An Assessment Using Stable Isotopes

Estuaries and Coasts: J CERF (2008) 31:53–69 DOI 10.1007/s12237-007-9029-0

Distribution and Trophic Importance of Anthropogenic Nitrogen in Narragansett Bay: An Assessment Using Stable Isotopes Autumn Oczkowski & Scott Nixon & Kelly Henry & Peter DiMilla & Michael Pilson & Stephen Granger & Betty Buckley & Carol Thornber & Richard McKinney & Joaquin Chaves

Received: 6 June 2007 / Revised: 30 August 2007 / Accepted: 7 September 2007 / Published online: 11 January 2008 # Coastal and Estuarine Research Federation 2007

Abstract Narragansett Bay has been heavily influenced by human activities for more than 200 years. In recent decades, it has been one of the more intensively fertilized estuaries in the USA, with most of the anthropogenic nutrient load originating from sewage treatment plants (STP). This will soon change as tertiary treatment upgrades reduce nitrogen (N) loads by about one third or more during the summer. Before these reductions take place, we sought to characterize the sewage N signature in primary (macroalgae) and secondary (the hard clam, Mercenaria mercenaria) producers in the bay using stable isotopes of N (δ15N) and carbon (δ13C). The δ15N signatures of the macroalgae show a clear gradient of approximately 4‰ from north to south, A. Oczkowski (*) : S. Nixon : K. Henry : P. DiMilla : M. Pilson : S. Granger : B. Buckley Graduate School of Oceanography, The University of Rhode Island, South Ferry Road, Narragansett 02882 RI, USA e-mail: [email protected] C. Thornber Department of Biological Sciences, The University of Rhode Island, 100 Flagg Road, Kingston 02881 RI, USA R. McKinney U.S. Environmental Protection Agency, Atlantic Ecology Division, 27 Tarzwell Drive, Narragansett, RI 02882, USA J. Chaves The Ecosystems Center, Marine Biological Laboratory, Woods Hole, MA 02543, USA

i.e., high to low point source loading. There is also evidence of a west to east gradient of heavy to light values of δ15N in the bay consistent with circulation patterns and residual flows. The Providence River Estuary, just north of Narragansett Bay proper, receives 85% of STP inputs to Narragansett Bay, and lower δ15N values in macroalgae there reflected preferential uptake of 14N in this heavily fertilized area. Differences in pH from N stimulated photosynthesis and related shifts in predominance of dissolved C species may control the observed δ13C signatures. Unlike the macroalgae, the clams were remarkably uniform in both δ15N (13.2±0.54‰ SD) and δ13C (−16.76±0.61‰ SD) throughout the bay, and the δ15N values were 2–5‰ heavier than in clams collected outside the bay. We suggest that this remarkable uniformity reflects a food source of anthropogenically heavy phytoplankton formed in the upper bay and supported by sewage derived N. We estimate that approximately half of the N in the clams throughout Narragansett Bay may be from anthropogenic sources. Keywords Nitrogen . Carbon . Stable isotope . Narragansett Bay . Sewage . Macroalgae . Hard clams . Eutrophication

Introduction The anthropogenic fertilization of Narragansett Bay began in earnest on Thanksgiving Day, November 30, 1871, with a celebration to mark the opening of the public water supply to the city of Providence (Nixon et al. 2008). The construction of a sewer system began soon thereafter, and the addition of nitrogen (N) and phosphorus (P) to the bay rose rapidly as public health infrastructure spread throughout

54

the urban areas at the head of the bay and within the watershed. By about the middle of the twentieth century, total N inputs reached a plateau, where they remain today, whereas P inputs have declined since the 1970s (Nixon et al. 2008). There was a change in the relative abundance of the organic and inorganic N forms in the 1970s, as sewage treatment plants upgraded to full secondary treatment, and dissolved organic N declined from about 60 to 25% of direct sewage N loading (Nixon et al. 2005). At current rates of nutrient input, mostly from point sources, Narragansett Bay is among the more intensively fertilized estuaries in the USA (Nixon and Pilson 1983). This history contrasts sharply with that of many coastal systems of the southeast and Gulf of Mexico where nonpoint sources of nutrients are most important, and fertilization only became significant with rising fossil fuel combustion and the use of synthetic fertilizer after the Second World War (Galloway and Cowling 2002). The situation of the last four to five decades of high and relatively steady N loading to Narragansett Bay is about to change as sewage treatment plants are being required to add biological denitrification during the months of May through October, inclusive. The final level of N reduction for each plant is still uncertain, and the effect of N reductions by treatment plants in the watershed is obscured by uncertainties about N attenuation in the rivers (e.g., Seitzinger et al. 2002; Van Breemen et al. 2002). Based on a detailed inventory of N inputs to the bay and to the rivers, we estimate that the reduction in total N input to the bay during the May–October period will ultimately be about 35% or higher (Nixon et al. 2008). The purpose of the N reduction is to reduce hypoxia in the bottom waters of the Seekonk and Providence River Estuaries at the head of the bay, in the upper bay, and in Greenwich Bay, a side arm of Narragansett Bay (Fig. 1). While hypoxia is a common summer feature in the estuaries, it is episodic in the upper bay and Greenwich Bay and appears in association with weaker neap tides in late summer (Bergondo et al. 2005). As there is strong experimental (Oviatt et al. 1995) and field stoichiometric evidence (Kremer and Nixon 1978; Pilson 1985a) that N is the nutrient whose supply most limits primary production in Narragansett Bay during summer, such a marked reduction in N may reduce hypoxic events in the bay. There is a wide body of experimental and field evidence suggesting that various aspects of secondary production in Narragansett Bay may be limited by food supply during late summer and fall. Field studies by Durbin and Durbin (1981) and Campbell (1993) suggested foodlimited production in the dominant summer copepod species, and Grassle and Grassle (1984) and Rudnick et al. (1985) found field evidence of food limitation of the benthos in late summer, especially for surface feeders and some deposit feeders (e.g., nematodes). Experiments using

Estuaries and Coasts: J CERF (2008) 31:53–69

the large (13 m3) Marine Ecosystems Research Laboratory (MERL) mesocosms replicating Narragansett Bay conditions have shown food limitation of zooplankton (Sullivan and Ritacco 1985), as well as of benthos (Maughan 1986; Beatty 1991), and the growth of juvenile menhaden (Keller et al. 1990). Given this site-specific evidence, as well as more general correlative evidence of the positive relationship between primary production and the yields of fish (Nixon 1988; Iverson 1990; Nixon and Buckley 2002) and the standing crop of benthos (at least when anoxia is not involved; Kemp et al. 2005), it is reasonable to ask if the upcoming N reductions might also have a negative effect on secondary production in Narragansett Bay. Before this question can be addressed, it is essential to understand how pervasive and widespread the N from secondarily treated effluent is in the primary and secondary producers. In this paper, we report the results of a study to address this question using the relative abundance of the stable isotopes of N (15N:14N) and C (13C:12C) in the tissues of macroalgae and the hard clam or quahog, Mercenaria mercenaria, collected throughout Narragansett Bay (Fig. 1). These ratios are hereafter described using the standard delta notation, δ15N and δ13C, expressed as per mill (‰; Fry 2006). A similar approach has been applied in other coastal systems with varying success (e.g., Rau et al. 1981; Cifuentes et al. 1988; Spies et al. 1989; Hobbie et al. 1990; Carmichael 2004), but Narragansett Bay is particularly well suited for such a study for many reasons. Despite the bay’s history of intensive N enrichment and an incomplete understanding of the distribution and fate of anthropogenic N in the bay, the N budget for the bay has been described and the inputs are well known (Nixon et al. 1995, 2008). Second, the major anthropogenic sources are geographically concentrated at the head of the system. While there are six relatively small sewage treatment plants (STP) distributed around the bay, 82% of the sewage N that is discharged directly to Narragansett Bay proper enters the Seekonk River and Providence River Estuaries at the head of the bay from three secondary treatment plants. A single facility, the Fields Point treatment plant, provides almost 50% of the direct sewage N input (Nixon et al. 2008). About 90% of the N discharged into Narragansett Bay proper from the rivers also enters through these two estuaries. As rivers and STPs represent the major sources of N loading to Narragansett Bay, the vast majority of N enters the bay at the head of the system. Direct groundwater input to the bay is very small relative to surface drainage (Nowicki and Gold 2008; Pilson 2008). The direct discharge of sewage N accounts for about 30% of the total N input on an annual basis and STPs in the watershed account for a large amount of the N brought into the bay by rivers. Sewage N could account for close to 100% of the


55

Fig. 1 Maps of study locations showing sampling sites in and around Narragansett Bay, Rhode Island. a Narragansett Bay and Providence River Estuary. b Coastal lagoons or salt ponds. c Block Island

total N delivered by the two largest rivers that enter Narragansett Bay proper, the Blackstone and Pawtuxet (Nixon et al. 2008). Direct atmospheric deposition of N onto the bay accounts for about 5% of the total input

(Nixon et al. 1995, 2008). While there is a significant amount of dissolved inorganic nitrogen (DIN) that enters the bay from Rhode Island Sound in the estuarine or gravitational circulation (Nixon and Pilson 1983; Chaves

56

Mean annual primary production (g m-2 d-1)

6

Mean summer primary production (g m-2 d-1)

2004), the stable isotopic ratio of the ammonia and nitrate in this source differs considerably from that of the rivers and STPs. Ammonia, the dominant form of N released from the sewage treatment facilities, is characterized by particularly heavy (typically >10‰) δ15N values, a characteristic of secondary treatment effluent (Sheats 2000). Offshore δ15N values of DIN are much lighter, typically on the order of 5‰ (see Fry 2002; Chaves 2004). The combination of anthropogenic sources at the head of the bay and relatively unaffected Rhode Island Sound at the mouth leads to strong upper to lower bay gradients in virtually all pollutants examined in the system (e.g., Valente et al. 1992; Oviatt et al. 2002). Third, the residence time of the water in the bay averages more than 30 days during the summer growing season (Pilson 1985b), a time sufficiently long that virtually all of the anthropogenic N is taken up by the biological system. During summer, concentrations of DIN in the surface waters of the mid and lower bay are commonly below 1 μM (Kremer and Nixon 1978; Pilson 1985a; http:// www.gso.uri.edu/phytoplankton/). Under these conditions, isotopic fractionation by the primary producers may be minimal. Fourth, Narragansett Bay is a phytoplankton-based ecosystem in which production by macrophytes and allocthonous inputs from emergent wetlands are of little importance (Kremer and Nixon 1978). There are strong gradients in phytoplankton abundance and productivity from the enriched upper bay to the mouth (e.g., Kremer and Nixon 1978; Oviatt et al. 2002; Fig. 2) that one might expect to be reflected in the isotopic signatures of sedentary filter feeding animals in the bay. Last, compared with some other heavily enriched systems where stable isotopes have been examined (e.g., Delaware Estuary), Narragansett Bay proper is a highsalinity system over virtually its entire length (from about 32–33 at the mouth of the bay to 18 at the head of the Providence River; Doering et al. 1990). As a result, there is little opportunity for flocculation during mixing, and there is also no distinct turbidity maximum that might affect N transport and cycling. While N isotopes are obviously relevant to the purpose of this paper, our interest in δ13C may require some explanation. During a MERL mesocosm experiment involving the fertilization of lower Narragansett Bay waters with inorganic nutrients (DIN, dissolved inorganic P, and Si), Gearing et al. (1991) observed that δ13C values of phytoplankton in the most nutrient-enriched mesocosm were an average of 4.3‰ heavier than phytoplankton in control mesocosms receiving no nutrient enrichment (δ13C=−17.3‰ compared with −21.6‰, respectively). This isotopic enrichment was also reflected in zooplankton and in all feeding types of the benthos. Hard clams from a


Providence River Estuary West Passage East Passage

5 4 3 2 1 0 6 5 4 3 2 1 0 0

10 20 30 40 Distance from Fields Point (km) Fig. 2 Mean annual primary production (top panel) and summer primary production (June, July, August; bottom panel) plotted against distance from Fields Point. The line at approximately 8 km marks the mouth of the Providence River Estuary. Productivity data from the Providence River Estuary are shown to the left of the line in gray. The gray point at 39 km from Fields Point was collected at a station at the very bottom of Narragansett Bay and is representative of both the East and West Passages. Data were collected approximately biweekly between 1997 and 1998 and are from Oviatt et al. (2002). Chlorophyll a data were also collected as part of this study and show nearly identical trends to the productivity

control mesocosm had δ13C of −21.5‰ compared with −17.1‰ in the nutrient-enriched mesocosm. The conclusion of Gearing et al. (1991, p. 300) that “The intense phytoplankton blooms caused by addition of inorganic nutrients resulted in carbon with a distinctive isotope ratio” was consistent with the earlier studies by Fry and Wainright (1991) who observed heavier δ13C values during spring blooms on Georges Bank compared to adjacent deeper waters and by Smith and Kroopnick (1981) who used corals to demonstrate that “Metabolically active aquatic communities…can generate variations in both the chemical and isotopic compositions of the water surrounding them.” An ingenious study by Schell (2000) found a lightening of δ13C values in annual winter deposits of baleen plates from bowhead whales, which were used as a proxy for phytoplankton. The decline in δ13C values was attributed to a decline in primary productivity in the whales’ feeding ground, the Bering and Chukchi Seas, associated with


57

Concentration (mmol kg-1)

2.5 HCO3-

2.0

1.5

CO2

1.0

0.5 CO320.0 6.0

6.5

7.0

7.5

8.0

8.5

9.0

pH 2 Fig. 3 Concentrations of HCO 3 , CO2, and CO3 over a pH range of 6–9 at a salinity of 30. The pH change is driven by changes in the concentration of total CO2 such as could be caused by net respiration or net photosynthetic production. Calculated using relationships in Pilson (1998)

60

Chlorophyll a (mg m-3)

50 Enriched 40 30 20 Control

10 0 8.8 8.6

Enriched 8.4

pH at Dusk

climate change. Chanton and Lewis (1999) also discuss how δ13C values in estuarine systems may reflect whether the system is net heterotrophic or autotrophic. The mechanism by which inorganic nutrient enrichment and the resulting high rates of carbon fixation produced the increase in 13C content of the food web in the MERL experiment may have been increasing use by phytoplankton of HCO 3 compared with free CO2 at the higher pH resulting from enhanced photosynthesis. At a salinity of 30 (characteristic of much of Narragansett Bay and the MERL systems), there is virtually no free CO2 above a pH of 7.8, while HCO 3 is abundant (Fig. 3). Measurements of pH taken during the MERL experiments showed that pH values in the most intensively fertilized mesocosms exceeded 8.2 during blooms, whereas the mean pH in three control mesocosms was almost always below 8.0 (Fig. 4; Keller et al. 1987). The isotopic fractionation of inorganic C in sea water is such that free dissolved CO2 has a δ13C of about −9‰, whereas that of bicarbonate is between 0 and 1‰ at summer temperatures (Zhang et al. 1995; Fry 2006). Recent experiments have provided convincing evidence that both marine phytoplankton and many marine macroalgae can take up bicarbonate directly (Cook et al. 1986; Drechsler et al. 1993; Larsson et al. 1997; Larsson and Axelsson 1999; Cassar et al. 2004). As the large nutrient inputs to Narragansett Bay enter through the Providence River Estuary and there is a strong gradient in phytoplankton production from high values in the upper bay to much lower rates in the lower bay, particularly during the summer (Oviatt et al. 2002; Fig. 2), δ13C measurements may provide an additional indication of where in the bay the phytoplankton supporting the growth of the hard clams are formed. We also measured δ13C in the

8.2 Control

8.0 7.8 7.6 7.4 7.2 7/2

7/16

7/30

8/13

8/27

9/10

9/24

Date Fig. 4 Approximately weekly mean (n=3) chlorophyll a concentrations and pH measurements made at dusk during a MERL mesocosm experiment performed by Keller et al. (1987) during July, August, and September 1984. The enriched mesocosms received an eightfold enrichment of inorganic nutrients (N and P) over control tanks

macroalgae around the bay to see if, in fact, the gradient in metabolism within the system was reflected in the C isotopic composition of primary producers. We are not the first to recognize the attractions of Narragansett Bay as a site where the distribution of stable isotopes might be useful in documenting human effects. The first measurements of δ15N in the bay were made by Garber (1982) on sediment cores and benthic infauna collected near the mouth of the Providence River Estuary and in the mid bay. He found little variation with depth in the sediments (to 30 cm at one station and about 10 cm at another) or between locations. He also found no dramatic difference between locations for the benthic animals examined and they appeared similar to the sediments, with δ15N of 9.4–11.4‰. The first use of δ13C to describe food chains in the bay also showed no difference between sediments in the West and East Passages or with depth in the sediments (Gearing et al. 1984, 1991). An important finding from these studies was that the heavy δ13C signature in benthic animals in the bay suggested that they were feeding preferentially on diatoms rather than

58

nanoplankton. This is consistent with the observations of Fry and Wainright (1991) that diatoms are not only heavier than other phytoplankton but also a very important food source for zooplankton in the Gulf of Maine. More recently published studies examined δ15N in intertidal marshes within Narragansett Bay and reached varying conclusions that local N sources are reflected in salt marsh plants and animals (McKinney et al. 2001; Cole et al. 2004) or that the overall pollution gradient in the bay was reflected in the δ15N of some species examined in three marshes located in the upper, mid, and lower bay (Pruell et al. 2006). The one unpublished study of open water particulate matter in the bay found little seasonal variation in δ15N, but a spatial distribution of heavy values in the upper bay and lighter values offshore suggested a conservative mixing of the isotopes (Chaves 2004). The same study also examined δ15N in whole tissues of intertidal blue mussels (Mytilus edulis) along the gradient of the bay and found that the animals with the heaviest δ15N were in the upper bay, becoming progressively lighter both toward the mouth and up into the Providence River Estuary. The latter was attributed to assimilation of sewage particulate matter with relatively light N isotope values (Chaves 2004). Building on this background, our purpose was to attempt to measure the spatial extent of anthropogenic heavy N in the Providence River Estuary and throughout Narragansett Bay proper, to attempt to quantify the extent to which an important sedentary animal appeared to be dependent on organic matter produced with the heavy anthropogenic N. We also wanted to expand on the work of Gearing et al. (1991) to see if the clams growing in the anthropogenically fertilized and more highly productive upper bay regions were enriched in δ13C versus marine phytoplankton (δ13C≈–22‰) or terrestrial C3 organic matter (≈–28‰; Fry 2006). The most straightforward approach would have been to measure the isotopic ratios of DIN in the water throughout the system. During summer, the concentrations of ammonia and nitrate over much of the mid and lower bay are very low and variable, with a mean DIN concentration on the order of 1.3 μM (http://www.gso.uri.edu/phytoplankton/). With the analytical and financial resources available to us, it would not have been possible to obtain enough samples, in high enough concentrations, to adequately characterize the δ15N in the DIN throughout the bay. Analyzing the isotopic ratios of the phytoplankton is very difficult in shallow coastal systems where the phytoplankton comprise only a small fraction of the particulate matter in the water. A recent analysis by DiMilla (2006) of total suspended solids less than 150 μm from a fall collection found that phytoplankton may account for less than 1% of the material filtered from the surface water in Narragansett Bay. We


chose to sample attached macroalgae that occur throughout the bay. Macroalgae are sensitive indicators of N loading, and their δ15N reflects the isotopic signature of their N sources (Hobbie et al. 1990; Costanzo et al. 2001; Savage and Elmgren 2004; Martinetto et al. 2006; Thornber et al. in press). The choice of hard clam was simpler, as it is the signature animal of the bay, widely distributed throughout the system, and a sedentary filter feeder.

Methods Study Sites and Sample Collection We collected attached macroalgae at 19 sites during very low tides between September 8 and 21, 2006 from rocks above and just below the water surface at low tide. All conspicuous species were sampled, and multiple individuals (at least three) were collected for each species. Collection locations were chosen to avoid any local terrestrial influences from potential groundwater seeps. Collections were always from exposed bedrock or at the base of mid-channel lighthouses built on rock fill (Fig. 1). Macroalgae were also collected from Great Salt Pond, a coastal lagoon on Block Island, approximately 16 km offshore from the mouth of Narragansett Bay. We had previously (2001) collected macroalgae from two mainland coastal lagoons (locally called salt ponds), Ninigret and Quonochontaug, which are connected to the relatively unpolluted Block Island Sound (Fig. 1). Macroalgal samples were rinsed with tap water, and epiphytic algae and organisms were removed. Specimens were photographed for later identification, and the samples made up of multiple individuals were dried at 65°C for at least 24 h and ground to a fine powder with a mortar and pestle; subsamples were weighed in tin capsules for isotopic analysis. Narragansett Bay hard clams were collected from ten stations, primarily by the Rhode Island Department of Environmental Management, using a hydraulic shellfish dredge between late June and mid September 2005 and in August 2006 or by divers using SCUBA (Table 1). All Narragansett Bay clams were collected from subtidal locations deeper than 2 m and from a variety of sediments ranging from clayey silt to gravelly sand. Clam lengths and widths ranged from 40 to 100 mm and 22 to 65 mm, respectively, with an average clam length of 74 mm and width of 46 mm. We also collected hard clams from four sites outside of Narragansett Bay, including Ninigret and Quonochontaug lagoons, and from Old Harbor and Great Salt Pond on Block Island (Fig. 1). Ninigret lagoon is 6.92 km2, with a mean depth of approximately 1.2 m and mean salinity of 24. While Quonochontaug is smaller (2.96 km2), it is


59

Table 1 Sampling dates, depths, and methods for hard clams (Mercenaria mercenaria) analyzed in this study. The number of individual animals analyzed at each station (n) is also given along with the mean±standard deviation of the δ15N and δ13C Station Narragansett Bay GSO Dyer Island Brenton Cove Conimicut Ohio Ledge Wickford Calf Pasture Providence River West Jamestown Bristol Whole bay Block Island B.I. Old Harbor B.I. Great Salt Pond Coastal ponds Ninigret Quonochontaug a

Dates sampled

Mean depth (m)

Collection

15

N

SD

n

13

C

SD

n

September 1–2, 2005 September 1–2, 2005 September 1–2, 2005 June 24–September 19, June 24–September 19, June 24–September 19, June 24–September 19, June 24–September 19, June 24–September 19, August 16, 2006

8 2 5 4 8 5 6 4 8 4

Diver Diver Diver Dredgea Dredgea Dredgea Dredgea Dredgea Dredgea Dredgea

13.4 13.3 12.9 13.1 13.2 13.4 13.1 12.6 13.8 13.3 13.2

0.1 0.3 0.5 0.5 0.4 0.4 0.5 0.5 0.4 0.4 0.5

75 24 45 61 30 37 60 95 48 10 485

−17.1 −17.3 −16.2 −16.4 −16.9 −17.4 −17.2 −16.2 −16.8 −16.5 −16.8

0.3 0.4 0.3 0.4 0.5 0.3 0.7 0.5 0.3 0.1 0.6

15 19 20 36 20 26 31 56 14 10 247

November 7, 2005 September 2006

3

Diver Diver

11.1 9.1

0.6 0.3

25 12

−17.6 −15.9

0.5 0.2

15 12

June 2006 June 2006

5 8

Divera Divera

8.3 9.0

0.6 0.4

19 36

−19.6 −17.9

0.4 0.3

19 36

2005 2005 2005 2005 2005 2005

Collected by the Rhode Island Department of Environmental Management, Division of Marine Fisheries

slightly deeper (approximately 1.8 m) and more saline (28). While no public STPs discharge into any of the lagoons, they receive anthropogenic N through groundwater contaminated by individual septic systems, and excess macroalgal growth and eutrophication in isolated areas have been documented (Lee and Olsen 1985). The foot muscle of each hard clam was removed and dried at 65°C for at least 48 h. Dried samples were ground to a fine powder with a mortar and pestle and stored in acid-washed scintillation vials in a desiccator until analysis.

for instrument drift in each run, and no drift was observed in analyzing the samples discussed here. We analyzed 20% of the hard clam samples and all macroalgae samples in duplicate, with clam standard deviations generally less than 0.10‰ for δ13C and 0.20‰ for δ15N. The macroalgae samples collected earlier from Ninigret and Quonochontaug Lagoons were analyzed at the Boston University Stable Isotope Facility (http://www.bu.edu/sil/ index.htm). Statistical Analyses

Isotopic Analysis We determined C and N stable isotopic values for macroalgae and hard clams using a Carlo-Erba NA 1500 Series II elemental analyzer interfaced to a Micromass Optima mass spectrometer with a precision of better than ±0.3‰ at the US Environmental Protection Agency, Atlantic Ecology Division in Narragansett, Rhode Island. The C isotopic composition was expressed as a part per thousand (per mill) deviation (δ13C ‰) from the reference standard PDB, and the N isotopic composition (δ15N ‰) was expressed as a part per thousand (per mill) difference from the composition of N2 in air (Mariotti 1983) as follows: d X ¼ ½ðRsample =Rstandard Þ 1 103 where X is δ13C or δ15N and R is the ratio 13C/12C or N/14N. Samples were analyzed randomly in batches of approximately 30. We used laboratory standards to check 15

We performed linear regressions on the macroalgae data for δ15N and δ13C with distance from Fields Point (see Fig. 1). After using a one-way analysis of variance (ANOVA) and a paired Student’s t test to confirm that the Hog Island Station was significantly lighter than those stations to the north and south of it, we then reperformed the linear regressions without the Providence River Estuary stations and the Hog Island station (see “Results and Discussion” for more explanation). To identify differences among macroalgal phyla and families, we used an analysis of covariance (ANCOVA) for both δ15N and δ13C, again with distance down the bay. We used a t test to determine significant differences between δ13C values in the Providence River Estuary versus the rest of Narragansett Bay. To look for differences among stations in hard clam δ15N and δ13C values, we performed a one-way ANOVA and then a paired Student’s t test for both N and C isotopes. The ANCOVA was performed in SAS (SAS Institute Inc. 1982), whereas

60


JMP (JMP Release 6.0.0 2005) was used for all other statistics.

Results and Discussion Macroalgae Nitrogen There were significant spatial patterns in δ15N of macroalgae (Fig. 5) despite variation among species (Tables 2 and 3). The gradient from generally heavier δ15N in upper bay algae to lighter at the mouth of Narragansett Bay (R2 =0.43, p40 μM) even during summer in the Providence River Estuary surface water compared with very low values throughout the rest of the bay (Kremer and Nixon 1978; Doering et al. 1990). We attribute the lighter δ15N of the macroalgae in the Providence River Estuary to fractionation favoring uptake of the lighter isotope (Fry 2006; York et al. 2007). This process may also be evident in the algae collected at the Hog Island lighthouse (Fig. 1), which sits directly in the outflow from Mt. Hope Bay (Kincaid 2006) and the sewage outfall from the city of Fall River, about 15 km away. Fall River is the second largest source of sewage N that enters Narragansett Bay directly, amounting to approximately 40% of the Fields Point wastewater treatment facility N input (at Fields Point, adjacent to the Fuller Rock station; see Fig. 1; Nixon et al. 1995), and the Hog Island station was significantly lighter than the stations immediately to the north and south (2 and 1.5‰, respectively). With the Providence River Estuary and Hog Island stations removed, the fit of the regression line improves (R2 =0.68, p0.05).

δ13C, Level

δ13C, Mean

N

Phaeophyta Chlorophyta Rhodophyta

−14.9 −16.6 −19.1

14 20 16

A

Fucaceae Codiaceae Ulvaceae Dasyaceae Gigartinaceae Rhodomelaceae Areschougiaceae

−14.8 −15.4 −16.9 −17.7 −18.6 −21.3 −22.1

14 3 17 5 7 2 2

A A A A

B C

B B B

C C

D D


63

Carbon

Fig. 6 Mean δ15N values of macroalgae are shown at collection sites in the East and West Passages of lower Narragansett Bay. Standard deviations are shown in parenthesis; see Table 2 for number of samples. At the Rose Island Station, only one species of macroalgae was collected, so no standard deviation is given

δ15N values of the macroalgae in the West Passage reflect this circulation pattern, with declining values from north to south in the whole bay and from west to east in the lower bay. At the stations exposed to incoming water from Rhode Island Sound, macroalgae had lower mean values of δ15N of 7.9 and 7.8‰ for Castle Hill lighthouse and Beavertail, respectively. These values are similar to those from stations outside of Narragansett Bay in Block Island Sound, including Great Salt Pond on Block Island (7.9‰) and Ninigret and Quonochontaug lagoons on the south shore of Rhode Island (7.6 and 7.9‰, respectively). These values from outside of the bay are still heavier than those observed by Wozniak et al. (2006) in macroalgae with δ15N signatures ranging from 3.2‰ to 7.4‰ in Massachusetts salt marshes. Heavy anthropogenic DIN appears to be present throughout the surface water of virtually the entire bay even when concentrations are very low during the summer growing season. In the East Passage, the δ15N was elevated compared to offshore conditions at least as far south as Newport (Rose Island lighthouse) on the eastern side where the residual flow is generally north into the bay. We did not collect any samples on the western side of the East Passage where circulation records show a southward residual flow under some, but not all, wind conditions (Kincaid et al. 2008). In the West Passage, the δ15N was elevated in the middle and western side of the passage all the way to Rhode Island Sound.

Marine macroalgae are notoriously variable in their C isotopic ratios (e.g., Fry and Sherr 1984; Raven 1997), which can be highly species-dependent (Raven et al. 2002); δ13C values can vary by over 10‰ in the thallus of an individual of some brown macroalgae. Macroalgae with values less than −30‰ tend to be red, whereas very heavy macroalgae (values greater than −10‰) are usually green (Raven et al. 2002). The Narragansett Bay species were no exception to this variability (Fig. 5, Table 2). The range in mean δ13C among macroalgae at the family level (7.2‰) and at the species level (11.9‰) was greater than for δ15N (4.0 and 6.1‰, respectively; Tables 2 and 3). Within-site variability was usually much greater for C isotope ratios. Coefficients of variation for site mean δ13C were often two to three times higher than for δ15N (Table 2). At Castle Hill lighthouse, where six species were collected, the coefficient of variation for δ13C was 10% compared with 3% for δ15N. At least in part because of this variability, there was no overall gradient in δ13C in macroalgae down the length of the bay as there was for δ15N (R2 =0.03, p