test the effect of the number of gaps between patches and proportion of total distance of each transect containing plants on the proportion of patch-only transect containing flowers to patch- only transect lacking flowers.

Overall Capacity for Flowering

The percentages of greenhouse-grown Hudson River genotypes that flowered were compared in a nested ANOVA, with Site as a random effect nested within Region as a fixed effect.

Analysis – Pollination Potential

Potential limitations to pollination were assessed by examining isolation of sexes among and within sites. First, in both July shoot-level and August surface sampling, presence versus absence of each sex at each site was noted.

VI-22

Next, in the shoot level sampling, when both sexes were present at sites the sex ratio of reproductive shoots was calculated, where an equal sex ratio (0.5) is maximal and

increasing dominance by either sex goes to 0. A chi square test was also used to compare

sites in terms of the number of individuals that were assigned to each reproductive

category (male, female, non-flowering).

The spatial distribution of males and females within sites was then quantified. In

the shoot-level samples the minimum Euclidean distance between samples containing

each of the sexes was calculated. If male and female shoots were found together in a

sample, the distance was set to 0 m and the number of times this occurred in each site

was noted.

On surface sampling transect lines spatial proximity of the sexes was assessed by

counting the total number of segments containing both male and female inflorescences.

The small number of male flowers precluded formal statistical tests.

Whether potential for pollination at Hudson River sites was predicted by genotypic

diversity or salinity was also tested. Chesapeake Bay samples were excluded from this

analysis to prevent confounding genotypic and environmental differences that exist

between Maryland and New York. The effect of site-level genotypic diversity on the

site-level sex ratio above was assessed using a linear regression. It was then asked

whether the magnitude of deviation from an equal sex ratio differed among the tidal-

saline, tidal-fresh, and non-tidal sites in a one-way ANOVA.

VI-23

RESULTS

Flowering

The July shoot-level sampling yielded 219 inflorescences on 2,175 shoots representing 1,515 individuals in the Hudson River and 937 inflorescences on 682 shoots

representing 587 individuals in the Chesapeake Bay. This stark difference in flowering in

the two regions carried through to averages of 15.64 (sd=13.78) versus 312.33 (sd=32.13)

inflorescences per site and of 0.71 (sd=1.28) versus 13.58 (sd=9.00) inflorescences per

sample (Figure 3) in the Hudson River and Chesapeake Bay, respectively. Differences in

number of inflorescences per site between regions (F=719.8, df=1, p=4.31e-14), per

sample between regions (F=695.3, df=1, p=5.57e-14), and per individual between regions

(F= 554.7, df=1, p= 2.92e-13) were significant based on nested ANOVAs. Even the sites with the most inflorescences per sample in the Hudson River (BPT and RIS) had far fewer inflorescences than samples at Chesapeake Bay sites (Figure 3).

Differences in number of inflorescences per site between the regions were

partially accounted for by larger numbers of shoots at Chesapeake Bay sites ( =227.3, sd=19.8) than at most Hudson River sites ( =155.4, sd=65.5; F= 3.393, df=1,𝑥𝑥 p=̅ 0.0853;

Figure 4). But more importantly, a much larger𝑥𝑥̅ number of shoots were flowering at each

Chesapeake Bay site ( =120.0 versus =12.4 in the Hudson River; F=285.7, df=1,

p=3.56e-11; Figure 4) 𝑥𝑥and̅ per sample 𝑥𝑥at̅ each site (F=289.9, df=1, p=3.21e-11). In the

Chesapeake Bay >45% of shoots at each site were flowering ( =53.1%) whereas no

Hudson River site had more than 23% of shoots flowering ( =7𝑥𝑥̅ .7%).

All inflorescences at all Hudson River sites were <2 𝑥𝑥c̅ m long at the time of

sampling. In the Chesapeake Bay, 2.2% (BC) -21% (WPP) of flowers sampled were ≥2

VI-24 cm. At AL, the longest male inflorescence was 13 cm (n=26) and the longest female was

9 cm (n=2). At BC the longest female inflorescence was 25 cm (n=22) and the longest male was 3 cm (n=5). At WPP the longest female inflorescence was 94 cm (n=9) and the longest male was 11 cm (n=37). Although they were not included in samples, many mature female flowers were observed on the water surface at BC and AL during sampling. In the Hudson River, only one mature female flower was observed outside of the sampling area at BPT on the day the shoot-level samples were collected.

Figure 3. Distribution of the number of inflorescences per 0.04 m2 sample at each site. Lines indicate median number of inflorescences and open diamonds are mean numbers. Regions (MD=light gray; NY=dark gray) were significantly different based on a nested ANOVA.

VI-25

Figure 4. Total number of male (black), female (medium gray), and non- flowering shoots (light gray) per site.

In the August surface sampling, the number of female flowers detected at sites ranged from 0 (STK and ACR) to 2,661 (BPT) and averaged 561. Despite extensive survey ≤6 female flowers were found at AL, NYK, CRO, and GIS (Table 1). Although the number of flowers found in some Hudson River sites exceeded what was found in the

Chesapeake Bay, the search effort required to find those flowers was much higher (Table

1). When standardized to 1000 m2, density of female flowers was significantly higher in the Chesapeake Bay ( = 0.97) than the Hudson River ( = 0.19) based on a nested

ANOVA (F=4.859, df=1𝑥𝑥̅ , p=0.0435; Figure 5) despite finding𝑥𝑥̅ very few female flowers at

Aquia Landing. The distance (m) of patch-only transect containing female flowers averaged 412.2 m and ranged from 0 m (STK) to 876.1 m (STC). The distance of patch- only transect not containing female flowers averaged 581.4 m and ranged from 0 m (BC

VI-26

and WPP) to 2009 m (CHV). Translating these data into the ratio of patch-only transect

containing flowers to not containing flowers, the average was 0.47 and ranged from 0

(STK) to 1.0 (BC and WPP).

Male flowers were observed at all sites except CHV, NYK, STW, STC, and STK.

Through this broader sampling, males at three sites that had been entirely female in the

shoot-level sampling (BPT, STC, MIS), and females at two sites that had been only male

(GIS and TPT) were documented. Males were also found in the surface sampling at

ACR, a site that had been without flowers in July; however, failure to find any female

flowers made ACR the only site at which both sexes were not detected. Further, although

both sexes were found at STK in July, none were seen during surface sampling.

Within transects, WPP and BC had the most segments containing male flowers (20

segments totaling 497.3 m and 18 segments totaling 139 m, respectively), whereas AL had 2 segments (160 m) and the Hudson River sites that contained males had an average of 3.78 segments or 84.0 m.

Only 11 fruits were found across all sites in the Hudson River (1 at MEC and 10

at BPT). Although many female flowers that were past anthesis were seen, no others

were forming fruit, indicating lack of pollination. By contrast, most flowers in the

Chesapeake Bay were maturing into fruits. The mesocosm study provided little

additional information on flowering because only two genotypes, Stockport 864 (2 male

inflorescences) and Croton CRV7A3 (2 female flowers), flowered during the study.

VI-27

Figure 5. Density of female inflorescences (per 1000m2) at each site based on the area of the full transect.

Testing Hypotheses for Levels of Flowering

Plant Size

The Chesapeake Bay had longer maximum leaf lengths per individual than the

Hudson River ( =68.1 cm, sd= 27.8 versus =40.66 cm, sd=24.8, respectively), and this difference was confirmed𝑥𝑥̅ by a nested ANOVA𝑥𝑥̅ (F=17.75, df=1, p=0.000753). In both regions, flowering individuals tended to be near the upper end of the distribution of maximum leaf lengths (Figure 6) with an overall average length of 74.5 cm (sd=25.0) for flowering individuals and 41.3 cm (sd=25.0) for non-flowering individuals. In the

VI-28

Hudson River, maximum leaf lengths of flowering individuals were ~19 cm shorter than

in the Chesapeake Bay ( =59.3 cm [sd=22.4] and =78.5 cm [sd=24.1], respectively).

Maximum leaf lengths on𝑥𝑥̅ non-flowering individuals𝑥𝑥̅ averaged 39.2 cm (sd=24.4) in the

Hudson River and 52.0 cm (sd=25.3) in the Chesapeake Bay. Beyond the average values, the differences in the distributions of leaf lengths in the two regions is striking, with values for Hudson River individuals being right skewed (Figure 6).

Flowering individuals also fell near the upper end of the distribution of leaf widths with an overall average maximum width of 7.70 mm (sd=1.37) for flowering individuals versus 6.06 mm (sd=1.68) for non-flowering individuals (Figure 6). The overall average leaf width was slightly higher in the Chesapeake Bay ( =6.67 mm, sd=1.75) than the Hudson River ( =6.36 mm, sd=1.75), but a nested ANOVA𝑥𝑥̅ revealed no statistically significant difference𝑥𝑥̅ (F= 0.481, df=1, p= 0.499). The mean leaf width for flowering individuals in the Hudson River ( =8.11 mm, sd=1.37) was higher than the

Chesapeake Bay ( =7.60 mm, sd=1.35). The𝑥𝑥̅ average leaf width for non-flowering individuals on the 𝑥𝑥Hudson̅ River was 6.22 mm (sd=1.71), and the mean leaf width for non-flowering individuals in the Chesapeake Bay was 5.24 mm (sd=1.25; Figure 6).

Despite having shorter leaves, there were more ramets overall on Hudson River individuals ( =3.22, sd=1.72; F= 6.98, df=1, p= 0.0185) compared to Chesapeake Bay individuals (𝑥𝑥̅=2.49, sd=1.10). In both the Hudson River and the Chesapeake Bay, flowering individuals𝑥𝑥̅ had more ramets ( =4.36, sd=1.67 and =2.64, sd=1.10, respectively) than non-flowering individuals𝑥𝑥̅ ( =3.13, sd=1.70𝑥𝑥 ̅and =2.25, sd=1.06, respectively; Figure 6). 𝑥𝑥̅ 𝑥𝑥̅

VI-29

Figure 6. Maximum leaf length, leaf width, and number of ramets distribution for flowering (black; 1) vs. non-flowering (gray; 0) individuals in Chesapeake Bay (left) and Hudson River (right). Dashed lines mark the leaf length at which an individual has a 95% chance probability of not flowering (10.5 cm) based on logistic regression predictions.

VI-30

A multivariate logistic regression (pseudo r2= 0.28) revealed that as the maximum leaf length (p value=2e-16), leaf width (p value=2e-16), and total number of ramets (p value= 0.0449) per individual increased, so did the probability of an individual flowering.

Individuals of maximum leaf length 10.5 cm, paired with leaf width 6.5 mm (average) and number of ramets 2.8 cm (average) had a 95% chance of not flowering (Figure 7).

For any given leaf length, varying leaf width had a large effect on the chance of

flowering. For example, a leaf width of 8.2 mm (mean+1sd) increased the chance of flowering from 5% to 13.3%, whereas 4.7 mm wide leaves (mean-1sd) paired with the same leaf length (10.5 cm) and number of ramets (2.8) had only a 2.1% chance of flowering (Figure 7). The number of ramets had a smaller effect on the probability of flowering. Probability of flowering at a leaf length of 10.5 cm only rose to 6.1% when the number of ramets was increased from 2.8 (mean) to 4.2 (mean+1sd) and only dropped to 4.7 % for plants with 1.5 ramets (mean-1sd) (Figure 8). Individuals with a maximum leaf length ≤ 10.5 cm were all from the Hudson River, further emphasizing the smaller stature of plants from this region as compared to the Chesapeake Bay (Figure 6).

VI-31

Figure 7. Probability of an individual flowering given varying leaf lengths paired with mean 2.8 ramets and three different leaf widths: mean- 1sd =4.7 mm (light gray line); mean=6.5 mm (black line); mean+1sd=8.2 mm (dark gray line)

Figure 8. Probability of an individual flowering given varying leaf lengths (X axis) paired with mean 6.5 mm leaf width and three different numbers of ramets: mean-1sd =1.5 (light gray line); mean=2.8 (black line); mean+1sd=4.2 (dark gray line).

VI-32

Environmental Conditions

Salinity

Maximum daily salinity at Piermont Pier (tidal-saline) ranged from 3.10 ppt to

11.3 ppt and at West Point it ranged from 0.1 ppt to 3.56 ppt (Figure 9). The tidal-saline sampling sites would have maximum daily salinities between these values. At Norrie

Point, maximum daily salinities were low and had a narrow range, from 0.10 ppt to 0.15 ppt. All remaining sites sampled would be below these values.

There was no significant difference among the Hudson River sites nested in the three environments spanning the river’s tidal and salinity gradient in terms of either the proportion of individuals that were flowering (F=1.562, df=2, p=0.253) in July or the density of female flowers (F=1.364, df=2, p=0.296) in August.

Temperature

Year-round minimum daily temperature ranged from 1.3 °C to 29.1 °C in the

Chesapeake Bay, and from -0.068 °C to 26.0 °C in the Hudson River. Between the end of March and early June, minimum temperature each day was on average of 4.5 °C warmer in the Chesapeake Bay and any given minimum temperature was reached first, on average 29.1 days earlier (Figure 10). This differential continues through mid-summer, at which time the differences decrease. Temperatures become increasingly similar from

September to November.

VI-33

Figure 9. Maximum daily salinity (ppt) at Norrie Point (black), West Point (light gray), and Piermont Pier (dark gray) HRECOS monitoring stations, averaged across years from 2008 to 2018.

Figure 10. Mean of the minimum daily temperature (°C) at four Hudson River monitoring stations (gray) and 16 Chesapeake Bay monitoring stations (black) averaged across years from 1984 to 2018.

VI-34

Genotypic Diversity

Genotypic diversity was not a significant predictor of flowering (success) versus not flowering (failure) at sites in the Hudson River when tested with a logistic regression

(pseudo r2= 0.01544292, p value = 0.318).

Bed Density

A low proportion of sample frames containing plants at MEC (0.33) and CRO

(0.55) indicates low patch density that contrasts with BC, WPP, GIS, MIS, CHV, and

BPT where all samples supported plants (Table 1). The overall mean number of shoots

per sample at sites, including samples that contained no plants, ranged from 2.6 to 13.8,

and averaged 6.5. When empty samples were excluded, densities within patches ranged from averages within sites of 3.2 to 13.8 shoots per sample, with an overall average of 7.5

shoots per sample. Bed patchiness, measured by CV of numbers of shoots among

samples, was lowest at CHV (0.34) and highest at TPT (1.06) and CRO (1.0) (Figure 11).

In a logistic regression, a higher proportion of sample frames containing plants (pseudo

r2= 0.2366674, p=0.027709) and a higher CV (p=0.000136) predicted a higher ratio of

flowering to non-flowering individuals, but mean number of shoots per sample was

insignificant (p= 0.488260).

VI-35

Figure 11. Coefficient of Variation (CV) of numbers of shoots among buckets. High CVs indicate greater variation in Vallisneria density within sites.

In surface sampling, the proportion of total distance of each transect containing

plants was highest at BC, BPT, CHV, GIS, MIS, and WPP (1.0) and lowest at MEC

(0.33) and CRO (0.52). The number of gaps between patches averaged 6.65, and ranged

from 0 (STK) to 25 (CRO). A lower proportion of total transect distance containing

plants (p=9.36e-12) and more gaps between beds (p < 2e-16) predicted a higher ratio of

patch-only transect containing flowers to not containing flowers in a logistic regression

(pseudo r2= 0.05922251).

Overall Capacity for Flowering

The mean number of multilocus genotypes per Hudson River site grown at the

Appalachian Laboratory was 12 and ranged from 4 (NYK) to 19 (BPT); 9 genotypes

VI-36 from WPP and 10 from BC were planted (Table 2). Variation in number of genotypes reflects genotypic diversity at the sample sites. From 50 to 100% of genotypes from the

11 Hudson River sites flowered. For 6 of those sites, the percentage was >85% (Table 2).

In the two Chesapeake Bay sites, 70% (BC) and 44% (WPP) genotypes flowered. A nested ANOVA revealed no significant difference between percentage of flowering genotypes between the Chesapeake Bay and Hudson River, although the sample sizes were highly unbalanced regions (F= 3.373, df=1, p= 0.0934).

Table 2. Number of multilocus genotypes from each site that flowered in the greenhouse at the Appalachian Laboratory. Sex and resulting deviation from a 50:50 sex ratio based on genotypes flowering are indicated.

Deviation # of from MLGs Proportion # 50:50 Site Grown Flowered Female # Male Sex Ratio Hudson River MEC 14 1.00 9 5 0.36 PHK 17 1.00 10 7 0.41 STK 6 1.00 3 3 0.50 BPT 19 0.63 12 0 0.00 RIS 17 0.65 8 3 0.27 MIS 11 1.00 9 2 0.18 CHV 6 1.00 4 2 0.33 TPT 13 0.85 6 5 0.45 GIS 9 0.67 2 4 0.33 CRO 19 0.89 3 14 0.18 NYK 4 0.50 2 0 0.00 Chesapeake Bay WPP 9 0.44 2 2 0.50 BC 10 0.70 3 4 0.43

Pollination Potential Isolation of males and females was seen in the Hudson River that contrasted with patterns in the Chesapeake Bay. Females and males were found at all three Chesapeake

Bay sites, whereas only one sex was detected in the fine-scale shoot sampling at five

VI-37

Hudson River sites. At TPT and GIS only males were found and at MIS, STC, and BPT only females were found. AL in the Chesapeake Bay and four Hudson River sites with both sexes had strongly biased sex ratios, deviating from the maximum value of 0.5 by at least 20% (Figure 12). In the Hudson River, only CRO, PHK, RIS, and MEC had relatively balanced numbers of males and females at the scale of the sampling transect

(i.e. sex ratios close to 0.5; Figure 12). Neither genotypic diversity (r2= 0.05386, p value= 0.425) nor salinity/tidal category (F= 2.826, df=2, p= 0.102) explained variation

in sex ratios.

A chi squared test comparing sites in terms of the total number of male, female,

and non-flowering individuals was significant (χ2=891.98, p-value < 2.2e-16). More females than expected were found at BPT and there were more males than expected at

AL and TPT. BC and WPP had both more males and females than expected relative to non-flowering individuals. All Hudson River sites except for BPT had more non- flowering individuals than expected.

Differences between regions were also assessed in terms of isolation of sexes among samples within the fine-scale transects. All Chesapeake Bay sites had multiple

samples with both males and females (AL, n=4; BC, n=7; WPP, n=12). By contrast, only

one sample each at CRO and MEC had both sexes. In the other six Hudson River sites in

which both males and females were found, the minimum distance to the opposite sex

ranged from 1.16 m (PHK) to 17.13 m (STK) and averaged 7.1 m. Thus, more isolation

of males and females was seen at the scale of the whole transect and within transects in

the Hudson River than in the Chesapeake Bay.

VI-38

Figure 12. Site-level deviation from 50:50 sex ratio. Values closer to 0 indicate dominance by either sex.

In surface sampling, flowers of both sexes were found at all sites except ACR.

The total number and length of segments containing both male and female flowers were

highest at BC (9 segments totaling 113.1 m) and WPP (20 segments totaling 497.3 m).

Males and females were found in the same segment at only six of the nine Hudson River sites at which both sexes were present on the surface transect. At these sites, the sexes

were never found together in more than two segments (Figure 13) and the total length of

these segments never exceeded 66.0 m.

Among the Appalachian lab greenhouse plants, deviation from a 50:50 sex ratio

was most extreme in genotypes from NYK and BPT where all flowering genotypes were

female. By contrast, PHK, STK, TPT, WPP and BC had relatively even sex ratios (Table

2).

VI-39

Figure 13. Number of segments containing both male and female inflorescences based on full area surveyed.

VI-40

DISCUSSION

The capacity for asexual reproduction is ubiquitous in aquatic angiosperms and is considered a key feature of their evolutionary and ecological success (Philbrick and Les

1996). The balance between asexual and sexual reproduction is a dominant factor in structuring genetic diversity. Sexual reproduction generates genetically unique individuals with variation that fuels natural selection, whereas asexual reproduction generates many copies of the same genotype, potentially leaving populations and species without the diversity needed to respond to environmental change (Eckert et al. 2016). As such, the paucity of sexual reproduction in Vallisneria americana in the Hudson River is concerning. Sexual reproduction is first compromised by significantly fewer flowers

(Figure 3) and flowering shoots (Figure 4) in fine-scale samples, as well as lower densities of female flowers and fewer encounters with male flowers along large-scale transects (Table 1, Figure 5) than in Chesapeake Bay sites. Of even greater importance is lack of fruiting by the vast majority of flowers in the Hudson River. On the positive side, the fact that Hudson River plants have not lost the inherent capacity to flower (Figures 3 and 4, Table 2) implicates environmental factors in driving the observed differences.

Neither low genotypic diversity nor high salinity were explanatory factors for limited flowering. Although there was variation across sites (Figures 3 and 4), low flowering throughout the Hudson River likely precluded finding patterns related to environmental and genotypic factors within the river if any existed. The lack of effect of high salinity was surprising given that ≥5 - 9 ppt is known to limit vegetative growth and reproduction (Doering et al. 1999; French and Moore 2003; Boustany et al. 2010) and these salinities were likely seen at high tides in the tidal saline sites (Figure 9). Bed

VI-41

density had a small effect on probability of flowering, but in the opposite direction than

was hypothesized, as measured by the CV of the number of shoots per sample, the

proportion of samples with plants, the number of gaps between beds, and the proportion

of each surface transect containing plants. This result was surprising. Higher density

was originally hypothesized to yield more flowering due to positive feedback loops

between establishment and expansion of SAV beds and water clarity (Gurbisz and Kemp

2014). Water clarity yields increased reproduction in V. americana (e.g., Carter et al.

1996; Doyle and Smart 2001), generating larger, denser beds that can further slow water movement (Koch 2001; de Boer 2007). In slow moving water, more suspended particles

can sink, further increasing water clarity (Ward 1985; Gacia and Duarte 2001; de Boer

2007); however, effects of density on water quality were not tested and if they did exist,

they did not manifest in terms of more flowering in this study.

The most important proximate predictor of flowering was plant size (Figures 6, 7,

and 8), which was not surprising given the well-known importance of biomass (Titus and

Hoover 1991) and leaf length (Engelhardt et al. 2014) in flowering in Vallisneria. Water

temperature differences between regions (Figure 10) coincide with the differences in leaf

length, ramet production, and flowering. Higher water temperatures in the spring and

early summer in the Chesapeake Bay may result in increased vegetative growth earlier in

the year, leading to plants reaching reproductive size more quickly than in the Hudson

River. The link between temperature and size was demonstrated by Marsden 2015 in a

growth chamber experiment in which Chesapeake Bay plants were smaller when grown

at Hudson River temperatures and Hudson River plants were larger when grown at

Potomac River temperatures. Further development of flowers was more advanced in the

VI-42

Chesapeake Bay fine scale samples, in that none of the Hudson River plants had flowers that had reached 2 cm at approximately the same time Chesapeake Bay female flowers were open at the water’s surface. Temperature could also have contributed to lack of flowering in the mesocosm experiment. Based on previous experiments that were conducted on benches in the same greenhouse, flowering was expected. In this case, plants were grown on the greenhouse floor, which may have kept water temperatures cooler than they would have been on the benches.

Effects of temperature on growth and phenology are broadly known (e.g., Roy and Sparks 2000) and are a function of enzyme kinetics (Trudgill et al. 2005). The relationship is so strong that a number of days above a species-specific base temperature is needed for phenological phenomena (e.g., germination, bud break, flowering, hatching of insects, etc.) to occur, termed Growth Degree Days (GDD). GDD is calculated as the daily difference between the midpoint of the maximum (Tmax) and minimum (Tmin) temperature and the base temperature (Tbase) for a species:

+ = 2 . 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 𝑇𝑇𝑚𝑚𝑚𝑚𝑚𝑚 𝐺𝐺𝐺𝐺𝐺𝐺 − 𝑇𝑇𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 Although V. americana flowering is known to coincide with increases in temperature

(Titus and Stephens 1983; Carter and Rybicki 1985; Best and Boyd 2001), the GDD needed to initiate flowering in V. americana is not known. McFarland (2006) suggested temperatures needed for flower initiation based on the aquatic species Lobelia dortmanna in which the peduncle supporting the female flowers begins elongating at a threshold temperature of 19.1 °C (Szmeja 1987); however the idea of GDD has not been employed.

Elements of temperature requirements are known, but the picture is incomplete. Rybicki and Carter 2002 found that >92% of turions planted in sediment germinated at all tested

VI-43 temperatures from 13 °C to 22 °C. Further, leaf extension at the higher temperatures was substantially higher (Rybicki and Carter 2002). Temperatures from 28 °C to 32 °C are optimal for vegetative growth in the field (Titus and Adams 1979; Barko et al. 1982,

1984) and in the growth chamber (Marsden 2015).

Given the importance of temperature for development and phenology, the observed daily differences in minimum temperatures of 4.5 °C with warming in the

Chesapeake Bay preceding the Hudson River by ~29 days in the spring (Figure 10) could easily generate the observed growth differences. It is possible that the lower water temperatures in the Hudson River limit the growing season compared to the Chesapeake

Bay, affording plants little time to reach reproductive size, and causing flowers to bloom when there is little time for fruits to mature.

Such latitudinal differences in terrestrial environments are known as Hopkins’

Bioclimatic Law (Hopkins 1918) which predicts a four day delay in phenology for each poleward degree of latitude. These predictions have not been extensively tested in aquatic environments but they have been found to be accurate for Chara hispida (Calero et al. 2017) and Hydrilla verticillata (Spencer and Ksander 2001). Determining GDD needed for flowering and fruiting in Vallisneria would be extremely useful for predicting reproductive potential given annual temperatures.

If phenology is delayed by ~29 days, estimates of flowering from the fine-scale sampling would undoubtedly have been higher if flowers on individual ramets were measured in August; however, flowers developing that late will not reliably have sufficient time to develop mature fruits before the growing season ends in September due

VI-44 to declining photoperiod (Dawson 1980) and temperatures (Titus and Stephens 1983;

Titus and Hoover 1991).

Although temperature differences appear to account for the observed difference in flowering, the explanation for higher number of total ramets per individual in the Hudson

River (Figure 6) is not clear. The larger number of ramets is suggestive of increased investment in asexual reproduction. Turion production was not quantified and, thus, it is not certain if the ramets have an impact beyond the current growing season. Turion production is typically correlated with number of ramets (Titus and Hoover 1991) and genetic data indicate high levels of clonality (Table 1). It is possible that increased turion production simply results at lower temperatures; however, there is evidence for adaptive differences in ramet production. Marsden (2015) found Hudson River plants produced more turions than Potomac plants even in Potomac River growing conditions, and

Potomac plants did not produce more turions in Hudson River conditions. It is easy to see the fitness benefit of leaving more asexual propagules, especially in environments in which sexual reproduction is compromised.

The fact that Hudson River plants retain the capacity to flower yields optimism for future resilience – sexual reproduction is not precluded. That optimism is tempered by the fact that the vast majority of the female flowers that did bloom were not pollinated and failed to form fruits – in fact only 11 fruits were found in the >31 km that were surveyed within Hudson River Vallisneria beds (Table 1).

Abortion of flowers indicates pollen limitation, which can arise from myriad factors in a dioecious plant with complicated water pollination. Pollination is limited by deep water and high current, wind and waves (Sullivan and Titus 1996). Although depth

VI-45 does not vary systematically between all Chesapeake Bay and Hudson River sites, the differences in tidal range (0.6-0.8 m in the Chesapeake Bay and 1.2-1.8 m in the Hudson

River) will place female flowers at the surface where pollination takes place for less time each day in the Hudson River. Synchrony between males and females is also known to be critical for pollination (Titus and Hoover 1991).

Beyond these environmental factors, lack of pollination is likely related to strongly skewed sex ratios (Figures 4 and 12) and spatial isolation of male and female plants (Figure 13), which were much greater at more spatial scales in the Hudson River than in the Chesapeake Bay (although AL was highly male biased). Only two sites on the

Hudson River had any fine-scale samples with both male and female inflorescences in shoot-level sampling, whereas all three Chesapeake Bay sites had both sexes within the same sample. Presence of only one sex on fine-scale transects at five sites indicates males and females can be isolated by ~100 m. Later in the season, many segments along large-scale transects contained both male and female flowers in the Bay whereas this occurred in no more than 2 segments at each of six Hudson River sites. Biased sex ratios have been found in other Vallisneria populations (Doust and Laporte 1991; Lokker et al.

1994, 1997) so the Hudson River is not unique, but the number of sites throughout the river that are strongly biased is concerning.

Extensive asexual reproduction can be self-reinforcing, perpetuating more asexual reproduction while limiting future sexual reproduction (Barrett 2015). In locations with the predominance of one sex, asexual reproduction will predominate. As individual genotypes become more extensive, flowers of the opposite sex will become more and more isolated from one another, further reducing the potential for sexual reproduction. In

VI-46

populations with prolonged clonal reproduction and barriers to sexual reproduction,

patches may ultimately consist of just one or a few large clones (Eriksson 1989; Honnay

and Bossuyt 2005).

Clonal diversity is likely to interact with the spatial distribution of the species.

Dominance by a limited number of genotypes will be more likely in sites that are isolated such that dispersal is limited and rescue by genotypes is less likely. The source of the propagules also matters. Dispersal of sexual propagules leads to increased genotypic

diversity at neighboring sites and genetic rescue for nearby small, low-diversity

populations, whereas dispersal of asexual propagules amplifies the clonal extent of few

genotypes across multiple sites. Thus, it is important to consider the degree to which the

balance of clonality versus sexual reproduction represents long-term patterns in the

Hudson River and the degree to which it is the consequence of increased isolation from

the 2011 storms. If Vallisneria has survived in the Hudson River for millennia with low

genotypic diversity and low sexual reproduction, it is hard to argue that there have been

dire consequences for resilience. By contrast, if relatively few genotypes came to

dominate recently and persist (and even thrive) in the current environmental conditions,

they may be at risk for reduced resilience in the future if acclimation or adaptation is

precluded.

Conservation Implications

Reproduction is an essential component of resilience because it facilitates the

growth and expansion of beds. Demographic consequences of the lack of sexual

reproduction at so many sites on the Hudson River appear to be offset by asexual

reproduction. The immediate demographic benefits thus may have longer term

VI-47

consequences by precluding generation of new genotypes through recombination. Thus,

two key elements needed for natural selection to be effective, genetic diversity and

sexually produced progeny, are at low frequency or absent.

A key unknown is the amount of sexual reproduction necessary to generate sufficient genetic diversity and for resilience. Clearly sexual reproduction is low, but it does occur. Unpublished data (Neel and Engelhardt, Table 1) indicate that on average,

~37% of samples at sites were generated by sexual reproduction; sites varied from ~10-

79% with a standard deviation of ~18%. To begin to understand the amount of diversity

needed, one has to know about the breadth of environmental tolerances of individuals. If

individual genotypes have broad tolerances that allow acclimation, extensive clonality

will have little negative impact on resilience. By contrast, if individual genotypes vary in

their tolerance and persistence at a site depends on tolerances present in an array of

individuals, low genotypic diversity will create high risk. Thus, it is of particular interest

to determine whether extensive highly clonal genotypes have the same breadth of

environmental tolerance as arrays of genotypes generated through sexual reproduction.

Three factors lessen concern for Vallisneria in the Hudson River. First, aquatic

angiosperms are typically broadly tolerant of a range of environmental conditions and

abundant asexually produced individuals are potentially suited to conditions beyond what

currently exist (Philbrick and Les 1996). At the same time aquatic environments are

considered to be more stable than terrestrial environments (Philbrick and Les 1996). It is

common to see limited sexual reproduction and low genotypic diversity at higher

latitudes at edges of species geographic ranges (Yakimowski and Barrett 2014; Eckert et

al. 2016), and yet many of these species are often ecologically successful in these

VI-48

situations. Clearly, Vallisneria in the Hudson River is an ecological success story having

recovered rapidly after devastating storms. Genetic data collected in 2015 (Neel and

Engelhardt unpublished data, Table 1) indicate that much of this recovery can be attributed to asexual reproduction. Thus, increasing extent through stoloniferous growth within seasons that is translated across seasons through production of turions compensates for a lack of seed production.

Second, the temperature increases predicted under climate change are likely to benefit Vallisneria by generating warmer temperatures earlier in the season (Badeck et al.

2004) that allow sufficient growth for flowering and thus lengthening the growing season. Such beneficial effects on temperate seagrasses have been noted as a contrast to tropical species (Short et al. 2016). Although climate change also brings risks of storms of increased magnitude and sea level rise, in the particular case of seasonality of temperature there is a rare bit of good news.

Third, sampling each site only twice in one season provides a useful but incomplete picture of reproductive potential. In particular, the large proportion of non-

flowering ramets may yield misestimates of the true sex ratios in populations. The

proportions of ramets observed are within ranges documented elsewhere (5% (Lokker et

al. 1994), to 24% (Titus and Stephens 1983), to 42% (Doust and Laporte 1991), to 69%

(Lokker et al. 1997)). Further, females have been found to have a higher proportion of ramets with flowers (Doust and Laporte 1991). Thus, if large expanses of apparently non-reproductive plants found in the August sampling included males that did not release flowers exactly at the right time to be observed, perhaps prospects for pollination are higher than it might appear.

VI-49

Management Options

If managers decide that increasing sexual reproduction and generating new

genotypes is a desired goal, a relatively easy management solution would be to bring

males and females into closer proximity to see if fruit set increases. All but one site

(ACR) supported both sexes so such movement could be accomplished with no

introduction of genotypes across sites. One could hand pollinate females with mature

male flowers to generate seeds. Alternatively, moving different numbers of shoots of the

opposite sex into extensive single sex patches could provide longer-term benefit. If plantings were done at different densities, information on spatial distribution of sexes needed to facilitate pollination could be gained. A downside of this approach is that siring by few fathers can yield many seeds that are full or half sibs. In the Chesapeake

Bay, female fruits on average contain seeds sired by seven fathers and even so there is significant biparental inbreeding (Lloyd et al. 2018). The potential implications of many seeds being sired by one or a small number of fathers would need to be weighed in planning such an endeavor. If it was deemed necessary to increase genetic diversity beyond an individual bed, introducing males from neighboring sites to female-skewed

sites and introducing female plants from neighboring sites to male-skewed sites might be viable options for increasing the potential for sexual reproduction; however, transplanting between sites should be considered with an abundance of caution. Care should be taken to only transplant between proximal sites, given the risk of out-breeding depression

(Marsden et al. 2013) and transplanting should commence only after looking more closely at genetic relationships between populations.

VI-50

Future Studies

The present study provided a first look at mode of reproduction in V. americana in the Hudson River; however, additional studies are needed to provide a comprehensive picture of the potential of V. americana to reproduce sexually. Assessing flowering in

July and August provided information on the flower production and proximity of flowers for pollination, but how widely pollination actually occurred and led to fruit set by the end of the growing season is still unknown. Assessing flowering through to senescence in the fall and counting how many fruits are set throughout the growing season would more accurately assess the contribution of sexual reproduction in the river. Also, developing the relationship between GDD and plant and flower development by sampling more thoroughly through the early growing season would allow prediction of annual flowering potential based on water temperatures monitored by HRECOS. It is possible that sufficient sexual reproduction occurs in years with warmer spring temperatures.

GDD predictions could provide great insight into long-term potential for the balance of sexual and asexual reproduction in the Hudson River. Further, determining if the relationship varies among Vallisneria genotypes, across the salinity gradient, or among geographic areas would provide insight into how different individuals might contribute differently to resilience and about local adaptation that might affect restoration choices.

Beyond variation in GDD, assessing whether these genotypes vary in breadth of environmental tolerance will increase understanding of the risks associated with extensive clonal reproduction. Specifically, it would be instructive to ask if the extensive clones identified in the Hudson River genetic diversity sampling in 2015 have broader environmental tolerances than clones with limited distributions and to compare the tolerances of widespread versus rare genotypes to environmental conditions projected VI-51 under climate change. Examining the balance between flowering and vegetative growth in widespread versus rare genotypes across a range of growing season water temperatures would be fruitful. Taken together, these future studies would build on the intriguing results of this Polgar Fellowship to help predict the potential for sexual reproduction of V. americana in the Hudson River.

VI-52

ACKNOWLEDGMENTS

We thank the Hudson River Foundation for providing funding through the Tibor T.

Polgar fellowship, and the Cary Institute of Ecosystem Studies for hosting us in August.

Special thanks to lab member Natalia Noyes, 2018 Polgar Fellow Virginia Caponera, and

Patrick Gardullo, who assisted with field sampling, and David K. Thulman who assisted

with measuring plants. Thanks also to Katia Engelhardt, who provided data on flowering

on plants she propagated at the Appalachian Laboratory.

VI-53

REFERENCES

Badeck, F.W., A. Bondeau, K. Bottcher, D. Doktor, W. Lucht, J. Schaber, and S. Sitch. 2004. Responses of spring phenology to climate change. New Phytologist 162:295–309.

Barko, J.W., D.G. Hardin, and M.S. Matthews. 1982. Growth and morphology of submersed freshwater macrophytes in relation to light and temperature. Canadian Journal of Botany 60:877–887.

Barko, J.W., D.G. Hardin, and M.S. Matthews. 1984. Interactive influences of light and temperature on the growth and morphology of submersed freshwater macrophytes. Technical Report A-84-3. U.S. Army Corps of Engineers, Aquatic Plant Control Research Program, Vicksburg, MS.

Barrett, S.C.H. 2015. Influences of clonality on plant sexual reproduction. Proceedings of the National Academy of Sciences 112:8859–66.

Bayley, S., V.D. Stotts, P.F. Springer, and J. Steenis. 1978. Changes in submerged aquatic macrophyte populations at the head of Chesapeake Bay, 1958-1975. Estuaries 1:171.

Best, E.P.H., and W.A. Boyd. 2001. A simulation model for growth of the submersed aquatic macrophyte American Wildcelery (Vallisneria americana). Final Report ERDC/EL TR-01-5. U.S. Army Corps of Engineers, Aquatic Plant Control Research Program, Vicksburg, MS. de Boer, W.F. 2007. Seagrass–sediment interactions, positive feedbacks and critical thresholds for occurrence: a review. Hydrobiologia 591:5–24.

Boustany, R.G., T.C. Michot, and R.F. Moss. 2010. Effects of salinity and light on biomass and growth of Vallisneria americana from Lower St. Johns River, FL, USA. Wetlands Ecology and Management 18:203–217.

Brush, G.S., and W.B. Hilgartner. 2000. Paleoecology of submerged macrophytes in the Upper Chesapeake Bay. Ecological Monographs 70:645.

Calero, S., D. Auderset Joye, A. Rey-Boissezon, and M.A. Rodrigo. 2017. Time and heat for sexual reproduction: comparing the phenology of Chara hispida of two populations at different latitudes. Aquatic Botany 136:71–81.

Carter, V., and N.B. Rybicki. 1985. The effects of grazers and light penetration on the survival of transplants of Vallisneria americana Michx. in the tidal Potomac River, Maryland. Aquatic Botany 23:197–213.

VI-54

Carter, V., N.B. Rybicki, and M. Turtora. 1996. Effect of increasing photon irradiance on the growth of Vallisneria americana in the tidal Potomac River. Aquatic Botany 54:337–345.

Chesapeake Bay Program. 2018. https://www.chesapeakebay.net/what/downloads/cbp_water_quality_database_1984 _present.

Davis, F.W. 1985. Historical changes in submerged macrophyte communities of Upper Chesapeake Bay. Ecology 66:981–993.

Dawson, F.H. 1980. Flowering of Ranunculus penicillatus (Dum.) Bab. var. calcareus (R.W. Butcher) C. D. K. Cook in the river piddle (Dorset, England). Aquatic Botany 9:145–157.

Doering, P.H., R.H. Chamberlain, K.M. Donohue, and A.D. Steinman. 1999. Effect of salinity on the growth of Vallisneria americana Michx. from the Caloosahatchee Estuary, Florida. Quarterly Journal of the Florida Academy of Sciences 62:89–105.

Doust, J.L., and G. Laporte. 1991. Population sex ratios, population mixtures and fecundity in a clonal dioecious macrophyte, Vallisneria Americana. The Journal of Ecology 79:477.

Doyle, R.D., and R.M. Smart. 2001. Impacts of water column turbidity on the survival and growth of Vallisneria americana winterbuds and seedlings. Lake and Reservoir Management 17:17–28.

Eckert, C.G., and S.C.H. Barrett. 1993. Clonal reproduction and patterns of genotypic diversity in Decodon verticillatus (Lythraceae). American Journal of Botany 80:1175–1182.

Eckert, C.G., M.E. Dorken, and S.C.H. Barrett. 2016. Ecological and evolutionary consequences of sexual and clonal reproduction in aquatic plants. Aquatic Botany 135:46–61.

Engelhardt, K.A.M., M.W. Lloyd, and M.C. Neel. 2014. Effects of genetic diversity on conservation and restoration potential at individual, population, and regional scales. Biological Conservation 179:6–16.

Eriksson, O. 1989. Seedling dynamics and life histories in clonal plants. Oikos 55:231.

Findlay, S.E.G., 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.

VI-55

Findlay, S.E.G., D.L. Strayer, S.D. Smith, and N. Curri. 2014. Magnitude and patterns of change in submerged aquatic vegetation of the tidal freshwater Hudson River. Estuaries and Coasts 37:1233–1242.

French, G.T., and K.A. Moore. 2003. Interactive effects of light and salinity stress on the growth, reproduction, and photosynthetic capabilities of Vallisneria americana (wild celery). Estuaries 26:1255–1268.

Gacia, E., and C.M. Duarte. 2001. Sediment retention by a Mediterranean Posidonia oceanica meadow: the balance between deposition and resuspension. Estuarine, Coastal and Shelf Science 52:505–514.

Gurbisz, C., and W.M. Kemp. 2014. Unexpected resurgence of a large submersed plant bed in Chesapeake Bay: analysis of time series data. Limnology and Oceanography 59:482–494.

Hamberg, J., S.E.G. Findlay, K.E. Limburg, and S.A.W. Diemont. 2017. Post-storm sediment burial and herbivory of Vallisneria americana in the Hudson River estuary: mechanisms of loss and implications for restoration. Restoration Ecology 25:629–639.

Heck, K.L., and T.A. Thoman. 1984. The nursery role of seagrass meadows in the Upper and Lower reaches of the Chesapeake Bay. Estuaries 7:70.

Holling, C.S. 1973. Resilience and stability of ecological systems. Annual Review of Ecology and Systematics 4:1–23.

Honnay, O., and B. Bossuyt. 2005. Prolonged clonal growth: escape route or route to extinction? Oikos 108:427–432.

Hopkins, A.D. 1918. Periodical events and natural law as guides to agricultural research and practice. Washington, Govt. print. off.

Hudson River Environmental Conditions Observing Network (HRECOS). 2018. Water Quality Monitoring Historical Data. http://www.hrecos.org/index.php?option=com_content&view=article&id=143&Item id=54. Accessed 2018.

Hughes, A.R., and J.J. Stachowicz. 2004. Genetic diversity enhances the resistance of a seagrass ecosystem to disturbance. Proceedings of the National Academy of Sciences of the United States of America 101:8998–9002.

Koch, E.W. 2001. Beyond light: physical, geological, and geochemical parameters as possible submersed aquatic vegetation habitat requirements. Estuaries 24:1.

VI-56

Lloyd, M.W., H.R. Tumas, and M.C. Neel. 2018. Limited pollen dispersal, small genetic neighborhoods, and biparental inbreeding in Vallisneria americana. American Journal of Botany 105:227–240.

Lokker, C., L. Lovett-Doust, and J. Lovett-Doust. 1997. Seed output and the seed bank in Vallisneria Americana (Hydrocharitaceae). American Journal of Botany 84:1420–1428.

Lokker, C., D. Susko, L. Lovett-Doust, and J. Lovett-Doust. 1994. Population genetic structure of Vallisneria americana, a dioecious clonal macrophyte. American Journal of Botany 81:1004.

Lubbers, L., W.R.W. Boynton, and W.M.W. Kemp. 1990. Variations in structure of estuarine fish communities in relation to abundance of submersed vascular plants. Marine Ecology Progress Series 65:1–14.

Marsden, B.W. 2015. Enhancing Vallisneria americana restoration using knowledge of genotypic and phenotypic diversity. University of Maryland College Park.

Marsden, B.W., K.A.M. Engelhardt, and M.C. Neel. 2013. Genetic rescue versus outbreeding depression in Vallisneria americana: implications for mixing seed sources for restoration. Biological Conservation 167:203–214.

McDonald, T. 2016. SDraw: spatially balanced sample draws for spatial objects.

Nieder, W.C., E. Barnaba, S.E.G. Findlay, S. Hoskins, N. Holochuck, and E.A. Blair. 2004. Distribution and abundance of submerged aquatic vegetation and Trapa natans in the Hudson River Estuary. Journal of Coastal Research 10045:150–161.

Nieder, W.C., S. Hoskins, S.D. Smith, and S.E.G. Findlay. 2009. Distribution and spatial change of Hudson River Estuary submerged aquatic vegetation: implications for coastal management and natural resource protection. Pages 259–277 in X. Yang, editor. Remote Sensing and Geospatial Technologies for Coastal Ecosystem Assessment and Management. Springer Berlin, Heidelberg.

Orth, R.J., T.J.B. Carruthers, W.C. Dennison, C.M. Duarte, J.W. Fourqurean, K.L. Heck, A.R. Hughes, G.A. Kendrick, W.J. Kenworthy, S. Olyarnik, F.T. Short, M. Waycott, and S.L. Williams. 2006. A global crisis for seagrass ecosystems. BioScience 56:987–996.

Orth, R.J., W.C. Dennison, J.S. Lefcheck, C. Gurbisz, M. Hannam, J. Keisman, J.B. Landry, K.A. Moore, R.R. Murphy, C.J. Patrick, J. Testa, D.E. Weller, and D.J. Wilcox. 2017. Submersed aquatic vegetation in Chesapeake Bay: sentinel species in a changing world. BioScience 67:698–712.

Orth, R.J., and K.A. Moore. 1983. Chesapeake Bay: an unprecedented decline in

VI-57

submerged aquatic vegetation. Science 222:51–3.

Pebesma, E., and R. Bivand. 2018. sp: Classes and Methods for Spatial Data.

Philbrick, C.T., and D.H. Les. 1996. Evolution of aquatic angiosperm reproductive systems. BioScience 46:813–826.

Procaccini, G., and L. Piazzi. 2001. Genetic polymorphism and transplantation success in the Mediterranean seagrass Posidonia oceanica. Restoration Ecology 9:332–338.

R Core Team. 2018. R: a language and environment for statistical computing. Vienna, Austria.

Roy, D.B., and T.H. Sparks. 2000. Phenology of British butterflies and climate change. Global Change Biology 6:407–416.

Rybicki, N.B., and V. Carter. 2002. Light and temperature effects on the growth of wild celery and hydrilla. Journal of Aquatic Plant Management 40:92–99.

Sand‐jensen, K. 1998. Influence of submerged macrophytes on sediment composition and near‐bed flow in lowland streams. Freshwater Biology 39:663–679.

Sgrò, C.M., A.J. Lowe, and A.A. Hoffmann. 2011. Building evolutionary resilience for conserving biodiversity under climate change. Evolutionary applications 4:326–37.

Short, F.T., S. Kosten, P.A. Morgan, S. Malone, and G.E. Moore. 2016. Impacts of climate change on submerged and emergent wetland plants. Aquatic Botany 135:3– 17.

Spencer, D.F., and G.G. Ksander. 2001. Field evaluation of degree-day based equations for predicting sprouting of hydrilla (Hydrilla verticillata) turions and tubers. Journal of Freshwater Ecology 16:479–486.

Sullivan, G., and J.E. Titus. 1996. Physical site characteristics limit pollination and fruit set in the dioecious hydrophilous species, Vallisneria americana. Oecologia 108:285–292.

Szmeja, J. 1987. The structure of a population of Lobelia dortmanna L. along a gradient of increasing depth in an oligotrophic lake. Aquatic Botany 28:1–13.

Titus, J.E., and M.S. Adams. 1979. Coexistence and the comparative light relations of the submersed macrophytes Myriophyllum spicatum L. and Vallisneria americana Michx. Oecologia 40:273–286.

Titus, J.E., and D.T. Hoover. 1991. Toward predicting reproductive success in

VI-58

submersed freshwater angiosperms. Aquatic Botany 41:111–136.

Titus, J.E., and M.D. Stephens. 1983. Neighbor influences and seasonal growth patterns for Vallisneria americana in a mesotrophic lake. Oecologia 56:23–29.

Trudgill, D.L., A. Honek, D. Li, and N.M. Straalen. 2005. Thermal time - concepts and utility. Annals of Applied Biology 146:1–14.

Ward, L.G. 1985. The influence of wind waves and tidal currents on sediment resuspension in Middle Chesapeake Bay. Geo-Marine Letters 5:71–75.

Yakimowski, S.B., and S.C.H. Barrett. 2014. Clonal genetic structure and diversity in populations of an aquatic plant with combined vs. separate sexes. Molecular Ecology 23:2914–2928.

VI-59

PAINTED TURTLE ECOLOGY IN A FRESHWATER TIDAL MARSH: CONCLUDING SURVEY

A Final Report of the Tibor T. Polgar Fellowship Program

Virginia Caponera

Polgar Fellow

Biodiversity, Earth, and Environmental Sciences Department

Drexel University

Philadelphia, PA

Project Advisor:

Dr. Erik Kiviat

Hudsonia, Ltd.

Annandale-On-Hudson, NY 12504

Caponera V. and E. Kiviat. 2020. Painted Turtle Ecology in a Freshwater Tidal Marsh: Concluding Survey. Section VII: pp.1-32. In S.H. Fernald, D.J. Yozzo, and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

VII-1

ABSTRACT

The painted turtle (Chrysemys picta) is a common species of freshwater turtle in

New York State, but little is known of their ecology and habitat use in tidal wetlands, which are considerably different from their typical, non-tidal habitats. Understanding how painted turtles, and freshwater turtles in general, persist in tidal habitats is especially relevant in the Hudson River Estuary, because it hosts over half of the amphibian and reptile species present in New York State. Habitat use and home range data of turtles in a freshwater tidal habitat were collected over most of the active season in 2018 (April -

September) via radiotelemetry. Home range estimates were constructed using the 100% minimum convex polygon method. Microhabitat and tidal zone selection were also assessed. Turtle home ranges from 2018 were compared to data collected from 2015 through 2017, and home range varied significantly among some years. Additionally, turtles selected microhabitat based on tide stage. These data are informative to painted turtle behavioral ecology and may also provide insight into how typically non-tidal species may adapt to tidal conditions.

VII-2 TABLE OF CONTENTS

Abstract ...... VII-2

Table of Contents ...... VII-3

List of Figures and Tables...... VII-4

Introduction ...... VII-5

Methods...... VII-8

Results ...... VII-14

Discussion ...... VII-23

Acknowledgements ...... VII-29

References ...... VII-30

VII-3 LIST OF FIGURES AND TABLES

Figure 1 - Topographic map of Tivoli North Bay...... VII-9

Figure 2 - Home range map for turtles 29-40 ...... VII-19

Figure 3 - Home range map for turtles 41-46 ...... VII-20

Figure 4 - Inter-year comparison of home range values ...... VII-21

Figure 5 - Correlation between mean daily movement and Julian date ...... VII-21

Figure 6 - Tide zone selection across tide stages ...... VII-22

Figure 7 - Comparison of mean water depth between tidal zone selections ...... VII-22

Table 1 - Mass, carapace length and plastron length values of captured turtles .. VII-14

Table 2 - Mean home range areas for female and male turtles ...... VII-16

Table 3 - Individual home range values ...... VII-17

Table 4 - P-values of tidal zone selection comparisons between tide stages ...... VII-18

VII-4 INTRODUCTION

The painted turtle (Chrysemys picta) is a species of freshwater turtle commonly found throughout the eastern United States and southern Canada (Ernst and Lovich

2009). Typically, these turtles are found in still or slow-moving, fresh water bodies, such as sluggish stream reaches, marshes, lakes or ponds, in often dense populations (Van Dijk

2011; Ernst and Lovich 2009). Due to the species’ abundance and wide distribution, painted turtle ecology is well studied, including habitat use in non-tidal ecosystems.

Several studies have been conducted assessing painted turtle habitat use and movement patterns, concluding that turtles frequently move long distances (MacCulloch and Secoy

1983) and are capable of homeward orientation (Emlen 1969) in non-tidal habitats.

While painted turtles are most commonly found in slow moving fresh water bodies, they are also common in freshwater tidal habitats (Odum et al. 1984). In this distinct ecosystem, little is known of turtle habitat use or movement. A tidal wetland differs substantially from non-tidal water bodies in a number of ways. The constant fluctuation of water level, contrasting ebb and flow at various times of day, different microhabitat types between tidal zones, and potentially fast-moving currents make a tidal wetland a complex habitat that may require behavioral adaptations for reptiles (Kiviat

1989). Painted turtle ecology has not been studied extensively within a tidal wetland, but previous research conducted in Tivoli North Bay, a freshwater tidal marsh of the Hudson

River, has indicated that turtles may be altering behavior in response to tidal conditions

(Bacon and Kiviat 2018). The population density of painted turtles in Tivoli North Bay is far lower that what is typically expected in a non-tidal habitat, which may be due to high levels of emigration or mortality (Rozycki and Kiviat 1996). Radiotelemetry analysis has

VII-5 found that painted turtle home ranges in Tivoli North Bay are large and variable among

individuals, between 0.04 and 45.44 ha (Bacon and Kiviat 2018). Home range size can

serve as a measure of the quality of a habitat (Akresh et al. 2017); however, home range

is not well defined for turtles in non-tidal habitats, thus prohibiting comparison with tidal

habitats (Ernst and Lovich 2009). The size and variability of home range found in

Bacon’s 2015 work may indicate some level of patchiness of habitat quality in Tivoli

North Bay specifically, or that tidal conditions necessitate movement across long

distances. Additionally, reversing tidal currents or fluctuating water levels may disorient

turtles. In mammals, home range size has been found to negatively correlate with

seasonal resource stability (Viana et al. 2018) and resource abundance (Corriale et al.

2013), indicating that as resources become scarce, home range size increases. Similarly,

in bullsnakes (Pituophis catenifer sayi), home range area increases in poor quality

habitats (Kapfer et al. 2010). The variation in size of range between individuals, along

with low population density, suggest that Tivoli North Bay, and perhaps tidal wetlands in

general, are not ideal habitats for painted turtles; however, the presence of turtles in this

marsh and other freshwater tidal marshes does indicate that they are able to persist, and

so the ways in which they are utilizing microhabitats in the marsh to compensate for

potentially non-ideal conditions become of interest.

In a tidal wetland, the constant ebb and flow of water creates a dynamic environment, where the same geographic location can become multiple different habitats over the course of a tide cycle (Swarth and Kiviat 2009). For turtles to be able to properly thermoregulate, forage, reproduce, and avoid predation, it may be necessary to select different microhabitats as tidal conditions induce change in the environment. Eastern box

VII-6 turtles (Terrapene carolina carolina) alter habitat selection to prefer non-tidal habitats to tidal habitats when air temperatures are high (Marchand et al. 2004) and red-bellied

turtles (Pseudemys rubriventris) alter basking behavior in response to water depth and

tidal flow (Swarth 2004). Snapping turtles (Chelydra serpentina), a species also present

in Tivoli North Bay, alter habitat selection in accordance with the tide cycle by using

different tidal zones for thermoregulation at different tide stages (Kiviat 1980). Thus,

painted turtles may also exhibit similar dynamic habitat selection patterns under tidal

conditions. Understanding what microhabitats turtles are selecting, both within the larger

home range and over the course of the tide cycle, will help to define habitat requirements

within a tidal wetland and contribute to understanding how freshwater animals adapt to

estuarine wetlands. In this study, microhabitats were defined as the lower, middle, and

upper tidal zones, each occupying a third of the vertical tide range and with vegetation

and flooding duration distinct from the others. Discovering that painted turtles are, in

fact, altering microhabitat selection based on tidal conditions would be a novel finding

for the species, as painted turtles generally live in lentic or sluggish lotic water bodies,

and thus would not normally be confronted with constantly changing water levels and

fluctuating resource availability on a daily scale. Further, identifying areas of favored

habitat within the marsh may help to quantify availability of high-quality habitat, and explain some of the variation in home range size found in Bacon’s 2015 work.

In this study, painted turtle home range was quantified, and microhabitat selection was assessed within home range. Environmental variables, such as tide stage and water

depth, within microhabitats were considered. The objective of the work was to identify

VII-7 patterns between the microhabitats turtles were selecting and the tide cycle in the marsh, in order to understand how turtles might be altering behavior to persist under tidal conditions. Additionally, home range size and distance travelled per day were quantified, in order to identify changes in activity level over the active season which might be specific to tidal conditions. It was hypothesized that turtles would select microhabitats based on water availability, thus moving into low tidal zones at low tide and high tidal zones at high tide. Turtle activity (measured as mean distance travelled per day) was predicted to decrease as the active season progressed, as activity in other similar species decreases over active season duration (Beaudry et al. 2009; Kiviat 1980).

METHODS

Study Site

Tivoli North Bay is a 150 ha freshwater tidal wetland on the eastern edge of the

Hudson River (Figure 1) in Dutchess County, New York. It is bordered by upland forest to the east and the Amtrak railroad and Hudson River to the west. Approximately 24 % of the marsh is composed of tidal creeks and pools (Rozycki and Kiviat 1996). Dominant plants include narrowleaf cattail (Typha angustifolia), spatterdock (Nuphar advena), arrow arum (Peltandra virginica), wild-celery (Vallisneria americana), watermilfoil

(Myriophyllum spicatum), and pickerelweed (Ponderia cordata). The mean tidal amplitude is 1.2 m (Kiviat 1980).

Trap and Hand Capture

Turtles were captured for transmitter attachment 5 May 2018 through 22 June 2018.

Small hoopnet traps (Promar collapsible minnow traps, size medium) with floats attached

VII-8 were set in areas where painted turtle activity had been previously observed and were checked daily. Traps were baited with canned sardines in soybean oil (Frazer et al. 1990).

Environmental variables were not collected for the initial capture of turtles in traps, as tide stage, water level, temperature, and other habitat data are time sensitive, and time of turtle occupancy in the traps was unknown. To capture turtles by hand, mudflats and shallow channels were visually scanned or traversed on foot at low tide. Painted turtles

Figure 1.

Topographic map of Tivoli North Bay. The bay is bordered to the west by the Amtrak Railroad, and to the east by upland forest. The total area of the marsh is ~150 ha. Map created by Elise Heffernan, Hudsonia.

VII-9 were detected on mudflats by a partially exposed carapace or by following turtle footprints in mud. Habitat condition data were collected for individuals caught by hand.

Upon capture, body condition data were collected. Straight line carapace length and plastron length were measured to the nearest millimeter using calipers. Body mass was measured to the nearest gram using a precision spring scale (Pesola) calibrated against digital scales. Turtles were approximately aged by counting scute annuli present on the plastron. Scute annuli can reliably be used to age painted turtles up to 4 years old, after which turtles are considered mature (Stone and Babb 2005). Turtles were examined for anomalous scutes on the carapace and plastron, which included extra scutes or asymmetry, and evidence of predation or other injury, which included missing limbs, pitting, and scars on the carapace and plastron. Turtle sex was identified via body size and anterior claw length. Males can be identified by smaller body size and extended anterior claws (Ernst and Lovich 2009). Maturity of turtles was indicated by size, with turtles having plastron length > 90 mm classified as adults (Ernst 1971). Gravidity of females was determined by palpating for eggs through the soft parts anterior to the hind legs (Wilbur 1975).

Before release, turtles were assigned a unique notch code, which was used for identification upon recapture (Cagle 1939). Turtles were also outfitted with radio transmitter packages (Advanced Telemetry Systems, Isanti, Minnesota), which did not exceed 5% of the turtle’s total body weight. To attach the transmitters, two holes were drilled in the 9th and 10th marginal scutes of the turtle’s carapace. Bolts were inserted into the drilled holes, and a piece of rubber garage door gasket (approx. 1.5 x 2.5 cm) was placed over the top of the carapace in line with the bolts and secured with two # 8 size

VII-10 nuts and washers. The radio transmitter was attached to the piece of rubber gasket with epoxy putty (Water Weld). Attaching the transmitters in this way was necessary because painted turtles shed entire scutes during molt. In previous years, when transmitters were attached directly to the carapace using only epoxy, scute shedding resulted in detachment of transmitters. Turtles were released in the same location within 24 hours of capture.

Individuals are referred to by their radio channel numbers.

Radiotelemetry

Each turtle was tracked via radiotelemetry following release. A Communication

Systems receiver (model R-1000) and vinyl-coated antenna (RA-23K, Telonics, Inc) were used to detect signals of nearby individuals. Each transmitter was detectable within a radius of 100 meters. Upon locating a turtle, coordinates were recorded with a Garmin

GPS 12. Turtles were tracked approximately once per week for the duration of the study period, between 11 April and 15 August 2018. Individuals that were captured and tagged in 2018, beginning on 5 May, were tracked from the date of capture through 15 August.

Environmental Variable Data Collection

Upon hand capture or radio location, environmental variable data were recorded.

Environmental variables included air temperature, water temperature, cloud cover, wind speed (Beaufort scale), vegetation, tidal zone occupied, and water depth. Water depth was a measure of the depth of water in the bay at a given point in the tide cycle measured from the nearest adjacent low tide zone, not the depth of water above the turtle in its tracked location. While tide stage can be used to approximately represent water depth, measuring depth directly controls for fluctuations in water flow due to weather or the

VII-11 lunar cycle. The marsh was divided into three tidal zones: lower, middle, and upper. For this study, the lower tidal zone classification contained mudflats, spatterdock and pickerelweed beds, and, in some cases, shallow subtidal habitats. The lower tidal zone is

exposed at the lowest tide stages (spring low tides), including the uppermost subtidal

portions of the habitat, which are slightly below Mean Low Water (MLW). The middle

tidal zone was the transition area between lower and upper tide zone. This area was at a

steep angle (~45 º) in many places, contained sparse vegetation, and was exposed and

submerged periodically during the tide cycle. The upper tidal zone consisted of cattail

stands on the creek banks and some small, shallow channels deeper in the stands. This

zone was only submerged at the highest tide stages.

Statistical Analyses

Turtle home range size was calculated using the 100% minimum convex polygon

method (ArcMap version 10) (Powell 2000). The Spearman rank correlation was used to

assess the relationship between home range size and body mass, and home range size and

number of radio locations. A 95% confidence interval was used to determine differences

in home range size between males and females from 2018, and female range size between

2018 and 2015. An unpaired t-test was used to compare body size between 2015 and

2018 females. To identify changes in activity over the course of the season, mean

distance travelled per day was calculated by dividing the total distance between two

consecutive tracked locations by the number of days that had elapsed between radio

locations (Beaudry et al. 2009). The Spearman rank correlation was used to evaluate

relationships between activity (measured as mean daily movement), day of year, and air

VII-12 temperature at time of location. Differences in distance travelled per day between males

and females were assessed using the Welch two sample t-test. Chi-squared tests and pairwise t-tests were used to determine significance in tide zone selection and tide stage.

The Welch two sample t-test was used to determine significance between water level, air and water temperature, and tidal zone selection. Middle tide zone data were eliminated from analysis due to low number of observations (n=4). Means are presented ± SD.

VII-13 RESULTS

Capture

Sixteen turtles were captured, including four males, ten females, and two juveniles. One female (# 29) was a recapture from 2015, as detected from her notch code. Seven turtles were caught by hand and seven were caught in traps. The mean and

SD of turtle masses were: females (n=10), 436.5 ± 99.8 g; males (n=4), 298.8 ± 94.2 g; and juveniles (n=2), 71.5 ± 12.0 g. The mean carapace length (CL) and plastron length

(PL) of turtles were: females, 150.7 ± 12.8 mm (CL) ,140.5 ± 12.3 mm (PL); males,

135.6 ± 15.6 mm (CL), 122.8 ± 16.2 mm (PL); juveniles, 75.5 ± 2.1 mm (CL), 69 ± 3.0 mm (PL) (Table 1).

Table 1. Mass, carapace length (CL) and plastron length (PL) values of captured turtles.

Female Male Juvenile Number 10 4 2 Mass (g) 436.5 ± 99.8 298.8 ± 94.2 71.5 ± 12.0 CL (mm) 436.5 ± 99.8 135.6 ± 15.6 75.5 ± 2.1 PL (mm) 140.5 ± 12.3 122.8 ± 16.2 69 ± 3.0

Inter-year Comparison of Body Size

Body mass of females did not differ between 2015 (468 ± 68.5, n=9) and 2018

(436.5 ± 99.8, n=10, p=0.44, unpaired t-test). Female carapace length also did not differ between years: 2015 (150 ± 7.7, n=9); 2018 (150.7 ± 12.8, n=10, p=0.91, unpaired t-test).

Body size was not compared between years for males due to small sample size.

Predation and Scute Anomaly

VII-14 Evidence of predation attempts or infection (scars, pitting) were detected on nine of the 14 captured turtles. Two of the turtles had missing limbs: a female missing both front feet, and a male missing the left front foot. There was also a high incidence of scute anomaly. Out of 18 captures, five turtles had anomalous scutes, two turtles had 13 marginal scutes on both sides of the carapace, and three turtles had asymmetrical carapacial scutes.

Home Range Size

Sixteen turtles were tracked in 2018. Fourteen were adults captured in 2018, and two were females tagged in 2017 with active transmitters. Turtles were each tracked 4-17 times over the study period. Variation in the number of radio-locations was due to differences in when transmitters were attached (turtles that were captured earlier in the season had more radio-locations), and difficulty consistently locating some individuals due to the large expanse of marsh with limited canoeable channels. Turtles that had been previously tagged during 2017 were inconsistently located after June, indicating that their transmitters lost function. Data from these individuals were excluded from home range analyses, in order to assess activity over the entire duration of the active period. Data from these individuals were included in comparisons between mean daily movement and date, and habitat selection analyses. The mean number of radio-locations for an individual was 9.3. The mean home range size was 5.25±5.35 ha. For males (n=4), the smallest home range was 1.75 ha, the largest was 20.48 ha, and the mean was 8.04 ± 8.47 ha. For females (n=10), the smallest home range was 0.04 ha, the largest was 11.59 ha, and the mean was 4.14 ± 3.55 (Tables 2 and 3). The number of times an individual was

VII-15 located was not significantly correlated with the home range size (rho=0.358, p=0.172, n=14). The one female recapture (# 29) had a home range area of 11.59 ha, which was larger than her home range area of 1.76 ha in 2015 (Bacon and Kiviat 2018).

Interestingly, this turtle's geographic range was within the same region of the marsh between years.

Home range size was not significantly correlated with carapace length for males

(rho= -0.6, p=0.42, n=4) or females (rho=-0.2, p=0.58, n=10). Male home range size was not significantly different from female home range size (95% CI: males: 1.07-15.06; females: 2.29-5.47). Female home ranges were compared using historic tracking data from 2015-2018. Female home ranges from 2015 (n=7) were significantly larger than female home ranges from 2018 (n=10) (95% CI: 2015:15.28-29.21; 2018: 2.29-5.99,

Figure 4). Home range data collected during the intervening years (2016, 2017) were also compared. Female home range was smaller during 2016 (n=16) compared to 2015 (n=7)

(95% CI: 2016: -6.33-7.599, 2015:15.28-29.21). There was no significant difference found between female home range size in 2017 (n=3) and any other year (95% CI: 2017:

3.62-17.56). Male home range size between 2015 and 2018 was not assessed due to lack of male tracking in 2015.

Table 2. Mean home range areas for female and male turtles.

Mean (ha) Minimum (ha) Maximum (ha) Total Sample 5.25±5.35 0.04 20.48 Females 4.14 ± 3.55 0.04 11.59 Males 8.04 ± 8.47 1.75 20.48

VII-16 Table 3. Individual home range values. Individuals are identified by their radio channel numbers (30-46), and . sorted in order of increasing size. Turtle Range (ha) Sex 40 0.04 Female 41 0.09 Female 43 0.91 Female 36 1.75 Male 42 3.10 Female 46 3.13 Female 30 3.99 Male 31 4.93 Female 32 5.03 Female 44 5.31 Female 34 5.96 Male 45 7.26 Female 29 11.59 Female 33 20.48 Male

Distance Travelled Per Day

The mean distance between tracked locations was 203.9 ± 340.4 m, and the mean number of days between radio locations was 8.2 ± 5.6. The hypothetical mean distance travelled per day was: 33.2 ± 60.9 m for all turtles, 40.9 ± 63.0 m for males, and 30.9 ±

60.4 m for females. The maximum distance calculated was 534.0 m (a female), and minimum was 0.32 m (also a female). Mean distance travelled per day did not differ significantly between males and females (p=0.44, t=0.78, df=48.142, n=132). There was no correlation between day of year (Julian date) and distance travelled per day, though the longest distances travelled did occur later in the season (rho=0.112, p=0.197, n=132,

Figure 5).

Tidal Zone Selection

Each time a turtle was located, tidal zone and tide stage were recorded. The

VII-17 proportion of turtles occupying upper and lower tidal zones changed significantly with

some tide stages (x2=33.398, p=0.0025, df=14, n=151). Middle tidal zone data were eliminated from analyses because very few turtles were found in this zone (n=4). Turtles chose the upper tidal zone significantly more often during high tide than during low tide

(p=0.0001) or late ebb tide (p=0.0206). Turtles chose the lower tidal zone significantly more frequently during low tide than during mid-flood (p=0.036) or high tide (p=0.0001, n=151, pairwise t-test, Bonferroni adjustment, Figure 6, Table 4)

Water depth (independent of the direction of movement of water in the marsh) was also considered. Turtles chose the low tidal zone when water levels were lower, and the high tidal zone when water levels were higher (p=0.0026, t=-3.0634, df=133.88, n=141, Welch two sample t-test, Figure 7). Water temperature (p=0.99, t=-0.0005, df=141.82, n=147) and air temperature (p=0.457, t=-0.746, df=136.03, n=147, Welch two

sample t-test) were not significantly different between upper or lower tidal zone

selection.

Table 4. P-values of tidal zone selection comparisons between tide stages (pairwise t-test with Bonferroni adjustment). Significance indicated with an asterisk. Tide stages that are significantly different from one another (*) have different proportions of turtles utilizing the upper vs lower tidal zone.

Mid Late Early Low Early Flood Flood Flood High Ebb Mid Ebb Low ------Early Flood 1.000 ------Mid Flood 0.036* 1.000 - - - - - Late Flood 0.3257 1.000 1.000 - - - - High 0.0001* 1.000 1.000 1.000 - - - Early Ebb 0.7928 1.000 1.000 1.000 1.000 - - Mid Ebb 1.000 1.000 1.000 1.000 1.000 1.000 - Late Ebb 1.000 1.000 1.000 1.000 0.0206* 1.000 1.000

VII-18

Figure 2 . Home range map for turtles 29-40. Individual turtles are represented by different colors.

VII-19 Figure 3: Home range map for turtles 41-46. Individual turtles are represented by different colors.

VII-20

Figure 4. Inter-year comparison of home range values. Boxplots compare home range of females tracked from 2015 through 2018. (Bold line=median, box upper= third quartile, box lower= first quartile, whiskers = minimum and maximum range values, dots = outliers. Female home ranges were significantly smaller during 2018 (n=10, 95% CI: 2.29- 5.99) and 2016 (n=4, 95% CI: -6.33-7.60) compared to 2015 (n=7, 95% CI:15.28-29.21). Home ranges from 2017 did not differ significantly from any other year (n=3, 95% CI: 3.62-17.56).

Figure 5. Correlation between mean daily movement and Julian date. Mean daily movement is defined as the distance between two subsequent tracked locations, divided by the number of days elapsed between radio locations. Colors represent individual turtles. Length of mean daily movement (m=33.2±60.9) did not change over time (p=0.197, rho=0.113, n=133).

VII-21 Figure 6. Tide zone selection across tide stages. Upper tidal zone selection is depicted in yellow, lower tidal zone selection in red. Letters on bars represent significance relationships. Zone selection was significantly different between low and high tide (A), high and late ebb tide (B), and low and mid flood tide (C).

Figure 7. Comparison of mean water depth between tidal zone selections. Bold line= median, box upper= third quartile, box lower= first quartile, whiskers = minimum and maximum range values, dots = outliers. When turtles chose the upper tidal zone (yellow), mean water depth was higher, and when turtles chose the lower tidal zone (red), mean water depth was lower (p=0.0026, t=-3.0634, df=133.88, n=141, Welch two sample t-test).

VII-22 DISCUSSION

From this survey, it is apparent that turtle home range size is highly variable among individuals, and between years. The persistence of turtle #29 within the same general area of the marsh 2015-2017 indicates that at least some individuals do have a

sense of direction and “home” within a larger ecosystem. The difference in home range

for this individual was 11.59 ha in 2018 and 1.76 ha in 2015 (Bacon and Kiviat 2018).

However compelling the behavior of turtle #29 may be, the behavior of one turtle may

not be used to draw conclusions about the overall quality of the marsh.

Generally, the mean home range size of individuals was smaller and less variable

during 2018 (4.14 ± 3.55 ha) than during 2015 (14.57 ± 15.94 ha), and home range can

fluctuate between consecutive years (Figure 4). Most home range polygons included a

number of different habitat types, including lower intertidal spatterdock beds, upper

intertidal inter-creek cattail marsh, and intertidal and subtidal pools and creeks (Figures 2

and 3). The smaller home range found during 2018 compared to 2015 may indicate that

the marsh has increased in habitat quality (Kapfer et al. 2010; Viana et al. 2018).

Estuaries are especially dynamic systems and food, or another factor could have changed.

The conflicting results of decreased mean range size for the population but increased

range for an individual (#29) found between 2015 and 2018 indicate the need for further

home range analysis work. In this system, an inter-year comparison of home range for each individual, rather than mean home ranges, might be more valuable for identifying annual fluctuations in habitat quality. While home range size decreased significantly, which may indicate higher resource availability, mean turtle mass did not differ significantly between 2015 and 2018; however, turtle biomass does not necessarily

VII-23 correlate with productivity of an ecosystem (Iverson 1982), and so other indicators of habitat quality, such as vegetation productivity or prey availability, should be considered in future work. An assessment of patchiness of low and high quality habitat between years and seasonally within a year may also help to explain some of the variation in home range size and long distance forays.

Neither sex nor body size was a significant predictor of home range. While mean male home range (8.04 ha) was larger than the mean female home range (4.14 ha), these values for each group were not significantly different from each other. Similarly, male painted turtles have been found to travel farther than female painted turtles in a river, and so a larger sample size may reveal differences in home range based on sex (MacCulloch and Secoy 1983). Home range size was positively correlated with number of radio locations, though not significantly (rho=0.358, p=0.172, n=16).

The goal of identifying microhabitats within home range was to determine what microhabitats within the larger ecosystem were necessary to turtles, and how turtles might be altering behavior in response to tidal conditions. Within the marsh, the microhabitat distinctions considered were between lower, middle, and upper tidal zones.

Turtles most often occupied upper and lower tidal zones, and more interestingly, preferred different tidal zones at different stages of the tide cycle. Turtles chose the lower tidal zone over the upper tidal zone at low tide, and the upper tidal zone over the lower tidal zone at high tide. Further, when mean water level was considered, separate from tide stage and direction of water movement, turtles more frequently chose the upper tidal zone when more water was available, and the lower tidal zone when water levels in the marsh were lower. Few turtles were found in the middle tidal zone, perhaps because this area

VII-24 was steep and did not offer much surface area in the parts of the marsh where turtles were

tracked. It is novel that turtles are moving between two very different lower and upper

tidal zone microhabitats and taking advantage of each based on the availability of water,

especially when the more homogenous nature of their typical, non-tidal habitat is

considered. This behavior may be serving a similar function to tidal movements made by

snapping turtles, in which individuals have been found to move from upper tidal zones to

lower tidal zones as water flows out of the habitat, and the reverse as the tide turns

(Kiviat 1980). If turtles did not appear to show differentiation of tidal zone selection, and

instead occupied the lower tidal zone at all water depths and tide stages, then it may be

inferred that they are not adapting behavior to a tidal habitat. Movement between

intertidal zones is not required in a non-tidal wetland (for obvious reasons), and so the alteration of habitat selection behavior indicates adaptation necessary to persist.

While knowing that turtles are using each tidal zone and altering selection of the tidal zone based on tide stage, it is unclear what exactly is driving the turtles to move between the upper and lower zones. In the lower tidal zone, turtles have access to benthic prey objects and aquatic vegetation, and do not risk aerial exposure as the tide flows out of the marsh. The upper tidal zone, however, is distinct from the lower tidal zone both in its elevation, period of submergence, and vegetation (and presumably prey options).

Painted turtles are generalist foragers, with a diet including macroinvertebrates and vegetation (Ernst and Lovich 2009). The seeds of spatterdock have been found to be an important component of painted turtle diets, along with invertebrate prey (Padgett 2010).

Spatterdock is abundant in the lower tidal zone, which does not explain occupancy of the upper tidal zone; however, as spatterdock seeds are only seasonally available, turtles may

VII-25 have to change foraging strategies based on food availability. Arrow arum, a plant

common in the middle and upper tidal zone in Tivoli North Bay, has been found in the

snapping turtle diet (Lagler 1943), and so it is not unlikely that painted turtles may also

be utilizing it. Dietary analysis of painted turtles should be conducted to determine if

turtles may be harvesting some prey from upper tidal zone vegetation. The upper tidal

zone may also assist turtles in making more direct movements between different areas of

the marsh, rather than navigating a long series of winding lower tidal zone channels.

Additionally, the upper tidal zone does contain shallow channels, which may provide

refuge from predators and competition, such as snapping turtles, which are mainly

benthic dwellers.

Although sample sizes are small, the high proportions of scars and anomalous

scutes suggest that North Bay may be a stressful environment for painted turtles. Scars

that resemble scratches were probably teethmarks from mammalian predation attempts,

and the most likely predator is the raccoon (Procyon lotor), the population of which has increased greatly in the Hudson Valley during the past fifty years. Anomalous scutes indicate developmental stress that could be caused by environmental contaminants (Bell et al. 2006)

Turtle activity did not decrease with the progression of the active season. For some individuals, activity (measured as mean daily movement) increased, though not significantly (Figure 5). Spotted turtles and Blanding’s turtles both exhibit a distinct pattern of seasonal activity, with activity peaking in early summer and falling in late summer (Beaudry et al. 2009). The stability of activity levels through the summer in

Tivoli North Bay poses interesting questions about the energy requirements necessary in

VII-26 a tidal wetland, or the overall habitat quality of Tivoli North Bay. Future work to assess activity over the active period might include higher sampling effort and sampling though late summer and fall, to determine a definite “end” to activity. A necessary next step may be to create an index of what a “high quality” habitat is, in order to determine if the behaviors exhibited by this population can be due to the tidal nature, low quality resource availability, or both.

Additionally, the ability of painted turtles to alter behavior significantly is an indication of the adaptability of the species and provides insight into their behavioral ecology. While this study did not provide a definitive home range area for individuals, the continued variability in home range size between individuals and years does pose further questions about the quality of the habitat, or painted turtle ecology in general.

Although the painted turtle is not a species of conservation concern, studying the habitat requirements of a freshwater turtle in a tidal ecosystem can provide a framework for the future conservation efforts of other, less common freshwater turtle species, such as the eastern red-bellied turtle (Pseudemys rubriventris) or the spotted turtle (Clemmys guttata), in tidal wetlands. It is not uncommon for a species’ distribution to include both tidal and non-tidal wetlands, and so this work assessing adaptations to a tidal existence may be the first step in characterizing larger scale adaptive behavior (Swarth and Kiviat

2009). Understanding how organisms interact with tidal conditions, and how habitat requirements may change in tidal conditions, will be critical to effective conservation

(Breisch 2011). Additionally, the changing availability of freshwater tidal marsh habitat in the face of climate change must be considered. As climate change results in rising sea levels, freshwater tidal marshes will be especially at risk due to habitat loss and increased

VII-27 salinity, and so understanding how species are currently utilizing freshwater marshes will help predict the impact of habitat loss in the future (Barendregt and Swarth 2013).

VII-28 ACKNOWLEDGEMENTS

We thank Elise Heffernan for assistance in map construction; Jason Tesauro, Lisa

Kelley, and Angela Sirois-Pitel for equipment lending and advice; Emma Kelsick, Erik

Hedlund, Steve Coleman, Hannah Kowalsky, Sarah Welz, Patrick Baker, Alice Lindsay,

Aldo Grifo-Hahn, Carrie Perkins, Rory Kuczek, and Olive Chen for their assistance in field data collection; and the Hudson River Foundation for supporting this research.

Turtles were handled and radio-tracked under New York State License to Collect or

Possess: Scientific 871 and a Temporary Revocable Permit, both from the New York

State Department of Environmental Conservation. This paper is a Hudsonia – Bard

College Field Station Contribution.

VII-29 REFERENCES

Akresh, M.E., D.I. King, B.C Timm, and R.T Brooks. 2017. Fuels management and habitat restoration activities benefit eastern hognose snakes (Heterodon platirhinos) in a disturbance-dependent ecosystem. Journal of Herpetology 51: 468-476.

Bacon, R. J., and E. Kiviat. 2018. Ecology of painted turtles in a freshwater tidal marsh, Tivoli North Bay, New York. Section II:1-29 pp. In S.H. Fernald, D.J. Yozzo and H. Andreyko (Eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2015. Hudson River Foundation, New York

Barendregt, A., and C.W. Swarth. 2013. Tidal freshwater wetlands: Variation and changes. Estuaries and Coasts 36: 1-12.

Beaudry, F., P.G. deMaynadier, M.L. Hunter, and M.L. Hunter, Jr. 2009. Seasonally dynamic habitat use by spotted (Clemmys guttata) and Blanding’s turtles (Emydoidea blandingii) in Maine. Journal of Herpetology 43: 636-645.

Bell, B., Spotila, J.R., Congdon, J. 2006. High incidence of deformity in aquatic turtles in the John Heinz National Wildlife Refuge. Environmental Pollution 142: 457-465.

Breisch, A. 2011. Herpetofauna of the Hudson River watershed: A short history. pp. 41- 51 in Robert E. Henshaw (Ed.), Environmental History of the Hudson River. State University of New York Press..

Cagle, F. 1939. A system of marking turtles for future identification. Copeia 3:170-173.

Corriale, M.J., E. Muschetto, and E.A. Herrera. 2013. Influence of group sizes and food resources in home-range sizes of capybaras from Argentina. Journal of Mammalogy 94:19-28.

Emlen, S. 1969. Homing ability and orientation in the painted turtle Chrysemys picta marginata. Behaviour 33: 58-76.

Ernst, C.H. 1971. Growth of the painted Turtle, Chrysemys picta, in southeastern Pennsylvania. Herpetologica. 27: 135-141.

Ernst, C.H., and J.E. Lovich. 2009. Turtles of the United States and Canada. Second edition. Johns Hopkins University Press, Baltimore, MD. pp. 184-211.

Frazer, N.B., J.W. Gibbons, and T.J. Owens. 1990. Turtle trapping: Preliminary tests of conventional wisdom. Copeia 1990 (4): 1150-1152.

Iverson, J.B. 1982. Biomass in turtle populations: A neglected subject. Oecologia 55: 69- 76.

VII-30

Kapfer, J.M., C.W. Pekar, D.M. Reineke, J.R. Coggins, and R. Hay. 2010. Modeling the relationship between habitat preferences and home-range size: a case study on a large mobile colubrid snake from North America. Journal of Zoology 282: 13-20.

Kiviat, E. 1980. A Hudson River tidemarsh snapping turtle population. Transactions of the Northeast Section, Wildlife Society 37:158-168.

Kiviat, E. 1989. The role of wildlife in estuarine ecosystems. pp. 437-475 in J.W. Day (Ed.), Estuarine Ecology. John Wiley & Sons, New York.

Lagler, K.F. 1943. Food habits and economic relations of the turtles of Michigan with special reference to fish management. American Midland Naturalist 29: 257-312.

MacCulloch, R., and D. Secoy. 1983. Movement in a river population of Chrysemys picta bellii in Southern Saskatchewan. Journal of Herpetology 17: 283-285. doi:10.2307/1563834

Marchand, N.M., M.M. Quinlan, and C.W. Swarth. 2004. Movement patterns and habitat use of eastern box turtles at the Jug Bay Wetlands Sanctuary, Maryland. pp. 55-62 in C.W. Swarth, W. Roosenburg, and E. Kiviat (Eds). Conservation and Ecology of Turtles of the Mid-Atlantic region. Bibliomania!, Salt Lake City, UT.

Odum, W., T. Smith III, J. Hoover, and C. McIvor. 1984. The ecology of tidal freshwater marshes of the United States east coast: A community profile. U.S. Fish and Wildlife Service. FWS/OBS-83/17. 177 pp.

Padgett, D.J., J.J. Carboni, and D.J. Schepis. 2010. The dietary composition of Chrysemys picta picta (eastern painted turtles) with special reference to the seeds of aquatic macrophytes. Northeastern Naturalist 17: 305-312.

Powell, R.A. 2000. Animal Home Ranges and Territories and Home Range Estimators. pp. 65-103. in L. Boitani and T.K. Fuller (Eds), Research Techniques in Animal Ecology. Columbia University Press, New York.

Rozycki, C., and E. Kiviat. 1996. A low density, tidal marsh, painted turtle population. Section VII: 26 pp. In J.R.Waldman, W.C. Neider, and E.A. Blair (Eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 1995. Hudson River Foundation, New York.

Stone, P.A., and M.E. Babb, 2005. A Test of the annual growth line hypothesis in Trachemys scripa elegans. Herpetologica 61: 409-414.

Swarth, C.W. 2004. Natural history and reproductive biology of the red-bellied turtle (Pseudemys rubriventris). pp.73-83 in C.W. Swarth, E. Kiviat, W. Roosenburg, Eds. Conservation and Ecology of Turtles of the Mid-Atlantic

VII-31 Region.Bibliomania! Salt Lake City, UT.

Swarth, C.W., and Kiviat, E. 2009. Animals of tidal freshwater wetlands in North America. pp. 71-88 in A. Barendregt, D. Whigham, A. Baldwin (Eds), Tidal Freshwater Wetlands. Backhuys Publishers, Leiden, The Netherlands.

Viana, D.S., Granados, J.E., Fandos, P. Perez, J.M., Cano-Manuel F.J., Burton, D., Fandons, G., Angeles Parraga Aguado, M., Figuerola, J., and Soriguer, R.C. 2018. Linking seasonal home range size with habitat selection and movement in a mountain ungulate. Movement Ecology 6 (1): 1-11.

Van Dijk, P.P. 2011. Chrysemys picta. The IUCN Red List of Threatened Species 2011. doi: e.T163467A97410447.

Wilbur, H. 1975. A growth model for the turtle Chrysemys picta. Copeia 2:337-343.

VII-32 HUMAN IMPACT ON RAMSHORN-LIVINGSTON, A HUDSON RIVER FRESHWATER TIDAL MARSH

A Final Report of the Tibor T. Polgar Fellowship Program

Elizabeth Thompson

Polgar Fellow

Environmental Biology Columbia University New York, NY 10027

Project Advisor:

Dr. Dorothy Peteet Lamont-Doherty Earth Observatory Palisades, NY 10964

Thompson, E. and D. Peteet. 2020. Human Impact on Ramshorn-Livingston, a Hudson River Freshwater Tidal Marsh. Section VIII: pp.1-40. In S.H. Fernald, D.J. Yozzo and H. Andreyko (eds.), Final Reports of the Tibor T. Polgar Fellowship Program, 2018. Hudson River Foundation.

VIII-1 ABSTRACT

A historical approach to a site can provide insight into how past events have shaped it as well as the state of the ecosystem in question. Analysis of a 1-m sediment core at Ramshorn-Livingston

Marsh, one of the largest and most well-preserved freshwater tidal marshes in New York State, reconstructed about 250 years of ecosystem response to human settlement. From 100-68 cm, large and heavy seeds are found in the core, possibly indicating a heavier flood regime capable of carrying these seeds. From 68 -20 cm, Asteraceae seeds and violets are present in the macrofossil profile and there is an increase in inorganic sediment from 0.12 g to 0.28 g per sample, both indicators of human

settlement as weedy species colonized a more open landscape with upland erosion into the marsh.

There is a sharp decrease in inorganic sediment from 0-20 cm depth in the core from a weight of 0.33

g to 0.05 g per sample and macrofossil analysis shows a switch from a diverse community of sedges

to a monoculture of invasive cattail during the same period. Lead, copper, and calcium profiles

obtained through x-ray fluorescence spectroscopy (XRF) all reflect human activity. Lead increases from 50 cm to 18 cm depth and subsequently decreases rapidly. Copper has increased steadily from

30 cm to 10 cm depth, when it sharply declines again to the present. XRF analysis also reflects a

decrease in inorganic sediment, as aluminum, titanium, and silica all decrease in concentration from

20 cm to the top of the core. These findings reveal that an important freshwater tidal marsh which is

thought to be healthy and intact has undergone serious transitions over the past 300 years that may

put its ecosystem services at risk.

VIII-2 TABLE OF CONTENTS

Abstract………………………………………………………………………… VIII-2

Table of Contents………………………………………………………………. VIII-3 List of figures…………………………………………………………………... VIII-4

Introduction…………………………………………………………………….. VIII-5

Methods………………………………………………………………………… VIII-15

Sediment coring………………………………………………………… VIII-15

Loss-on-ignition………………………………………………………… VIII-16

Macrofossils……………………………………………………………. . VIII-17

X-ray fluorescence……………………………………………………… VIII-17

Results…………………………………………………………………………... VIII-17

Organic/inorganic matter content……………………………………….. VIII-18

X-ray fluorescence……………………………………………………… VIII-20

Macrofossils…………………………………………………………….. VIII-22

Discussion………………………………………………………………………. VIII-23

Zones 1, 2, and 3………………………………………………………… VIII-24

Soil accretion……………………………………………………………. VIII-25

Heavy metals……………………………………………………………. VIII-27

Calcium sources…………………………………………………………. VIII-27

Conclusions……………………………………………………………………… VIII-29

Recommendations……………………………………………………………….. VIII-30

Acknowledgements……………………………………………………………… VIII-31

References.……………………………………………………………………… VIII-33

Appendix A……………………………………………………………………… VIII-40

VIII-3

LIST OF FIGURES

Figure 1 – Cross-section through a freshwater marsh …………………………… VIII-5

Figure 2 – Spawning area of the American shad…………………………………. VIII-6

Figure 3 –Influx of sediment reduces microtopographic variation and a sedge meadow is replaced by 1 to 2 opportunistic species ………………… VIII-11

Figure 4 –Ramshorn-Livingston Marsh in 1770 and satellite image of the coring site, showing the town of Catskill north of the marsh ..……………… VIII-13

Figure 5 – Dachnowski corer…………………………………………………….. VIII-16

Figure 6 – Core collected from Ramshorn-Livingston Marsh…………………… VIII-18

Figure 7 – Inorganic and organic content of the core by depth………………….. VIII-19

Figure 8 –Normalized relative x-ray fluorescence showing abundance of silica, titanium, and potassium as a function of depth in the sample core….. VIII-20

Figure 9 - Normalized relative x-ray fluorescence showing abundance of lead, zinc, and copper as a function of depth in the sample core ………… VIII-21

Figure 10 - Normalized relative x-ray fluorescence showing abundance of calcium, chlorine, and bromine as a function of depth in the sample core ………………………………………………………. VIII-21

Figure 11 – Macrofossil profile of Ramshorn-Livingston Marsh…………..…… VIII-22

Figure 12 –A comparison between the change in inorganic weight per sample in the Ramshorn-Livingston core to the sedimentation rate at Chesapeake Bay……………..……………………………………... VIII-26

Figure 13 – Relationship between calcium and chlorine…………………..……. VIII-28

VIII-4 INTRODUCTION

Nestled between the booming metropolis of New York City and the vast wilderness of

the Adirondacks are 10,000 acres of freshwater tidal marshes along the Hudson River

(Friedlander 2016). Freshwater tidal marshes make up a tiny fraction of New York’s natural ecosystems, occurring mainly along a 160 km stretch of the Hudson River between Newburgh and Troy (Edinger et al. 2014). They are characterized by a salinity level lower than 0.5 psu, a tidally flooded hydrological regime, and a plant community adapted to tidal action and submergence (Tiner 2013) (Figure 1). This unique combination of factors shapes an ecosystem that is high in biodiversity, an effective carbon sink, and a considerable filter for pollutants and sediments in runoff.

Figure 1: Cross-section through a freshwater marsh (Mitsch et al. 2009).

Tidal marshes have high nutrient fluxes driven by a rapid turnover of vegetation, a slow release of nutrients captured in runoff, and nutrients brought in by regular flooding (Mitsch et al.

2009). This creates an abundance of nutrients at the bottom of the food chain that extends

VIII-5

upward, supporting a diverse community of plants, fish, and birds. Economically valuable fish,

such as smallmouth bass (Micropterus dolomeieu), largemouth bass (Micropterus salmoides),

American eel (Anguilla rostrata), and American shad (Alosa saidissima) utilize freshwater tidal marshes for their availability of food and protection from the higher flow rates and predation in the river (NOAA Fisheries 2013). The American shad was once emblematic of New York. It generated a huge fishing industry and a fierce following among residents as it was valued for its flesh and roe. Their decline from 138 distinct populations along the Hudson River to 68 is partly due to the loss of wetland habitat in the freshwater tidal portion of the Hudson River, where they spawn and spend their first few years of development (Woods 2012) (Figure 2).

Figure 2: Spawning area of the American shad. Adapted from Nack et al. (2014).

Freshwater tidal marshes also attract many migratory ducks and waterfowl. These species, such as the common loon (Gavia immer), green heron (Butorides virescens), and wood

VIII-6

duck (Aix sponsa) make use of the habitats found in tidal marshes (Wells et al. 2008). Just as the

shad are vulnerable to loss of habitat from damming and hardening of the river’s edges, the

marsh breeding birds and waterfowl are vulnerable to loss of habitat from changes in the

hydrology. Edge habitats are formed when different flood regimes along the elevation gradient

of a marsh support different plant communities. Channelizing the rivers disrupts these edge

habitats. The average water level may decrease, or flooding patterns may change as a result of

deepening the river’s bottom (Ralston et al. 2019). Marshes may also experience rapid accretion

to the influx of sediment in the river, which would elevate parts of it above tidal influence.

Marsh breeding birds are vulnerable to loss of habitat from invasive monocultures of Phragmites

australis that have been replacing native plant communities in the northeast United States

(Benoit and Askins 1999).

Wetlands have become recognized as a valuable carbon sink. Marshes, in particular,

have demonstrated one of the largest capacities for carbon storage (Mitsch et al. 2009). The high

primary productivity and the subsequent burial of organic matter allows a marsh to first take up large amounts of atmospheric CO2 and then store it, rather than releasing it back to the

atmosphere when the plant matter decomposes (Mcleod et al. 2011). Healthy marshes have an

impressive capacity to store carbon because their sediments accrete vertically as sea levels rise

(Chmura et al. 2003).

This matrix of plant detritus, trapped sediments, and dense vegetation slows runoff and acts as a filter to catch pollutants and sediments before they reach larger bodies of water, such as rivers or estuaries. As the water slows, fine sediments settle out and increased residence time

allows chemical reactions to take place, such as metal adsorption to clay particles (Berner and

Berner 2012). In the Scheldt estuary, marshes were found to remove 25%-50% of all metals

VIII-7

from runoff (Teuchies et al. 2013). Marshes are also effective at removing excess nitrogen and

phosphorus, nutrients associated with eutrophication (Khan and Brush 1994).

These important functions are being lost due to a global decline of freshwater marshes

(Kearney and Grace 1988; Strayer and Dudgeon 2010; Dahl and Stedman 2013; Murray et al.

2014). Freshwater tidal marshes fall within the tidal zone but above the salt wedge along the

river. The tidal zone on the Hudson River is artificially capped at Troy (250 km upstream from

NYC), where a dam prevents further tidal movement (Geyer and Chant 2006). Saltwater from

the Atlantic is carried in with the tide. While the position of the salt front can shift in years of

drought or with seawater rise, it typically does not exceed 130 km from the mouth of the river

(Peteet et al. 2011). Tidal marshes are often located on rivers with a high discharge but low

gradient, which allows for a high influx of sediment and nutrients but also a chance for

sediments to settle out of suspension and support marsh development (Mitsch et al. 2009). These

conditions are also favorable to human settlement, which benefits from the nutrient-rich soils

along the banks, the easy terrain, and abundant food. As such, freshwater tidal marshes are

especially vulnerable to degradation due to human activity, whether it be from rising sea levels,

damming rivers, hardening the shores, or the introducing invasive species (Kingsford et al.

2016).

Local sea levels are affected by three factors. Eustatic sea level rise is expected as the

polar ice caps and land ice melt and the ocean basins thermodynamically expand. Tectonic

rebound is occurring along the eastern coast of the United States, causing the northeast region of

the United States to lift and the mid-Atlantic region to sink. Finally, vertical movement of land

from local processes such as erosion or land-subsidence is changing the sea level in relation to

the land (Perry and Atkinson 2009). This results in variable sea level rise along the Atlantic

VIII-8

coast, from 2.11 mm/yr at Fernandina Beach, Florida to 5.92 mm/yr at Hampton Roads, VA.

Sea level change at Sandy Hook near the mouth of the Hudson River is estimated to be 4.05

+.21mm/year (NOAA Tides and Currents 2018). An increase in water volume within the

Hudson River Estuary can result in a river that is both more saline and deeper. Without a matching influx of freshwater from the source of the river, the salt wedge can move further north up the Hudson River, converting freshwater tidal marshes to salt marshes (Brinson et al. 1995).

In addition, the deeper water results in vegetation becoming more inundated and waves that can deliver more energy to the shoreline, perhaps resulting in erosion.

Wetlands are capable of responding to sea level rise through either the accretion of

inorganic sediments and plant and animal matter, which enables vertical growth, or through

migrating inland (Friedrichs and Perry 2001; Blum and Christian 2004). A response through

accretion has been observed at Sweet Hall Marsh in Chesapeake Bay, where an accretion rate of

1.0 g/cm2/yr exceeds the expected rate of sea level rise (Neubauer 2008). There has also been evidence of flexibility within these wetlands in their ability to shift from freshwater plant communities to oligohaline flora back to the original freshwater communities under changing salinity conditions (Odum et al. 1984; Davies 2004; Higinbotham et al. 2004; Bailey et al. 2006); however, salt intrusion has been accompanied by lower primary productivity and accretion rates, putting these marshes at risk of becoming inundated (Li and Pennings 2018; Wilson et al. 2018).

Rivers across the United States include more than 90,000 dams. The Hudson River itself

is impounded by the Federal Dam at Troy, opened in 1916 (ASCE 2017). Significant declines in

suspended sediment concentrations in rivers and streams have been highly correlated with the

construction of water-retaining dams (Weston 2014). Kirwan and Megonigal (2013) calculated

that dams and reservoirs retain 20% of the global sediment load, preventing it from reaching the

VIII-9

coast. This has placed marshes that have been able to adapt to sea-level rise in a new position of

vulnerability. A once-thriving region of marshes on the Yangtze River delta has reversed from expanding seaward since the seventh century to eroding inward after the construction of more than 50,000 dams (Kirwan and Megonigal 2013). There has been widespread loss in the United

States as well, with an estimated 25% of the wetlands in the Mississippi River delta being lost due to suspended sediment decline as caused by dams or relative sea level rise (Weston 2014).

Hardening of shorelines is the construction of structures parallel to the shore, such as

vertical seawalls, jetties, bulkheads, or railroads. Shoreline hardening is best predicted by

housing density and high GDP, which describes the Hudson River Estuary (Gittman et al. 2015).

Jamaica Bay marshes are disappearing due to hardening of shorelines, and increased nitrogen

loading from wastewater and stormwater discharge, as well as a decline of mineral sediment

(Peteet et al. 2018). Shoreline hardening reduces ecosystem services such as nutrient cycling,

carbon sequestration, and provision of habitat for native species. Bulkheads have been found to

reduce marsh width and decrease denitrification through marsh loss in North Carolina (O’Meara

et al. 2014). Biodiversity also decreases by 23% when seawalls are used rather than natural shorelines (Gittman et al. 2016). Channelizing rivers is another way to change the boundary between aquatic and terrestrial environments. Dredging, the construction of dykes, and the filling in of shores has decreased the total water area in the Hudson River by 30% when compared to the water area in 1910 and secondary channels have decreased by 70% (Collins and

Miller 2012). Similar channelization in northern Japan has resulted in increased water discharge and suspended sediment, which has resulted in transitions in vegetation from sedges to willow

trees (Nakamura et al. 1997).

VIII-10 Finally, wetlands have been globally affected by invasive plants that lower biodiversity, change nutrient cycling, and alter habitat structure (Zedler and Kercher 2004). Introduced species abundance in an ecosystem correlates positively with highly used dispersal routes for seeds (Tickner et al. 2001). The pulsing of water into the marsh from the river brings a high load of water-dispersed seeds, increasing the likelihood of exotic species. Typha x glauca, a hybrid between native Typha latifolia and introduced Typha angustifolia has spread through disturbed wetlands in the eastern United States. Frieswyk and Zedler (2007) found correlations between the expansion of Typha spp. and both water stabilization and urbanization, two processes that have occurred on the Hudson River. Water stabilization both decreases the amount of edge habitat created by regular pulsing and acts as a hydrological disturbance that may make the marsh inhospitable for habitat specialists that have adapted to specific hydrologic regimes

(Kercher and Zedler 2004). Urbanization can provide an influx of sediment, which can change the topography from heterogenous to homogenous (Figure 4). It also brings nutrients into the marsh, which favors Typha spp. (Zedler and Kercher 2004).

Figure 3: Influx of sediment reduces microtopographic variation and a sedge meadow is replaced by 1 to 2 opportunistic species (from Werner and Zedler 2002)

VIII-11

This study is part of a constellation of work that has been done in order to provide a more complete record and understanding of the Hudson River. There have been similar studies in the

past which has revealed important information about the Hudson River. Pederson et al. (2005)

identified periods of mega-droughts as well as European impact in the form of land clearance,

increased weedy plant cover, and an increase in inorganic particles in cores from Piermont

Marsh. Gill (2015) found clear anthropological effects in the form of exponential increases of

heavy metal concentrations in cores from Haverstraw Marsh; however, these studies were

performed in the brackish section of the Hudson River. Tivoli Bay is the only marsh in the

freshwater tidal section that has been examined similarly (Sritrairat et al. 2012). Tidal estuaries are already extremely dynamic, and conditions such as saltwater incursion, sediment accretion, or nutrient load can vary significantly between wetlands found at the mouth of the river as opposed to ones closer to the source.

The Site

Ramshorn-Livingston Marsh is located on the west bank of the Hudson River about 180 km from Battery Park at 40.741895 N, 73.989308 W. The marsh was recorded as early as 1770 by cartographer Henry Brace, Esq. (Brace 1770; Figure 4a). The same characteristic meandering

channel depicted by Brace can be seen today in satellite images. To the north of the marsh is

Catskill Creek, a larger tributary that feeds into the Hudson River (Figure 4b). Ramshorn-

Livingston is one of the largest freshwater tidal marshes on the Hudson River at 526 hectares

(NYSDOS 2012). While the area itself is protected as an Audubon wildlife refuge, much of the

surrounding land has been converted to agricultural or urban uses. Ramshorn- Livingston

acquired its protected status partly due to its high biodiversity. It hosts migratory ducks and

waterfowl in the autumn and winter, acts as a nursery to fish such as shad, bass, and eels in the

VIII-12

spring, and provides habitat for beavers (Castor canadensis) and river otters (Lontra canadensis)

(Scenic Hudson 2018).

a

N →

b

↑ N

Figure 4: a) Ramshorn-Livingston Marsh in 1770 (Brace 1770) and b) Satellite image of the coring site, showing the town of Catskill north of the marsh. Coring site marked by the red star and Catskill Creek drawn in blue.

VIII-13

Brace’s map clearly shows some of the first efforts to parcel the land for European

settlement. While the presence of European settlers predated this map by about a century, intensive development did not start until the beginning of the 1800s (McIntosh 1962). Industrial activities in this region included tanning with hemlock bark, glass production, ceramic factories, as well as widespread deforestation (McIntosh 1972). The early 19th century also brought about

vast hydrological changes to the Hudson River. New York State constructed dikes to block off

side channels and spur dikes to narrow the channel and increase velocities; these efforts were

expanded on by the United States Army Corps of Engineers which dredged and constructed larger longitudinal dikes along the Hudson River in the 1880s (Collins and Miller 2012). In

1916, the Federal Dam at Troy was opened. This system of locks and dams artificially maintained a certain depth on the river for water vessels and curtailed the tidal portion of the river north of the dam (Castagna 2016). By 1932, Congress approved dredging the river to a depth of 9.7 meters, changing the river from shallow and braided to a deeper, channelized river

(Collins and Miller 2012).

Pollution increased with settlement and population density, peaking in the 20th century

(Clarke 2011). Boyle (1969) described the untreated sewage from Utica, the discharge of hair,

grease, blood, and acids from tannery towns, the mats of pulp wastes that bob on the water, and thick deposits of decomposing sludge that could be observed in 1969 as “a pistol pointed at the

heart of the Hudson.” While the Clean Air and Water Acts have largely addressed these issues, the river has experienced ongoing difficulties in removing PCBs (Field et al. 2016); however,

Ramshorn-Livingston is regarded to be widely unaffected by this portion of the Hudson River’s history (NYSDOS 2012).

VIII-14

In this study, the analysis performed on the sediment core can inform us of past, present

and future problems. It was hypothesized that the analysis would show a distinctive lead profile

shared with other Northeastern marshes, a change in plant communities as caused by European settlement, and an increase in inorganic sediment also caused by European settlement. It was also expected to show a recent decrease in inorganic sediment due to the dredging of the Hudson

River and changed flood patterns. This information is useful in developing conservation or restoration strategies for this important tidal marsh. In terms of climate change mitigation, the wetlands along the Hudson River have a huge capacity to act as carbon sinks if properly cared for and maintained (Chmura et al. 2003). Finally, it can assist in anticipating how regional marshes will react to further settlement in the future.

A baseline to measure ecosystem health can be difficult to establish without a clear record of the history of an area. A historical record can identify how a marsh responds to inputs of sediment, nutrients, or heavy metals. This can help identify why freshwater marshes outside of the Hudson River Valley or ones without an available history may be experiencing shifts in vegetation or marsh decline.

METHODS

Sediment Coring

Ramshorn-Livingston was first surveyed for peat depth using fiberglass rods. Plants at

the site were identified using dichotomous keys and field guides. A core was extracted using a

10-cm diameter Dachnowski corer at the location with the deepest peat (Figure 5). The core was wrapped with polyethylene and an outer layer of aluminum foil and stored in a plastic tube. They

were then stored horizontally at 40°F in the Lamont-Doherty Core Repository. They were split vertically, one half stored for archival purposes and the other half used for analysis. The core was

VIII-15 segmented into 25 sections, each consisting of 4 cm of sediment depth. The age of the core was estimated using the lead profile and average accretion rates for Hudson River Valley as determined in other studies.

Loss-on-ignition

Loss-on-ignition was conducted at Lamont-Doherty. One cubic centimeter of sediment was extracted from each of the 25 segments and placed in a porcelain crucible. These samples were weighed and then dried at 100°C for 24 hours. The dried sample was weighed and then

Figure 5: Dachnowski corer (USEPA 2001).

VIII-16

combusted in a muffle furnace at 450°C for two hours and weighed once more. Loss-on-ignition was calculated according to the equation:

(%) = × 100

𝐷𝐷𝐷𝐷𝐷𝐷 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ𝑡𝑡 − 𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶 𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ𝑡𝑡 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 − 𝑜𝑜𝑜𝑜 − 𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 Macrofossils 𝐷𝐷𝐷𝐷𝐷𝐷 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ℎ𝑡𝑡

Macrofossils were identified using the USDA PLANTS database as well as the Peteet lab’s reference collection. Sediment from each section was washed through 500 and 125 μm screens. The remaining organic material was examined through 60x magnification and seeds, leaf fragments, needles, and any symmetrical material were removed and identified, then stored in vials and refrigerated.

X-ray fluorescence (XRF)

XRF was conducted using the ITRAXXRF spectrometer on the intact and hydrated core.

The core was first smoothed using saran wrap and plastic spatulas to minimize interference and allow a continuous reading from the spectrometer. The standard soil mode protocol was used to analyze Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, As, Se, Br, Rb, Zr, Mo, Ag, Cd, Sn, Hg, Pb, Bi, and

U. LEAP soil mode was also used to analyze P, S, Cl, K, Ca Ti, Cr, Mn, Fe, I, and Ba to extract

relative elemental concentrations. These protocols were described by Kenna et al. (2011) on the top meter of the cores.

RESULTS

There were two points of transition in the core: one at 20 cm depth and another at 64 cm depth.

The point at 64 cm depth signaled an increase in inorganic sediment accumulation and the

VIII-17

presence of many species of sedges. The point at 20 cm depth signaled a decrease in inorganic sediment accumulation and the replacement of sedges with Typha angustifolia. The color also

changed at these points in the core (Figure 6). Lead and zinc peaked between 64 and 20 cm

depth; silica, titanium, and aluminum decreased sharply between 20 cm and 0 cm; copper,

calcium, bromine, and chlorine all increased or had peaks between 20 cm and 0 cm. Age of the

bottom of the core was estimated to be about 250 years old based upon the increase in erosion

with settlement.

2018 CE 0 cm

~1970 CE 20 cm

Depth Year

~1860 CE 64 cm

1770 CE? 100 cm

Figure 6: Core collected from Ramshorn-Livingston Marsh.

Organic/inorganic matter content

Percent organic matter ranged between 9% and 65% (Figure 7a). There was a total

average of 22% organic matter per section with a standard deviation of 14%. The section

between 52-100 cm had a moderate percentage of organic matter (average=22%, st dev=6%).

The section between 16-52 cm had a fairly steady low percentage of organic matter

VIII-18 (average=11%, st dev=5%) due to the increase in inorganic sediment. There was a steep increase

in percentage of organic matter from 16-20 cm (9%) to the top of the core (65%).

Organic matter in weight ranged from 0.012 g to 0.036 g with an average of 0.025 g and standard deviation of 0.007 g (Figure 7b). Inorganic matter in weight ranged from 0.035 g to

0.331 g with an average of 0.144 g and a standard deviation of 0.073 g. There was an increase in inorganic material from 100 cm to 16 cm and then a sharp decrease from 0.331 g at 16-20 cm to an average of 0.046 g from 0-16 cm. Organic matter by weight did not dramatically increase nor decrease throughout the core.

Figure 7: Inorganic and organic content of the core by depth. Calculated for every 4 cm of the core. A) shows organic content by percent, B) shows weights of organic and inorganic content.

VIII-19

X-ray Fluorescence

Silica, titanium, and potassium all shared a similar depth profile (Figure 8). There was a

gradual increase in concentration until 23 cm and then a rapid decline to concentrations much

lower than seen previously. Silica and titanium had a high correlation with an R2 value of 0.97.

A multiple regression between silica, titanium, potassium, iron, and aluminum has an R2 value of

0.99.

Figure 8: Normalized relative x-ray fluorescence showing abundance of silica, titanium, and potassium as a function of depth in the sample core.

The amount of lead started to increase rapidly at 48 cm until it peaks at 17 cm (Figure 9).

There was a very rapid decline from 17 cm to 14 cm, after which the amount of lead is less than amounts observed between 50-100 cm. Zinc also declined following 17 cm with a peak at 6 cm, but the decline was more gradual than lead. Copper started increasing at 30 cm and peaked at 4 cm before declining to present. The amount of calcium found in the core was fairly consistent from 20-100 cm (Figure 10). There was a sharp peak at 19 cm, followed by two smaller ones located at 8 cm and 2 cm. Chlorine had a similar depth profile to calcium with peaks around 8 cm and 2 cm. Bromine increased in a similar manner to calcium after 20 cm but it was harder to discern any significant peaks.

VIII-20

Figure 9: Normalized relative x-ray fluorescence showing abundance of lead, zinc, and copper as a function of depth in the sample core.

Figure 10: Normalized relative x-ray fluorescence showing abundance of calcium, chlorine, and bromine as a function of depth in the sample core.

VIII-21

Macrofossils

The lowest section of the core (Zone 1, 100 - 68 cm) contained seeds of trees and shrubs

such as white pine (Pinus strobus), alder (Alnus) sp., ironwood (Carpinus caroliniana), hazel

(Corylus americana) and silky dogwood (Cornus ammomum), as well as herbs such as smartweed (Polygonum) sp., Asteraceae, violet (Viola) sp., buttercup (Ranunculus) sp., false nettle (Boehmeria) sp., sedge (Carex) spp. lenticular, sedge (Carex) sp. “shield”, sawgrass

(Cladium) sp., and arrowhead (Sagittaria latifolia )(Figure 11). Alnus sp. bracts, fungal sclerotia, and fungal hyphae were also found in this section. In Zone 2, from 68 cm to 20 cm, the peat included seeds of birch (Betula) sp., Polygonum sp., Asteraceae, Viola sp., Carex spp.

lenticular, Carex sp. “shield”, Carex spp. trigonous, three-way sedge (Dulichium) sp., bulrush

(Scirpus) sp., cattail (Typha angustifolia), sweet flag (Acorus) sp., fungal sclerotia, and fungal

hyphae.

Figure 11: Macrofossil profile of Ramshorn-Livingston Marsh.

Several unknown teardrop shaped seeds, fern leaflets, insects, and charcoal were also found. Zone 3, from 0-20 cm contained St. John’s wort (Hypericum perforatum), Asteraceae,

Carex spp. lenticular, Typha angustifolia, Sagittaria latifolia and the unknown “mouse” seeds, along with fern leaflets, insects, charcoal, and shells. Fungal sclerotia and Carex spp. were the most abundant macrofossils in Zone 1,. While lenticular and Carex spp. seeds were the most abundant between 20-68 in Zone 2, and Typha angustifolia seeds were the most abundant between 0-20 in Zone 3.

DISCUSSION

Three zones became apparent when comparing results from LOI, XRF, and macrofossils, with boundaries occurring at 20 cm and 68 cm. The establishment of these zones are supported by the change of the organic:inorganic ratio at 20 cm, the change in the XRF profiles of silica, titanium, potassium, lead at 20 cm, and the change of the macrofossil profiles at 68 cm and 20 cm. The macrofossil profile switched from heavy, large seeds to sedges at 68 cm and sedges to cattail (Typha angustifolia) at 20 cm.

Zone 1 – 100-68 cm

Zone I had the largest macrofossils. Silky dogwood (Cornus ammomum) and hazel

(Corylus americana) seeds are both over 3 cm long, with heavy, thick walls. The transport of these seeds to an area away from the tree line may indicate an active flooding cycle. The lack of seeds of the same size or weight later in the core may indicate a change in hydrology. This section of the core is also unique in that it has a high concentration of fungal sclerotia and hyphae, which correlates with the presence of buried wood and plants that grow in edge habitats

between terrestrial and aquatic environments (Miola et al. 2010). There are no abrupt increases of Asteraceae or violet (Viola) sp. seeds, which would indicate disturbance. VIII-23

Zone 2 – 68-20 cm

A diverse group of sedges was found in Zone 2, which are the bulk of the seeds identified. Asteraceae has a strong presence between 20 and 40 centimeters. As Asteraceae fossils are associated with early successional habitats, this sudden increase in seed count could be caused by disturbance in the area (Mitsch et al. 2009). Cattail (Typha angustifolia) seeds first appear in this section, but as a very minor component compared to the sedges. Lead starts to increase around 50 cm and peaks at 20 cm. This is a classic profile observed in many cores collected from the Hudson River Estuary’s marshes that can be used to establish human activity timelines (Chillrud et al. 2004; Sritrairat 2013; Peteet et al. 2018). Lead sources range from pigment plants, capacitor plants, extraction of lead ore from mines, and leaded gasoline (Chillrud et al. 2004). The decline of these industries, passage of the Clean Air Act, and the removal of lead from gasoline all acted to return lead levels to pre-industrial levels.

Zone 3 – 20-0 cm

Zone 3 has a dramatic switch from a sedge-dominated plant community to a near- monoculture of cattail (Typha angustifolia), an invasive species. Silica, titanium, and potassium all decrease markedly during this period. These elements are all associated with inorganic

sediments, and their decline is mirrored by the sharp decrease in inorganic material in the core.

The decrease in silica may be linked or enhanced by the switch from a sedge community to

Typha angustifolia. Cladium sp. uses silica for plant structure (Lorenzen et al. 2001). Typha sp.

does not utilize silica to the same degree: four times as much silica is stored in Cladium sp. as it

is in Typha sp. The sudden drop of available silica in the system from the lack of incoming

inorganic sediment may have weakened the existing population of sedges. In addition,

Ramshorn Marsh likely receives excess nitrogen and phosphorus from fertilizers in runoff. VIII-24

Typha sp. is also able to utilize nitrogen and phosphorus to increase aboveground biomass whereas sedges cannot (Woo and Zedler 2002). This gives Typha sp. an additional competitive advantage over the sedges (Woo and Zedler 2002). Finally, changes in hydrology have also been shown to cause shifts in plant communities. A more stable water regime encourages Typha sp. dominance whereas disturbance favors sedges (Boers and Zedler 2008).

Soil accretion

If it is assumed that the base of the core is about 1770 CE, then the average rate of sedimentation in this marsh would be 0.4 cm/yr (100 cm/250 years); however, there is no basal date, and the core could be older, resulting in a lower rate. Additionally, accumulation probably varies with decade, as seen in other cores. Using a solid chronology with pollen stratigraphy,

AMS dates, an XRF timelines for pollution, and stable isotope data, the decadal sedimentation rate will be established. Rates of sediment accretion in tidal freshwater marshes have been found to range from 0.11 to 2.19 cm/yr with an overall median rate of 0.76 cm/yr (Neubauer 2008), and recent rates in comparative tidal saltwater marsh Jamaica Bay are up to 0.4 cm/yr (Peteet et al.

2018). Accretion rates can be greatly affected by anthropogenic activity; a freshwater tidal marsh in Maryland showed an increase in accretion rate from 0.05 cm/yr to 0.50 cm/yr after

European settlement (Khan and Brush 1994). Khan and Brush connected their increase in sedimentation rates to soil erosion that was a result of settlement and clearing the land for farming. This study paralleled those findings with the increase of inorganic minerals and a greater content of inorganic soil entering the marsh in Zone 2. This would correspond with the late-1800s to early-1900s, a time of increased settlement in the Hudson River Valley.

The decrease of inorganic matter found in Zone 1 of the core may indicate slower future accretion rates. Organic matter is responsible for 62% of freshwater tidal marsh accretion and VIII-25

inorganic matter contributes about 38% along the east coast of the United States (Neubauer

2008). While the total organic carbon accumulation is four times higher in high-marsh sediments, such as the coring site, inorganic sediment still contributes a significant role in marsh accretion (Khan and Brush 1994).

There is an increase in K and Ti in freshwater and brackish marshes along the Hudson

River, which is associated with an increase in inorganic sediment with settlement (Pederson et al.

2005; Sritrairat et al. 2012) and continued disturbance; however, a study conducted in

Chesapeake Bay very closely parallels the findings at Ramshorn-Livingston Marsh (Figure 12).

Figure 12: A comparison between the change in inorganic weight per sample in the Ramshorn-Livingston core to the sedimentation rate at Chesapeake Bay (Hilgartner and Brush 2006).

The freshwater tidal marsh in Chesapeake Bay has a similar history; there was low disturbance before 1658 and accelerated European settlement in the 1700s. Hilgartner and Brush (2006) only VIII-26

found habitat shifts post-European settlement; the marsh remained stable over a period of 1500 years before settlement, indicating that habitat shifts were highly likely to be linked with human settlement. High sedimentation rates were followed by a vegetation shift from plants associated

with low marsh to plants associated with high shrub marsh. An increase in elevation is also

supported by the finding that organic content tends to increase and inorganic content decreases

with increasing elevation (Palinkas and Engelhardt 2015).

Heavy metals

Constructed wetlands have been used for pollution remediation because wetlands are so

effective at removing heavy metals from the water column. They are either removed through adsorption with solid particles such as clay or the biological uptake and storage in plants such as

Typha sp. (Dunbabin and Bowmer 1992). Marshes of similar age to Ramshorn Livingston are

better at capturing the anthropogenic signal of lead because of their lower background level

(Sritrairat 2013). While lead and zinc have both decreased as a result of recent legislation, copper continues to rise almost to the present. The XRF profiles only capture relative change and do not indicate quantifiable amounts of different elements, so it is not possible to tell where

the concentration of copper is relative to toxicity levels. Copper can be toxic at certain levels to some plants and especially some aquatic organisms (Brooks et al. 2007; Han et al. 2012; Sáez et al. 2013).

Calcium sources

Calcium showed several peaks between 0-20cm in the core, which might normally be associated with salt intrusion. Saltwater can be supersaturated with calcium carbonate and an intrusion would carry calcium, chlorine, and bromine into the marsh, all of which are observed in the upper 20 cm (Morse et al. 1978); however, saltwater as north as Poughkeepsie is a VIII-27

newsworthy event (Greer 1985). Ramshorn-Livingston is another 65 km north of Poughkeepsie

and it is highly unlikely saltwater would have reached the marsh unnoticed.

The profile of calcium and chlorine appear to be very similar (Figure 13). Upon future

investigation, the town of Catskill stored its road salt on a spit of land north of the marsh in the

open air (corr. with Jonathan Palmer, Vedder Research Institute). A common formulation for

road salt is CaCl2. Large rain events could wash the road salt into the marsh, creating these peaks. Although bromine has also increased in this period and is associated with saltwater, the signal is noisy and does not have a clear correlation with chlorine. Given the close relationship between calcium and chlorine, and the noisier signal of bromine, it appears more likely that the increase in calcium is caused by road salt rather than saltwater.

Figure 13: Relationship between calcium and chlorine.

VIII-28

CONCLUSIONS

Ramshorn-Livingston is regarded as a healthy and intact freshwater tidal marsh. As one peels back the years, a more complete picture of how the marsh has changed is provided. By using average accretion rates from cores analyzed in other Hudson River studies and dating the core based on the lead profile, it is estimated that the 1-m core spans about 250 years, but a firm chronology awaits further research. The organic matter profile declines from 30% organic at the base of the core to 10% as human disturbance resulted in increased silt and clay reaching the site.

This shift parallels the shift from more arboreal seeds to dominance of sedges and Asteraceae, along with heavy metal increases such as lead, copper, and zinc which reveal the industrial effects up into the 20th century. Most recent sediments in the upper part of the core indicate a decline in the heavy metals, and a switch from native sedges to invasive cattail.

It is natural for ecosystems to change, flux, or give way to a new order. Oligotrophic lakes fill with sediment, forming shallower, nutrient-rich bodies of water. Forests build up over land previously developed and tree roots make even concrete crumble. The slow rate at which these transformations occur set them apart from what has happened at Ramshorn-Livingston. As

Hilgartner and Brush (2006) found at Chesapeake Bay, these freshwater tidal marshes can exist in stasis for centuries. It is the introduction of human settlement that can provoke such a rapid change. A marsh’s status as an edge habitat between the terrestrial and aquatic means it receives feedbacks from both the land and the water. From land, human settlers are responsible for an increased nutrient load in the marsh and increased sediment from clearing land and moving earth. From the water, there is increased pollution from effluent and decreased sediment loads from dams. The hydrology of the river is changed by hardening its shorelines or breaking up ice floes in the winter. VIII-29

RECOMMENDATIONS

Pollen analysis can be conducted on the archived core from Ramshorn-Livingston, as

well as C13 and N15 isotope analysis. The pollen analysis will provide more regional details of

the vegetation present through time and the isotope analysis may provide a broader

understanding of any shifts from woody to herbaceous plant communities at the site, as well as

the nitrogen pollution history. Carbon dating can also provide an approximate bottom date of the

core. A changed hydrological regime is one potential explanation for several of the results, such as the heavy seeds transported into the marsh in Zone 1, the switch of plant composition from sedges to cattail, and the decrease in inorganic matter entering the marsh. Data on flood duration, frequency, and depth at Ramshorn-Livingston would be helpful in assessing this hypothesis. Deeper cores from the same site can be obtained, providing an older record of

Ramshorn-Livingston Sanctuary.

The core taken at Ramshorn-Livingston marsh is one of many taken in the Hudson River

Valley; however, this collection has a greater emphasis on brackish and salt marshes. More cores should be taken in the freshwater tidal section of the Hudson River. Sites close to the

Federal Dam at Troy are of particular interest because of the change in sediment and water dynamics. It would be informative to compare the loss of inorganic matter observed in the top

20 cm of the Ramshorn-Livingston core to others in the freshwater tidal portion of the Hudson

River. Marshes currently dominated by Typha sp. also warrant investigation to see if there was a

switch from previous vegetation to Typha.

The effects of human settlement in Ramshorn-Livingston were largely the result of

regional processes that did not originate in or near the marsh. Some policy changes would be

simple to implement, such as storing road ice inside a facility rather than in the open air, while VIII-30

others are not site-specific and apply to all wetlands along the Hudson River. In order to benefit

from the important and commercially viable ecosystem services that marshes such as Ramshorn-

Livingston offer, dredging and channelizing the river must continue to be closely monitored by local, state, and federal organizations.

In terms of site-specific problems, a monoculture of Typha angustifolia decreases the level of biodiversity, which affects accretion rates, soil chemistry, and the ecosystem food web.

An increase in nutrients from human settlement and a less frequent flood regime from rapid soil accretion both give T. angustifolia a competitive advantage over the sedges. If it was thought to

be beneficial to have a more diverse plant community, changes on a very large scale would have

to occur. The use of fertilizer and disposal of wastewater with phosphorus and nitrogen in the

watershed that feeds into Ramshorn Marsh would have to be minimized.

ACKNOWLEDGEMENTS

I would like to thank the Hudson River Foundation and the Ecology, Evolution, and

Environmental Biology department or their financial assistance in making this research possible.

I would also like to thank my mentor, Dr. Dorothy Peteet, for her invaluable help and support,

and for giving me an entirely new perspective on the world around me. I am grateful to Dr.

Hugh Ducklow for his generous and kind help guiding me through the writing process, and Dr.

Matthew Palmer for his helpful critiques. I am indebted to Clara Chang at Lamont-Doherty

Earth Observatory and Jonathan Palmer of the Vedder Research Institute, who provided valuable expertise. Clara amazed me with the amount of information she could extract from a simple peat core and Jonathan Palmer not only provided invaluable information on my site, but also

VIII-31

reinvigorated my investigation into Ramshorn-Livingston with his own love for the Hudson

River. I would also like to thank Dr. James Thompson for years of patience, faith, and

nutritional support. Finally, I am grateful to both my City College of San Francisco and

Columbia cohort for inspiring me through their own hard work and pursuit of excellence.

VIII-32

REFERENCES

American Society of Civil Engineers (ASCE). 2017. Dams Infrastructure Report Card, ASCE Foundation.

Bailey D.E., J.E. Perry and D.A. DeBerry. 2006. Aeschynomene virginica habitat in a tidal freshwater marsh, James City County, Virginia, USA. Banisteria 27:3-9.

Benoit, L.K. and R.A. Askins. 1999. Impact of the spread of Phragmites on the distribution of birds in Connecticut tidal marshes. Wetlands. 19:194-208.

Berner, E.K., and R.A. Berner. 2012. Global Environment: Water, Air and Geochemical Cycles. Princeton University, Oxfordshire and Princeton.

Blum, L.K., and R.R. Christian. 2004. Belowground production and decomposition along a tidal gradient in a Virginia salt marsh. Pages 44-47 in Ecogeomorphology of tidal marshes. S. Faghrazzi, M. Marani, and L.K. Blum, editors. Coastal and Estuarine Studies. American Geophysical Union, Washington, DC.

Boers, A.M., and J.B. Zedler. 2008. Stabilized Water Levels and Typha Invasiveness. Wetlands 28:3:676-685.

Boyle, R.H. 1969. Hudson River; a Natural and Unnatural History. Norton, New York.

Brace, H. 1770. The Loveridge patent. Provided by the Vedder Research Institute, 2017.

Brinson, M.M., R.R. Christian, and L.K. Blum. 1995. Multiple states in the sea level induced transition from terrestrial forest to estuary. Estuaries 18:648-659.

Brooks, M.L., J.S. Meyer, and D.M. McKnight. 2007. Photooxidation of wetland and riverine dissolved organic matter: altered copper complexation and organic composition. Hydrobiologia 579:95-113.

Castagna, J. 2016. Troy Lock and Dam NY reaches 100 year milestone. US Army Corps of Engineers New York District. https://www.nan.usace.army.mil/Media/News- Stories/Story-Article-View/Article/905916/troy-lock-dam-ny-reaches-100-year- milestone/ (accessed 2/19/19).

Chillrud, S.N., R.F. Bopp, J.M. Ross, D.A. Chaky, S. Hemming, E.L. Shuster, H.J. Simpson, and F. Estabrooks. 2004. Radiogenic lead isotopes and time stratigraphy in the Hudson River, New York. Water, Air & Soil Pollution 2-3:469–82.

Chmura G., S. Anisfeld, D. Cahoon, and J. Lynch. 2003. Global carbon sequestration in tidal, saline wetland soils. Global Biogeochemical Cycles 17:4. VIII-33

Clarke, D. 2011. Hudson River.870-872 in Environmental Encyclopedia. M. Bortman, P. Brimblecombe, M.A. Cunningham, W.P. Cunningham, and W. Freedman. Gale, Farmington Hills.

Collins, M.J., and D. Miller. 2012. Upper Hudson River estuary (USA) floodplain change over the 20th century. River Research and Applications 8:1246–1253.

Dahl, T.E., and S.M. Stedman. 2013. Status and trends of wetlands in the coastal watersheds of the conterminous United States 2004 to 2009. US Department of the Interior, Fish and Wildlife Service & National Oceanic and Atmospheric Administration, National Marine Fisheries Service.

Davies, S.B. 2004. Vegetation dynamics of a tidal freshwater marsh: long-term and inter-annual variability and their relation to salinity. Master’s Thesis. College of William and Mary, Virginia Institute of Marine Science. Gloucester Point, VA.

Dunbabin, J.S., and K.H. Bowmer. 1992. Potential use of constructed wetlands for treatment of industrial wastewaters containing metals. Science of The Total Environment 2-3:151– 168.

Edinger, G.J., D.J. Evans, S. Gebauer, T.G. Howard, D.M. Hunt, and A.M. Olivero. 2014. Ecological Communities of New York State. Natural Heritage Program, New York State Department of Environmental Conservation. Albany. Second Edition. A revised and expanded edition of Carol Reschke’s Ecological Communities of New York State.

Field, L.J., J.W. Kern, and L.B. Rosman. 2016. Re-visiting projections of PCBs in lower Hudson River fish using model emulation. Science of The Total Environment 557-558:489–501.

Friedlander, B. 2016. As sea level rises, Hudson River wetlands may expand. Cornell Chronicle. https://phys.org/news/2016-06-sea-hudson-river-wetlands.html (accessed 2/19/19),

Friedrichs, C.T., and J.E. Perry. 2001. Tidal salt marsh morphodynamics: a synthesis. Journal of Coastal Research 27:7-37.

Frieswyk, C., and J. Zedler. 2007. Vegetation change in Great Lakes coastal wetlands: deviation from the historical cycle. Journal of Great Lakes Research 33:336-380.

Geyer, W.R., and R. Chant. 2006. The physical oceanography processes in the Hudson River estuary. Pages 24-38 in J.S. Levinton and J.R. Waldman, editors. The Hudson River Estuary.. Cambridge University Press, Cambridge.

Gill, L. 2015. A soil story: reconstructing the paleoecological and anthropological history of Haverstraw tidal marshlands. Final report to Columbia University, New York, New York.

VIII-34

Gittman, R.K., S.B. Scyphers, C.S. Smith, I.P. Neylan, and J.H. Grabowski. 2016. Ecological consequences of shoreline hardening: a meta-analysis. BioScience 9:763-773.

Gittman, R.K., F.J. Fodrie, A.M. Popowich, and D.A. Keller. 2015. Engineering away our natural defenses: an analysis of shoreline hardening in the US. Frontiers in Ecology and the Environment 13:301-307.

Greer, W. 1985. Salt in Hudson imperils Poughkeepsie water. The New York Times, New York.

Han, R., I. Lefevre, C. Ruan, and N. Beukalaers. 2012. Effects of salinity on the response of the wetland halophyte Kosteletzkya virginica (L.) Presl. to copper toxicity. Water, Air, & Soil Pollution 3:1137–1150.

Hilgartner, W.B., and G.S. Brush. 2006. Prehistoric habitat stability and post-settlement habitat change in a Chesapeake Bay freshwater tidal wetland, USA. The Holocene 4:479–494.

Higinbotham, C.B., M. Alber, and A.G. Chalmers. 2004. Analysis of tidal marsh vegetation patterns in two Georgia estuaries using aerial photography and GIS. Estuaries 27:670- 683.

Kearney M.S., and R.E. Grace. 1988. Marsh loss in Nanticoke Estuary, Chesapeake Bay. Geographical Review 78:206-220.

Kenna, T.C., F. Nitsche, M. Herron, and B. Mailloux. 2011. Evaluation and calibration of a field portable x-ray fluorescence spectrometer for quantitative analysis of siliciclastic soils and sediments. Journal of Analytical Atomic Spectrometry 2:395-405.

Kercher, S.M., and J.B. Zedler. 2004. Flood tolerance in wetland angiosperms: a comparison of invasive and noninvasive species. Aquatic Botany 2:89-102.

Khan, H., and G.S. Brush. 1994. Nutrient and metal accumulation in a freshwater tidal marsh. Estuaries 2:345-360.

Kingsford, R.T., A. Basset, and L. Jackson. 2016. Wetlands: conservation’s poor cousin. Aquatic Conservation: Marine and Freshwater Ecosystems 5:892-916.

Kirwan, M.L., and J.P. Megonigal. 2013. Tidal wetland stability in the face of human impacts and sea-level rise. Nature 504:53-60.

Li, F., and S.C. Pennings. 2018. Responses of tidal freshwater and brackish marsh macrophytes to pulses of saline water simulating sea level rise and reduced discharge. Wetlands 5:885-891.

VIII-35

Lorenzen, B., H. Brix, I.A. Mendelssohn, K.L. McKee, S.L. Miao. 2001. Growth, biomass allocation, and nutrient use efficiency in Cladium jamaicense and Typha domingensis as affected by phosphorus and oxygen availability. Aquatic Botany 70:117-133.

McIntosh, R.P. 1962. The forest cover of the Catskill mountain region, New York, as indicated by land survey records. American Midland Naturalist 2:409-423.

McIntosh, R.P.F. 1972. Forests of the Catskill Mountains, New York. Ecological Monographs 2: 143-161.

Mcleod, E., G.L. Chmura, S. Bouillon, R. Salm, M. Bjork, C.M. Duarte, K.E. Lovelock, W.H. Schlesinger, and B.R. Sillman. 2011. A blueprint for blue carbon: toward an improved understanding of the role of vegetated coastal habitats in sequestering CO2. Frontiers in Ecology and the Environment 10:552-560.

Miola, A., S. Favaretto, I. Sostizzo, G. Valentini, and A. Asioli. 2010. Holocene salt marsh plant communities in the north Adriatic coastal plain (Italy) as reflected by pollen, non-pollen palynomorphs and plant macrofossil analyses. Vegetation History and Archaeobotany 5:513–529.

Mitsch, W.J., J.G. Gosselink, L. Zhang, and C.J. Anderson. 2009. Wetland Ecosystems. John Wiley & Sons, Hoboken, NJ.

Morse, J.W., J. Kanel, and H.L. Craig Jr. 1978. Literature review of the saturation state of seawater with respect to calcium carbonate and its possible significance for scale formation on OTEC heat exchangers. Ocean Engineering 3:297-315.

Murray, N.J., R.S. Clemens, S.R. Phinn, H.P. Possingham, and R.A. Fuller. 2014. Tracking the rapid loss of tidal wetlands in the Yellow Sea. Frontiers in Ecology and the Environment 12:267-272.

Nack, C.C., K.E. Limburg, and D. Miller. 2014. Assessing the quality of four inshore habitats used by post-yolk-sac Alosa sapidissima (Wilson 1811) in the Hudson River: A Prelude to Restoration. Restoration Ecology 23:57–64.

Nakamura, F., T. Sudo, S. Kameyama, and M. Jitsu. 1997. Influences of channelization on discharge of suspended sediment and wetland vegetation in Kushiro Marsh, Northern Japan. Geomorphology 3-4:279-289.

National Oceanic and Atmospheric Administration (NOAA). 2013. Tidal Freshwater Marshes: Underappreciated Part of the Estuary. NOAA Fisheries.

National Oceanic and Atmospheric Administration (NOAA). 2018. Tides and Currents: https://tidesandcurrents.noaa.gov/

VIII-36

Neubauer, S.C. 2008. Contributions of mineral and organic components to tidal freshwater marsh accretion. Estuarine, Coastal and Shelf Science 1:78–88.

New York State Department of State (NYSDOS). 2012. Ramshorn Marsh. Coastal Fish and Wildlife Rating Form, NYS Department of State - Natural Resources Bureau. https://www.dos.ny.gov/opd/programs/consistency/Habitats/HudsonRiver/Ramshorn_Ma rsh_FINAL.pdf (accessed 2/19/19).

Odum, W., T. Smith III, J.K. Hoover, and C.C. McIvor. 1984. The ecology of tidal freshwater marshes of the United States east coast: a community profile. U.S. Fish and Wildlife Service, Washington, DC.

O’Meara, T., S.P. Thompson, and M.F. Piehler. 2014. Effects of shoreline hardening on nitrogen processing in estuarine marshes of the U.S. mid-Atlantic coast. Wetlands Ecology and Management 23:385-394.

Palinkas, C.M., and K.A.M. Engelhardt. 2015. Spatial and temporal patterns of modern (~100 yr) sedimentation in a tidal freshwater marsh: implications for future sustainability.” Limnology and Oceanography 1:132–148.

Pederson, D.C., D.M. Peteet, D. Kurdyla, and T. Guilderson. 2005. Medieval warming, little ice age, and European impact on the environment during the last millennium in the lower Hudson Valley, New York, USA. Quaternary Research 3:238–249.

Perry, J.E., and R.B. Atkinson. 2009. York River tidal marshes. Journal of Coastal Research 57:40- 49.

Peteet, D.M., E.Markgraf, D.C. Pederson, and S. Sritrairat. 2011. Linking uplands to the Hudson River: Lake to marsh records of climate change and human impact over millennia. 123- 134 in Environmental History of the Hudson River: Human Uses That Changed the Ecology, Ecology that Changed Human Uses. Robert Henshaw. SUNY Press, Albany.

Peteet, D.M., J. Nichols, T. Kenna, C. Chang, J. Browne, M. Reza, S. Kovari, L. Liberman, and S. Stern-Protz. 2018. Sediment starvation destroys New York City’s marshes’ resistance to sea level rise. Proceedings of the National Academy of Sciences 115: 10281-10286.

Ralston, D.K., S. Talke, W.R. Geyer, H.A.M. Al-Zubaidi, and C.K. Sommerfield. 2019. Bigger tides, less flooding: Effects of dredging on barotropic dynamics in a highly modified estuary. Journal of Geophysical Research:Oceans 124:196-211

Sáez, M.I., J. Casas, J.L. Guil-Gerrero, M.D. Gil, R. Canero, and M.D. Suarez. 2013. Effects of organic matter, alkalinity and pH on acute copper toxicity to mosquitofish, Gambusia holbrooki: implications for a multipurpose management of irrigation ponds.” International Review of Hydrobiology 98:26-270.

VIII-37

Scenic Hudson. 2018. Ramshorn-Livingston Sanctuary. Proposed Hudson River Anchorages.

Sritrairat, S. 2013. Multiproxy analyses of past vegetation, climate, and sediment dynamics in Hudson River wetlands. Doctoral dissertation. Columbia University, New York, New York.

Sritrairat, S., D.M. Peteet, T.C. Kenna, , R.N. Sambrotto, R., K., D. Kurdyla, and T. Guilderson, T. 2012. A history of vegetation, biodiversity, sediment and nutrient dynamics at Tivoli North Bay, Hudson River, NY. Estuarine, Coastal, and Shelf Science 102-103:24-35.

Strayer, D.L., and D. Dudgeon. 2010. Freshwater biodiversity conservation: recent progress and future challenges. Journal of the North American Benthological Society 29: 344-358.

Teuchies, J., R. Carpentier, L. Bervoets, S. Temmerman, C. Wang, T. Maris, T.J.S. Cox, A. Van Braeckel, and P. Meire. 2013. Estuaries as filters: the role of tidal marshes in trace metal removal. PLoS ONE 8: 1–11.

Tickner, D.P., P.G. Angold, A.M. Gurnell, and J.O. Mountford. 2001. Riparian plant invasions: hydrogeomorphological control and ecological impacts. Progress in Physical Geography 1:22–52.

Tiner, R.W. 2013. Tidal Wetlands Primer: An Introduction to their Ecology, Natural History, Status, and Conservation. University of Massachusetts Press, Amherst.

United States Environmental Protection Agency (USEPA). 2001. Methods for collection, storage, and manipulation of sediments for chemical and toxicological analyses: technical manual. Office of Water EPA 823: B-01

Wells, A.W., W.C. Nieder, B.L. Swift, K.A. O’Connor, and C.A. Weiss. 2008. Temporal changes in the breeding bird community at four Hudson River tidal marshes. Journal of Coastal Research 55:221–235.

Weston, N.B. 2014. Declining sediments and rising seas: an unfortunate convergence for tidal wetlands. Estuaries and Coasts 37:1-23

Werner, K.J., and J. B. Zedler. 2002. How sedge meadow soils, microtopography, and vegetation respond to sedimentation. Wetlands 3:451-466.

Wilson, B., S. Servais, S. Charles, S.E. Davis, E. Gaiser, J. S. Kominoski, J. H. Richards, and T.G. Troxler. 2018. Declines in plant productivity drive carbon loss from brackish coastal wetland mesocosms exposed to saltwater intrusion. Estuaries and Coasts 41: 2147-2158.

VIII-38

Woo, I., and J.B. Zedler. 2002. Can nutrients alone shift a sedge meadow towards dominance by the invasive Typha x glauca. Wetlands 22:509-521.

Woods, L. 2012. The floundering fish: Hudson River shad. 2012. Cary Institute of Ecosystem Studies.

Zedler, J.B., and S. Kercher. 2004. Causes and consequences of invasive plants in wetlands: opportunities, opportunists, and outcomes. Critical Reviews in Plant Sciences 23:431- 452.

VIII-39

APPENDIX

Table A: Species from the Ramshorn-Livingston core and their common names

Species Common Name Pinus strobus Eastern white pine Alnus sp. Alder Betula sp. Birch Carpinus virginiana Ironwood Corylus americana Hazelnut Cornus ammomum Silky dogwood Polygonum sp. Knotweed Hypericum perforatum St. John’s wort Viola sp. Violet Ranunculus sp. Ranunculus Boehmeria sp. False nettle Carex sp. True sedge Dulichium sp. Threeway sedge Cladium sp. Fen-sedge Scirpus sp. Bulrush Typha latifolia Broadleaf cattail Typha angustifolia Narrowleaf cattail Sagittaria latifolia Broadleaf arrowhead Acorus sp. Calamus

VIII-40 2018 Polgar Fellows

From left to right: Jason Randall, Carrie E. Perkins, Michelle L. Zeliph, Kaili M. Gregory, Erika Bernal, Virginia Caponera, Elizabeth Thompson, Corey W. Rundquist

Special thanks to Sam Gordon for assistance in formatting manuscripts.