control of nutrient concentrations in the seekonk ......(qm + e,,) lower layer: predicted c ’ , =...

Estuaries Vol. 13, No. 4, p. 418-430 December 1990

Control of Nutrient Concentrations in the

Seekonk-Providence River Region of

Narragansett Bay, Rhode Island

P. H. DOERINC C. A. OVIATT M. E. Q. PILSON Marine Ecosystems Research Laboratory Graduate School of Oceanography University of Rhode Island Narragansett, Rhode Island 02882-l 197

ABSTRACT: Six synoptic samplings of nutrient concentrations of the water column and point-source inputs (rivers, sewage treatment plants) were conducted in the Seekonk-Providence River region of Narragansett Bay. Concentrations of nutrients (NH,+, NO,- + NO,-, P0,-3, dissolved silicon, particulate N, particulate C) were predicted using a conservative, two-layer box model in order to assess the relative influence of external inputs and internal processes on observed concentrations. Although most nutrients were clearly affected by processes internal to the system, external input and mixing explained most of the variability in and absolute magnitude of observed concentrations, especially for dissolved constituents. In the bay as a whole, two functionally distinct regions can now be identified: the Seekonk-Providence River, where dissolved nutrient concentrations are externally controlled and lower Narragansett Bay where internal processes regulate the behavior of nutrients. A preliminary nitrogen budget suggests that the Seekonk-Providence River exports some 95% of the nitrogen entering the system via point sources and bottom water from upper Narragansett Bay.

Introduction Owing to the present position of sea-level, most

rivers empty into estuaries or marginal seas rather than directly into the coastal ocean (Thurman 1985). Furthermore, major urban centers are often located on estuaries. As a result, estuaries re- ceive some of the highest inputs of nutrients (N and P) on an area1 basis of any class of ecosystems (Nixon et al. 1986). Understanding the behavior of nutrients in estuaries has important implications for global nutrient budgets (Wollast 1983; Kaul and Froelich 1984) and for controlling eutrophi- cation of these systems (e.g., Tippie 1984).

conservative box model reliant upon external input and mixing in the water column (Officer 1980). Predicted nutrient concentrations are compared to those measured in the field by functional regression (Ricker 1973). The slope of this regression is employed to calculate the proportion of observed concentration which may be attributable to external input and mixing. By inference, the remainder is due to internal processes. Finally preliminary estimates of net nitrogen flux through this portion of Narragansett Bay are presented.

Study Area

Perhaps one of the most pivotal questions con- cerning nutrients in estuaries is the degree to which estuaries behave as traps, retaining and recycling nutrients within the system (e.g., Smullen et al. 1982; Nixon 1987a). Related to this issue are the relative contributions of external nutrient supply and internal nutrient recycling to observed concentrations within the estuary (D’Elia et al. 1983; Oviatt et al. 1984; Pilson 1985a).

In this paper we report results of six synoptic surveys of nutrient concentrations, salinity, and nutrient inputs in the Seekonk-Providence River region of Narragansett Bay. Tidally averaged nutrient concentrations are estimated using a two-layer,

The Seekonk and Providence rivers lie at the head of Narragansett Bay, Rhode Island. Together they comprise about 7% of the total area of the bay and at mean low tide, hold some 3.3% of its water (Chinman and Nixon 1985). The mean depth of the Seekonk River is 1.29 m while that of the Providence River is 3.99 m (Chinman and Nixon 1985). However, a relatively deep (13- 14 m) ship- ping channel runs the length of the Providence River. A marginally navigable, much shallower (3- 4 m) channel also exists in the Seekonk.

The bay in general has been characterized as “well mixed” (Kremer and Nixon 1978) or “weak- ly stratified” (Weisberg and Sturges 1976), with

Q 1990 Estuarine Research Federation 418 0180-8347/90/040418-13$01.50/O

Control of Nutrient Concentrations 419

near surface and bottom salinities differing by about 2%~ (Pilson 1985b). By contrast, the Seekonk-Prov- idence River region exhibits considerably stronger stratification (Nixon 1987b; Spaulding 1987). The long-term average freshwater input to Narragan- sett Bay is estimated to be 105 m3 s-l with about 50% of the gauged input to the bay emptying into the Seekonk-Providence River (Pilson 1985b). The average residence time of freshwater in the bay is 26 d (range 10 to 40 d) and depends on freshwater input (Pilson 1985b). Best estimates for the See- konk-Providence River range from 3 d to 10 d (Spaulding 1987).

Because it is surrounded by the greater Provi- dence Metropolitan area, the Seekonk-Providence River receives considerable amounts of nutrients, trace metals, and hydrocarbons, mostly from rivers and sewage treatment plants (Oviatt et al. 1984). Comparison of bay-wide estimates of dissolved inorganic nitrogen loading (Nixon and Lee 1979) with those for the Seekonk-Providence River (Oviatt 1981) suggest that nearly 60% of the total loading to the bay enters in the Seekonk-Provi- dence River. Perhaps due to a combination of relatively small volume and relatively high pollutant loading, the Seekonk-Providence River is arguably the most degraded region of Narragansett Bay (Oviatt et al. 1984). These factors no doubt also contribute to the horizontal concentration gradients observed for many substances both in the water column and sediments of Narragansett Bay. Such gradients show a consistent pattern: high concentrations in the Seekonk-Providence River, de- creasing sharply toward the mouth of the bay (see Oviatt et al. 1984).

Methods SAMPLING

Six cruises were conducted at intervals of about two months (1986: October 11, December 15; 1987: March 11, April 22, June 27, August 12). Cruises were timed to occur during “dry weather”: no major rainfall (> 0.25 inches) within 4 d or 5 d of initial point-source sampling. The purpose of this constraint was to minimize the effects of nonpoint-sources upon the system. During each cruise 10 stations (Fig. 1) in the Seekonk (3) and Provi- dence rivers (7) were occupied at both high and low tide, In general, stations were sampled within + 1.5 h of slack tide, Discrete water samples were pumped (bellow or hand) from within 1 .O m of the surface and 1.0 m of the bottom through acid- rinsed (1% HCl) Teflon tubing. Vertical hydro- graphic profiles of temperature and salinity were obtained either with an Applied Microsystems, Inc. STD-12 (Providence River) or a Beckman Instru-

. ESTUARY

+ RIVER

Fig. 1. Station locations. Solid lines delimit boxes used in modelling effort. BR = Blackstone River, BV = Blackstone Val- ley Sewage Treatment Plant (STP), TM = Ten Mile River, MR = Moshassuck River, WR = Woonasquatucket River, FP = Field’s Point STP, EP = East Providence STP, PR = Pawtuxet River.

ments Inductive Salinometer (Seekonk River). The resolution of depth was about 0.5 m.

Coincidentally with each cruise, five rivers and three sewage treatment plants (Fig. 1) were sampled on the 3 d preceding each cruise. Rivers were sampled at low tide to minimize saltwater intrusion. Sampling was usually conducted from a bridge or other structure which allowed access to mid-stream. Samples were taken with a plastic bucket suspend- ed from a rope. An inverted funnel prevented con- tamination of the sample by drippings from the rope.

Composite (24 h) samples of effluent were collected fram the three sewage treatment plants by plant operators. These were refrigerated until returned to the laboratory. In general, samples were brought to the laboratory within 24 h of collection.

PROCESSING AND ANALYSIS

Samples for dissolved inorganic nutrients (NH,+, NOs- + NOB-, POdm3, dissolved silicon) were manually (60-ml plastic syringe) passed through 47 mm diameter, 0.4 pm pore size membrane filters (Nu-

420 P. H. Doering et al.

CONCEPTUAL BOX MODEL OF THE SEEKONK AND PROVIDENCE RIVERS

I v Ql v *_ $ Q3

Sl S2 s3

t

QVl EVl QV2 EV2 QV3 EV3

S’l Q’2 $72 4’3 s’3 +4’4

S’4

Fig. 2. Diagram of box model used to predict nutrient concentrations. S = salinity, R = freshwater input, Q’s = coefficient of advective transport, E’s = coefficient of nonadvective transport, V = vertical, number = box number. Superscripts: ’ = lower layer.

clepore) into 60-ml polypropylene jars. These were stored on ice until returned to the laboratory where they were frozen until analysis on a Technicon Autoanalyzer (Lambert and Oviatt 1986). Dupli- cate samples from station 1 were refrigerated for silicate analysis to avoid problems caused by freezing low salinity samples (Macdonald et al. 1986).

Particulate carbon and nitrogen samples were passed manually (60-ml plastic syringe) through 13 mm diameter Whatman GF/F glass-fiber filters (nominal pore size 0.7 pm) which had been com- busted at 425°C. Duplicate filters were stored on ice until returned to the laboratory where they were dried (40°C) and stored until analysis. The filtrate was collected and its weight determined in the laboratory. Carbon and nitrogen retained on the filters were determined by elemental analysis on a Carlo Erba Model 1106 Elemental (CHN) An- alyzer. The mean coefficient of variation for 2 18 analyses of replicate filters (n = 2) was 12% for carbon and 14.5% for nitrogen.

Discrete salinity samples were stored in 60-ml plastic bottles. Duplicate samples were analyzed on

TABLE 1. Calculation of advective (Q’s) and nonadvective (E’s) transport coefficients in the first two boxes of the model. The solutions for remaining box follow those for Box 2. For definition of terms see Fig. 2. Superscript ’ denotes lower layer. Subscripts refer to box number.

For Box 1:

Q, = R,.]S’,/(S’, - %)I Q’l = R,.lS,/(S’, - %)I Qv, = Q’2 E,, = (Q,% - Q~,s’,)/(s’, - S,)

For Box 2:

Qz = W&S’, - %)l.(Q, - Q’2 + RJ Q’s = l%./(S’, - %)l.(Q, - Q’z + R,)

TABLE 2. Equations used to predict the concentration of a given substance in the upper and lower layers of a given box. Subscript m refers to box number, and r to river, superscript ’ denotes lower layer, C = concentration. Other terms as in Fig. 2.

Upper layer: Predicted C, = [Q,,-ICm-, + R,C,, + C’,(Qvm + Evm)]/

(Qm + E,,)

Lower layer: Predicted C’, = (Q’m+lC’m+l + E&W(Q’, + Qvm + E,,)

an Autosal Model 8400 Inductive Salinometer. The mean coefficient of variation of 222 analyses of replicate (n = 2) samples was 0.84%.

Box MODEL

A two-layer conservative box model (Pritchard 1969; Officer 1980) was employed to 1) predict nutrient concentrations as a function of point source inputs and mixing, and 2) estimate net flux of nutrients through the system to upper Narragansett Bay (Fig. 2). The model, based on that given by Officer (1980), assumes steady-state conditions, that net circulation effects rather than tidal exchange dominate, and instantaneous mixing within each layer of each box. The model captures the gross features of net circulation in a stratified estuary: landward flow of saltwater in the lower layer and seaward flow of fresher water in the surface layer, with vertical exchange (advective and nonadvective) between layers. Vertical salinity profiles in Narragansett Bay are consistent with this classic, two-layer estuarine circulation (Pilson 1985b).

The equations used to calculate advective and nonadvective transport coefficients are given in Table 1. These solutions assume that volume and salt (see Aston 19’7s) are conserved. Once trans- ports have been calculated, the steady-state concentration of a given substance can be calculated from flux (transport x concentration) considera- tions: what goes into a layer must equal that which leaves. Solutions are given in Table 2.

The net flux of nutrients through the system to upper Narragansett Bay is calculated from the difference between inputs (point source + bottom water) and outputs (down estuary flux in the surface layer). Bottom water input and surface layer output can be calculated from observed concentrations and longitudinal water transport across the boundary between the Providence River and upper Narragansett Bay.

APPLICATION OF THE MODEL

The Seekonk and Providence rivers were divided into three boxes (Fig. 1, Table 3) following


TABLE 3. Characteristics of the boxes used in model calculations. Volume and depth data are from Chinman and Nixon (1985). For sampling station locations and freshwater inputs see Fig. 1.

Area (kd) Mean Depth (m) Sampling Stations Freshwater Inputs

1. Seekonk River 2.80 1.29 1,293 BR, BV, TM 2. Fox Pt. Reach 3.00 7.03 4, 5 WR, MR, FP 3. Sabin-Nayatt Reach 18.33 3.49 6,7,&g PR, EP

Chinman and Nixon (198.5). At least two sampling and mixing, the slope may also be interpreted as stations were included in each box. Station 10 quantifying the proportion of the observed conserved as a saltwater endmember. centration attributable to these processes.

Salinity in each layer was determined by first averaging high and low tide discrete samples at each station and averaging across stations within a box. Observed concentrations of nutrients were determined similarly. Freshwater input was determined from river flow data furnished by the United States Geological Survey and sewage treatment plant discharge records. Freshwater input and concentrations were taken as the mean of the 3 d preceding each cruise.

This interpretation is relatively straightforward when the slope is less than or equal to 1 .O. A slope of 1 .O indicates that 100% of the observed concentration can be explained by inputs and mixing. A slope less than 1.0 indicates that predicted concentrations are a fraction of the observed, suggesting addition to the system by some process.

Concentrations of ammonia, nitrate + nitrite, total dissolved inorganic nitrogen (DIN), phosphate, dissolved silicon, and particulate carbon and nitrogen were calculated for each cruise, yielding 36 predictions for each constituent equally divided between surface and bottom layers. The agree- ment between predicted and observed concentrations was assessed by computing the functional regression of predicted (y) on observed (x) (Ricker 1973).

The functional relationship between observed and predicted concentrations provides two pieces of information which bear on the relative importance of external input versus other processes (either external or internal): the R2 and the slope.

When the slope is greater than 1.0, observed concentrations are less than predicted and loss from the system is indicated. The inverse of the slope measures the fraction of predicted material which is actually observed, and assumed due to input and mixing. The remainder is the fraction lost due to some unquantified process.

Results FRESHWATER INPUT

Total freshwater discharge to the system varied (Fig. 3), being relatively high during the Decem- ber, March, and April cruises and relatively low during October, June, and August. Rivers were the major source of freshwater, comprising 94-96% of the total discharge during high flow conditions

The R* of the functional relation is the same whether observed is regressed on predicted or predicted on observed. Thus the R* indicates the amount of variability in observed concentrations which can be explained by predicted concentrations. Since predicted concentrations are comput- ed as a function of input and mixing of each layer of each box, by inference, the R* measures the amount of variability in observed concentrations which may be due to these processes. The data used in the regressions encompass both temporal (the six cruises) and spatial (the three boxes) variability. We have not tried to distinguish between them. A high R* would suggest that if input and mixing produce different concentrations at different times and places, then observed concentrations would also be correspondingly different.

FRESH WATER DISCHARGE

150E

The slope of the regression quantifies these cor- responding differences. Again, since predicted concentrations are assumed to result from input

OCT EC MAR API3 JUN AUG

Month of Cruise

Fig. 3. Mean total freshwater input to the Providence-See- konk River during the 3 d preceding each cruise.


SURFACE SALINITY

30'

24- 0

/' 18-

z a12-

6-

0 5 10 15 20 25

SURFACE AMMONIA SURFACE NITRATECNITRITE

0-j , I I , r 0 5 10 15 20 25

SURFACE PHOSPHATE

121

9

x z6

E

A3

T

0 5 10 15 20 25

SURFACE PARTICULATE NITROGEN

“‘“~

0.03 , , , , / 0 5 10 15 20 25

DISTANCE KM

140'

120-

100-

80-

60-

40-

20-

o--, , ,. ,., r 0 5 10 15 20 25

SURFACE SILICATE

12”1

80- \

0 5 10 15 20 25

SURFACE PARTICULATE CARBON

4<

3

7 m2 E

1

1:L 00

0 '5 10 15 20 25

DISTANCE KM

Fig. 4. Mean and range of salinity and nutrient concentrations observed in surface waters (depth = 1 .O m) during the six cruises, versus distance from the Main Street Bridge, Pawtucket, Rhode Island at the head of the Seekonk River.

.


BOTTOM SALINITY

BOTTOM AMMONIA BOTTOM NITRATE+NITRITE

i 20-

2 z

o-i ;l;r < $a-

0 5 10 15 20 25

BOTTOM PHOSPHATE

9-j

O-$ I I I 1

0 5 10 15 20 25

BOTTOM PARTICULATE NITROGEN

0.5-I I

0 5 10 15 20 25 0 5 10 15 20 25

DISTANCE KM DISTANCE KM

;1 5 10 15 20 25

BOTTOM SILICATE

60

x 2 40

E q20

I I I 1 4 0 5 10 15 20 25

BOTTOM PARTICULATE CARBON

0-l

Fig. 5. Mean and range of salinity and nutrient concentrations observed in bottom waters (1.0 m from bottom) during the six cruises, versus distance from the Main Street Bridge, Pawtucket, Rhode Island at the head of the Seekonk River.


HIGH RIVER DISCHAR0E

41.7 8.6 L8.7

I

c f c “YLF 2 46.2 106.2 186.4 +

f t 4.5 5.0 51.5 5.8 61.4 15.6

4.5 55.9 117.4

62.6 1 8.0 1 19.6 1

CRUISE MAR

3

I I I I

76.9 11.4 33.1

$ $ t cR%F4 85.0 154.0 297.9

t t t f t t 8.1 7.4 57.6 6.4 110.8 21.2

8.1 65.7 176.5

BOX 1 BOX 2 BOX3

Fig. 6. Transport coefficients (mS s-i) for three cruises (De- cember, March, and April) when freshwater input was high. Arrows correspond to Fig. 2.

and 78-82% during low flow conditions. The Blackstone River entering at the head of the See- konk was the largest source of freshwater. Second in importance was the Pawtuxet which enters around the middle of the Providence River.

CONCENTRATIONS

The salinity of bottom water always exceeded that near the surface (Figs. 4 and 5). Dissolved nutrient concentrations exhibited an opposite trend with surface values being greater than those near the bottom. There was no consistent relationship between surface and bottom concentrations of particulate nitrogen and carbon.

The salinity of surface waters generally increased down the estuary (Fig. 4). Lowest salinities were always observed at station 1 in the Seekonk River and highest salinities at station 10 in upper Narragansett Bay.

Highest concentrations of all nutrients measured in surface waters occurred in the Seekonk River (Fig. 4). The concentration of dissolved species tended to decrease from stations 1 or 2 in the See- konk River down the estuary. Secondary peaks in concentration, however, were evident especially for ammonia and phosphate.

TABLE 4. R* of linear relationships between tidally averaged nutrient concentrations and salinity for each cruise. n = 19 for each cruise. Correlation coefficients were all negative excepting particulate N on cruise 4. If R* > 0.208 then p < 0.05.

Cruise

Oct. 1 Dec. 2 Mar. 3 Apr. 4 Jun. 5 Aug. 6

I-$ + NO,- 0.990 0.988 0.994 0.990 0.986 0.958 +

Po4s 0.402 0.893 0.874 0.752 0.118 0.292 0.662 0.016 0.726 0.441 0.295 0.194

Dissolved Si 0.986 0.945 0.996 0.974 0.922 0.956 Particulate N 0.576 0.546 0.508 0.130 0.310 0.008 Particulate C 0.573 0.482 0.830 0.106 0.540 0.037

By contrast surface concentrations of particulate carbon and nitrogen, while behaving similarly to dissolved species in the Seekonk River, tended to increase between the head of the Providence River (station 4) and upper Narragansett Bay (station 10).

Bottom water salinities increased precipitously in the Seekonk River but remained relatively con- stant in the Providence River (Fig. 5). Highest concentrations of all nutrients occurred in the See- konk River (stations l-3). Concentrations declined from the head to the mouth of the estuary, but again secondary peaks in concentration were evident (Fig. 5).

Concentrations of all nutrients were significantly correlated with salinity on most cruises (Table 4). The R2 of the linear relationship was consistently greater than 90% on all six cruises only for nitrate + nitrite and dissolved silicon.

Model Results TRANSPORT COEFFICIENTS

Transport coefficients for the three high and low river discharge cruises are given in Figs. 6 and 7. Up-estuary transport of bottom water from the Fox Pt. Reach (Box 2) to the Seekonk River (Box 1) was small. About 90% (88-96%) of the bottom water entering Box 2 was advected vertically to the upper layer. Given the constricted passage between these boxes (Fig. l), this result seems intuitively reasonable.

NUTRIENT INPUTS

The ultimate sources of nutrients considered in the model were rivers, sewage treatment plants, and bottom water from upper Narragansett Bay. In the model the former two point-sources supply nutrients to the surface layer and the latter rep- resents the initial source of nutrients to bottom water. By way of comparison, fluxes (flow x concentration) from these sources are given for two representative cruises in Table 5.

Point-source inputs of dissolved inorganic nitrogen were an order of magnitude greater than the

TABLE 5. Nutrient inputs (flow x concentration) to the See- konk-Providence River from point sources and bottom water from upper Narragansett Bay. Units are moles s-i.

High Flow Low Flow Cruise 3 March Cruise 5 June

Point Bottom Point Bottom Sources Water Sources Water

NH,’ 6.31 0.21 4.82 0.84 NO,- + NO,- 4.28 0.09 1.54 0.13 PO,-3 0.36 0.10 0.34 0.16 Dissolved Si 10.76 0.30 2.00 1.54 Particulate C 13.49 4.14 5.64 3.37 Particulate N 1.71 0.64 0.80 0.50

contribution of upper Narragansett Bay bottom water. Both sources contributed significantly to the input of P04, particulate carbon, and particulate nitrogen. The relative importance of dissolved silicon input varied with freshwater discharge. Dur- ing high flow point-source input dominated, while during low flow conditions the two sources were roughly equivalent.

PREDICTED AND OBSERVED

CONCENTRATIONS

As an internal check of the model, mean salinities were predicted in each layer, substituting salinities for concentrations in the equations given in Table 2. Functional regressions of predicted on observed salinities for both layers yielded slopes equivalent to 1.0 and intercepts not statistically different from 0.0. Correlation coefficients exceeded 0.999 in both cases (Table 6).

Control of Nutrient Concentrations

LOW RTVER DlsxAROE

425

BOX 1 BOX 2 BOX 3

Fig. 7. Transport coefficients (mS s-i) for three cruises (Oc- tober, June, and August) when freshwater input was low. Ar- rows correspond to Fig. 2.

TABLE 6. Model results: slopes and intercepts of functional regressions of predicted on observed concentrations in surface and bottom layers. Errors are the 95% confidence interval. (*) slope does not overlap 1.0 (**) intercept does not overlap 0.0. DIN = NH,+ + NO,- + NO,-. % of concentration attributable to inputs and mixing calculated as described in text. Units are pmols 1-i for nutrients and o/o0 for salinity.

Surface layer Salinity DIN NH,’ NO,- + NO,-

PO,-3 Dissolved Si Particulate N Particulate C

Bottom layer Salinity

DIN NH,+ NO,- + NO,- PO,-3 Dissolved Si Particulate N Particulate C

Slope Intercept RP

1.00 -t 0.001 0.00 + 0.02 1.000 1.29 + 0.15* -11.16 + 11.32 0.950 1.74 + 0.59* -16.88 + 21.06 0.583 0.98 + 0.08 0.85 * 3.07 0.979

1.48 + 0.29* -1.25 + 1.44 0.858 1.20 + 0.14* -4.80 + 7.92 0.948 1.77 + 0.73* -0.07 ?z 0.11 0.399 1.77 k 0.68* -0.42 & 0.64 0.469

1.00 & 0.009 -0.01 + 0.23 0.999

0.91 * 0.08* 0.01 + 3.01 0.970 0.78 f 0.13* 0.32 xk 3.10 0.905 1.05 + 0.12 0.36 a! 1.54 0.952 0.84 + 0.12 0.05 t- 0.49 0.921 0.95 + 0.07 0.58 + 2.08 0.978 0.44 f 0.20* 0.05 5 0.03** 0.288 0.37 + 0.16* 0.34 + 0.16** 0.333

% Due to Inputs and Mixing

78 57

100

:: 56 56

-

;: 100

84 100

44 37


NITRATE + NITRITE IN SURFACE LAYER

n

3 $ 60. 5 40-

z

20 -

“I - 0 2b 4'0 6b 6'0 1

Observed pmoles/l

AMMONIA IN SURFACE LAYER

n

60 8

60 60 100

Observed poles/l

Fig. 8. Relationship between predicted (y) and observed (x) concentrations of nitrate + nitrite and ammonia in surface waters. Predicted concentrations were calculated from the box model.

Functional relationships between predicted and observed nutrient concentrations were all statistically significant at p < 0.05 (Table 6). Slopes of the regression were thus all different from zero. Nonzero intercepts were found only for particulate nitrogen and carbon in the bottom layer. R2 values of regressions were generally high, signifying little scatter in data points about the regression line. Regressions for particulate carbon and nitrogen and ammonia in surface water had particularly low R2 values. Results for ammonia and nitrate + ni-

trite in surface waters are given in Fig. 8 as ex- amples of relationships with high and low R* values.

The concentrations of nitrate + nitrite in surface waters and dissolved silicon and nitrate + nitrite in bottom waters could be explained entirely by the model, as a function of inputs and mixing (slopes = 1.0). Concentrations of the remaining dissolved and particulate nutrients in surface waters were less than predicted by the model (slopes greater than l.O), suggesting loss by unaccounted for processes. By contrast, concentrations of the remaining dissolved and particulate nutrients in bottom waters were greater than predicted (slopes less than l.O), suggesting enrichment by unaccounted for processes.

Judging from the regression slopes, inputs and mixing processes in surface layers produced concentrations of dissolved constituents which achieved 57-100% of their potential value. Loss processes reduced these potential concentrations by some O- 43%. In bottom waters, inputs and mixing processes explained between 78-100% of observed concentrations of dissolved constituents. Other processes increased observed concentrations, and must account for the remaining 0 to 22%.

NETFLUXOF NITROGEN

Net fluxes of dissolved inorganic (DIN) and particulate nitrogen were calculated from point-source inputs and longitudinal fluxes between the lower Providence River (Box 3) and upper Narragansett Bay. The data (Table 7) revealed no trends in retention or export for either DIN or particulate nitrogen which may be explained by changing sea- sons or river flow conditions.

Integrating these fluxes over the year (Table 8) indicated that about 96% of the DIN and 95% of the particulate N entering the system was exported. Of the total, perhaps 5% was retained or lost in some other form (e.g., dissolved organic or gas- eous).

Owing to lack of data for other than dissolved forms of phosphorus and silicon, useful budgets for these nutrients could not be constructed.

Discussion SALINITY AND NUTRIENT

CONCENTRATIONS

Spatial variability of nutrient concentration in the Seekonk-Providence River was at least in part a function of salinity. Excepting nitrate + nitrite and dissolved silicon, correlations with salinity were weak. Low association between salinity and nutrient concentration was not always caused by significant curvature of the relationship but more often


TABLE 7. Net flux of dissolved (DIN = NH,+ + NO,- + NO,-) and particulate nitrogen through the system. Input = contribution from point sources and bottom water from upper Narragansett Bay. For net flux (+) denotes export, (-) denotes retention. Units are moles s-i.

DIN Input

Point source Bottom water

Export Net

Particulate N Input

Point source Bottom water

Export NM

0.X. Dec. Mar. Apr. JIUL Aug.

5.05 8.00 10.58 9.39 6.36 4.58 2.05 1.81 0.30 0.21 0.97 0.70 8.06 10.42 9.85 10.20 4.93 4.25 0.96 0.61 -1.03 0.60 -2.40 -1.03

0.84 1.09 1.71 1.58 0.80 1.12 0.69 0.34 0.64 1.73 0.50 0.90 1.13 1.04 1.73 3.38 2.00 1.61

-0.40 -0.39 -0.62 0.07 0.70 -0.41

by scatter of the data points. The Seekonk-Prov- idence River system is relatively narrow and shal- low. Lateral inputs may be expected to significantly influence observed distributions. Under such conditions, property vs. salinity plots are ineffectual at distinguishing conservative and nonconservative behavior of constituents within the estuary (Fisher et al. 1988).

Box MODEL

We chose to use a box model to facilitate interpretation of nutrient distributions. Since lateral inputs are taken into account, conservative and nonconservative behavior within the estuary can be distinguished.

Salinity in the Seekonk-Providence River did not change gradually, rather there were precipitous changes in surface waters between the Seekonk and Providence rivers (Fig. 4). In bottom waters this discontinuity was displaced somewhat downstream (Fig. 5). Estimating hydrodynamic exchanges from a continuous curve model (e.g., Kaul and Froelich 1984; Smith et al. 1989) did not seem warranted. The changes in pattern of salinity distribution im- ply a step function which can be reasonably represented with a box model (Smith et al. 1989).

The delineation of boxes in the model is thus an important consideration. Boundaries should encompass spatial zones of regular salinity change and separate zones with different patterns of change. Boxes 1 and 2 (stations l-5) of the model separate regions of rapid salinity change in the Seekonk and upper Providence River from the lower Providence River (stations 6-l 0) where changes were more gradual.

The box model is two-dimensional rather than one-dimensional. This two-layer configuration is commensurate with the nearly 100 vertical profiles

of salinity taken during the six cruises. These profiles conform to the characterization given by Pil- son (1985b) for Narragansett Bay and are consistent with the net circulation pattern represented in the model.

PREDICTED AND OBSERVED NUTRIENT CONCENTRATIONS

Nutrient concentrations predicted by the model were all linearly related to observed concentrations. The strength of these relationships, as indicated by the R* values, was generally high, but in some cases (particulate nutrients and ammonia in surface waters) it was not. Given the modest nature of the data used in the model (surface and bottom samples) the statistical significance of the relationships was gratifying and suggests that our data are sufficient to provide a first-order picture of the Seekonk-Providence River region.

Aside from NO*- + N03- (which everywhere behaved according to model prediction) and dissolved silicon in bottom waters, both dissolved and particulate nutrients were depleted in surface waters and enriched in bottom waters relative to the model. Processes responsible for these deviations can only be surmised (Taft et al. 1978), but would appear to be internal to the system, as timing of the cruises minimized potential effects of nonpoint-

TABLE 8. Annual nitrogen budget for the Seekonk-Provi- dence River derived by integrating data in Table 8. Units are lo6 moles N yr-i. Input = point sources + bottom water.

Input Export Percent In lit

Exporte t;

DIN 267 257 96 Particulate N 62 90 Total 329 95


source inputs (rainfall, runoff, etc.) and major point resenting an average condition, There were too sources were measured. few data to dissect these sources of variation.

For dissolved substances in surface waters, pre- cipitation, adsorption (e.g., POdm3, dissolved silicon), and biological fixation into particulate matter may be invoked as explanations (Fanning and Pil- son 1973; Burton and Liss 1976; Aston 19’78; Ka- matani and Takano 1984). It is interesting to note that while NO*- + NO,- behaved as predicted, ammonia did not. The latter is preferred by phytoplankton (McCarthy et al. 1975) and this pro- clivity might explain the contrasting behavior of these two forms of nitrogen.

As a practical matter, the preponderant depen- dence of dissolved nutrient concentrations in the Seekonk-Providence River on external supply suggests that control of point sources would substantially improve water quality in this aleady degraded region. These conclusions, based on field data, agree with earlier manipulative experiments conducted in the MERL mesocosms (Oviatt et al. 1984).

In bottom waters the relative enrichment of PO,-3 and NH,+ may well be explained by remineralization either in the water itself or through flux from sediments fueled by diagenesis. Remineral- ization can significantly affect water column concentrations (Nixon 1981; D’Elia et al. 1983; Pilson 1985a; Anderson 1986).

Both dissolved silicon and nitrate + nitrite concentrations behaved as predicted in bottom waters. Efflux of the latter from sediments in Narragansett Bay is small (Kelly et al. 1985) and ammonia is the initial end product of nitrogen regeneration (Kaul and Froelich 1984). The fact that dissolved silicon was not enriched relative to prediction was sur- prising given the significant flux of this material from Narragansett Bay sediments (Nixon 198 1) and other estuarine sediments as well (D’Elia et al. 1983; Kamatani and Takano 1984).

Both particulate carbon and nitrogen were depleted in surface waters and enriched in bottom waters relative to the model predictions. Sinking from surface layers (e.g., Officer 1980) and resuspension of bottom sediments (Oviatt and Nixon 1975) may account for this pattern.

ESTUARINE FUNCTION AND FLUSHING TIME

The relative contribution of internal and external factors to dissolved nutrient concentrations varies spatially in Narragansett Bay. As shown in this study and an earlier mesocosm experiment (Oviatt et al. 1984), dissolved nutrient concentrations in the Seekonk-Providence River are subject to external control. This condition contrasts with circumstances in lower Narragansett Bay where nutrient cycles may be largely driven by activities internal to the system (Pilson 1985a).

This apparent regional difference in functioning of the bay may in part depend on flushing time. The freshwater in Narragansett Bay is replaced every 26 d (range 10 to 40 d; Pilson 1985b). Es- timates for the Seekonk-Providence River (3-10 d; Spaulding 1987) are considerably shorter. The extent to which a substance can be acted upon before it is washed out of the system depends on the relationship between the chemical or biochem- ical reaction rate and the flushing time (Officer 1980). The intensity of any internal biogeochemical signature should vary inversely with flushing time (Smith et al. 1989) and this appears to be the case for Narragansett Bay.

INTERNAL PROCESSES vs. EXTERNAL INPUTS AND MIXING

Much of the variability in and absolute magni-

NITROGEN BUDGET

tude of dissolved nutrient concentrations could be attributed to external inputs and subsequent mixing with lower Narragansett Bay water. Excepting ammonia in surface waters, the model explained better than 85% (see R* values) of the variability in dissolved nutrient concentrations. For particulate phases, this percentage was substantially lower. The slopes of the predicted on observed regressions indicated that some 60 to 100% of dissolved and 40 to 60% of particulate nutrient concentrations appear to be maintained by external inputs and mixing.

The comparisons between observed and predicted concentrations from which these estimates were made encompass both temporal and spatial variation. As such they should be viewed as rep-

The preliminary nitrogen budget for the See- konk-Providence River indicates that annually about 95% of the total nitrogen (DIN + particulate) entering the system is exported to upper Nar- ragansett Bay. The fate of the unaccounted nitrogen (5%) is unknown. Burial, denitrification, or export as dissolved organic nitrogen represent potential sinks. On an area1 basis the Seekonk-Prov- idence River apparently removes 0.7 1 mole N m-* yr-‘. Nixon et al. (1986) prepared bay-wide budgets for a variety of materials including nitrogen. We did not distinguish between loss processes but our estimate would include both denitrification and burial in the sediments. Nixon et al. (1986) estimate that these two processes account for some 0.55 mole N m-* yr-’ over the whole bay. A somewhat greater area1 removal rate in the Seekonk- Providence River, as compared with the whole bay,


concurs with available data. First, total nitrogen concentrations in the sediments are high in the Providence River and decrease down the bay, in- dicating greater net sedimentation in the Seekonk- Providence River (Oviatt et al. 1984; Seitzinger et al. 1984). Second, although denitrification in sediments was not demonstrably greater than in the rest of the bay, its rate in the Providence River has probably been underestimated (Seitzinger et al. 1984). Either or both of these processes could result in a greater area1 removal rate.

We can now at least provisionally put the See- konk-Providence River in perspective with respect to Narragansett Bay as a whole. This region holds somewhat over 3% of the water yet receives about 60% of the bay’s external nitrogen input. Although the effects of internal processes on dissolved nutrient concentrations are clearly evident, they do not appear as important as in the lower bay. Rather external inputs predominantly control concentrations.

As expected, the apparent loss of nitrogen from burial and denitrification on an area1 basis seems greater then in the rest of Narragansett Bay. Cer- tainly because of its smaller area and perhaps because of its higher flushing rate the Seekonk-Prov- idence River removes some 12% of the total nitrogen annually lost in Narragansett Bay and 3% of the total annual input.

Although considerable dilution of nutrient input occurs in this region, the Seekonk-Providence Riv- er does not buffer lower Narragansett Bay against high nitrogen loading. Under the conditions which we observed, the Seekonk-Providence River acted as a conduit passing nitrogen on to Narragansett Bay.

ACKNOWLEDGMENTS

We thank the captain, M. Gustafson, and crew of the R.V. Lauri Lee. The following were instrumental in completing the project: L. Weber, W. Warren, D. Cullen, C. Brown, E. Klos, E. Requintina, V. Banzon, N. Craig, E. Hoffman, G. Hoffman, and K. Schweitzer. This work was supported by the Rhode Island Department of Environmental Management, Narragan- sett Bay Project.

LITERATURE CITED

ANDERSON, G. F. 1986. Silica, diatoms and freshwater pro- ductivity maximum in Atlantic coastal plain estuaries, Ches- apeake Bay. Estuarine Coastal and Shelf Science 22:183-197.

ASTON, S. R. 1978. Estuarine chemistry, p. 362-440. In J. P. Riley and R. Chester (eds.), Chemical Oceanography, vol. 7, 2nd ed. Academic Press, New York.

BURTON, J. D. AND P. S. LISS. 1976. Estuarine Chemistry. Ac- ademic Press, New York, 229 p.

CHINMAN, R. A. AND S. W. NIXON. 1985. Depth-area-volume relationships in Narragansett Bay. Graduate School of Ocean- ography, University of Rhode Island. NOAA/Sea Grant Ma- rine Technical Report 87. 64 p.

D’ELIA, C. F., D. M. NELSON, AND W. R. BOYNTON. 1983. Chesapeake Bay nutrient and plankton dynamics: III. The annual cycle of dissolved silicon. Geochimica et Cosmochimica Acta 47:1945-1955.

FANNING, K. A. AND M. E. Q. PILSON. 1973. The lack of inorganic removal of dissolved silica during river-ocean mixing. Geochimica et Cosmochimica Acta 37:2405-2415.

FISHER, T. R., L. W. HARDING, JR., D. W. STANLEY, AND L. G. WARD. 1988. Phytoplankton, nutrients and turbidity in the Chesapeake, Delaware, and Hudson estuaries. Estuarine Coast- al and Shelj Science 27161-93.

KAMATANI, A. AND M. TAKANO. 1984. The behaviour of dissolved silica during the mixing of river and sea waters in Tokyo Bay. Estuarine Coastal and ShelfScience 19:505-512.

KAUL, L. W. AND P. N. FROELICH, JR. 1984. Modeling estuarine nutrient geochemistry in a simple system. Geochimica et Cos- mochimica Acta 48~1417-1433.

KULY, J. R., V. M. BEROUNSKY, S. W. NIXON, AND C. A. OVIA~. 1985. Benthic-pelagic coupling and nutrient cycling across an experimental eturophicationgradient. Marink Ecol&y Prog- ress Series 26:207-219.

KREMER, J. N. AND S. W. NIXON. 1978. A Coastal Marine Ecosystem, Simulation and Analysis. Springer-Verlag, New York. 217 p.

LAMBERT, C. E. AND C. A. OVIATT (EDS.). 1986. Manual of biological and geochemical techinques in coastal areas. Grad- uate School of Oceanography, University of Rhode Island, Narragansett, Rhode Island. MERL Series. Reoort No. 1. 2nd ed. 319 p.

I

MACDONALD, R. W., F. A. MCLAUGHLIN, AND C. S. WONG. 1986. The storage of reactive silicate samples by freezing. Limnology and Oceanography 3 1: 1139- 114 1.

MCCARTHY, I. I., W. R. TAYLOR, AND I. L. TAFT. 1975. The dynamics of-nitrogen and phospho&s cycling in the open waters of the Chesaoeake Bav. D. 664-681. In T. M. Church (ed.), Marine Chemistry in &id Coastal Environment. ACS Symp. Ser. No. 18. American Chemical Association, Wash- ington, D.C.

NIXON, S. W. 1981. Remineralization and nutrient cycling in coastal marine ecosystems, p. 111-138. In B. J. Neilson and L. E. Cronin (eds.), Estuaries and Nutrients. Humana Press, Clifton, New Jersey.

NIXON, S. W. 1987a. Chesapeake Bay nutrient budgets-A reassessment. Biogeochemistry 4:77-90.

NIXON, S. W. 1987b. Overview, p. 7-16. In NOAA Estuary of the Month Seminar No. 1. Narragansett Bay: Issues, Re- sources, Status and Management. NOAA Estuarine Program Office, Washington, D.C.

NIXON, S. W., C. D. HUNT, AND B. L. NOWICKI. 1986. The retention of nutrients (C, N, P), heavy metals (Mn, Cd, Pd, Cu) and petroleum hydrocarbons in Narragansett Bay, p. 99- 122. In J. M. Martin and P. Lasserre (eds.), Biogeochemical Processes at the Land-Sea Boundary. Elsevier Press, Amster- dam.

NIXON, S. W. AND V. LEE. 1979. Spatial and temporal pollution gradients in Narragansett Bay, p. l- 13. International Council for the Exploration of the Sea, Water Quality Committee. Warsaw, Poland. CM. 1979/E:56.

OFFICER, C. B. 1980. Box models revisisted, p. 65-l 14. In P. Hamilton and K. B. Macdonald (eds.), Estuarine and Wetland Processes with Emphasis on Modeling. Plenum Press, New York.

OVIATT, C. A. 1981. Some aspects of water quality in and pollution sources to the Providence River. Report for Region 1, Environmental Protection Agency. Sept. 1979-Sept. 1980. The Marine Ecosystem Research Laboratory, Graduaie School of Oceanoaranhv, Universitv of Rhode Island, Kinrrston, Rhode Islaid.’ p. ’

”

OVIATT, C. A. AND S. W. NIXON. 1975. Sediment resuspension


and deposition in Narragansett Bay. Estuarine and Coastal Marine Science 3:201-21’7.

OVIATT, C. A., M. E. Q. PILSON, S. W. NIXON, J. B. FRITHSEN, D. T. RUDNICK, J. R. KELLY, J. F. GRASSLE, AND J. P. GRASSLE. 1984. Recovery of a polluted estuarine system: A mesocosm experiment. Marine Ecology Progress Series 16:203-2 17.

PILSON, M. E. Q. 1985a. Annual cycles of nutrients and chlo- rophyll in Narragansett Bay, Rhode Island.Journal ofMarine Research 43:849-873.

PILSON, M. E. Q, 1985b. On the residence time of water in Narragansett Bay. Estuaries 8:2-14.

PRITCHARD, D. W. 1969. Dispersion and flushing of pollutants in estuaries. American Society of Civil Engineering, Journal of the Hydraulics Division 95, No. Hyl:l15-125.

RICKER, W. E. 1973. Linear regressions in fishery research. Journal of the Fisheries Research Board of Canada 30:409-434.

SEITZINGER, S. P., S. W. NIXON, AND M. E. Q. PILSON. 1984. Denitrification and nitrous oxide production in a coastal marine ecosystem. Limnology and Oceanography 29:73-83.

SMITH, S. V., J. T. HOLLIBAUGH, S. J. DOLLAR, AND S. VINK. 1989. Tomales Bay, California: A case for carbon-controlled nitrogen cycling. Limnology and Oceanography 34:37-52.

SMULLEN, J. T., J. TAFT, AND J. MACKORIS. 1982. Nutrient and sediment loads to the tidal Chesapeake Bay system, p. 147- 262. In E. G. Macalester, D. A. Barker, and M. Kaspter (eds.), Chesapeake Bay Program Technical Studies: A Synthesis. U.S. Environmental Protection Agency, Washington, D.C.

SPAULDING, M. 1987. Circulatian dynamics, p. 71-146. In NOAA Estuary of the Month Seminar No. 1, Narragansett Bay: Issues, Resources, Status and Management. NOAA Es- tuarine Program Office, Washington, D.C.

TAFT, J. L., A. J. ELLIOTT, AND W. R. TAYLOR. 1978. Box model analysis of Chesapeake Bay ammonium and nitrate fluxes, p. 115-130. In M. L. Wiley (ed.), Estuarine Interac- tions. Academic Press, New York.

THURMAN, H. V. 1985. Introductory Oceanography. 4th ed. Merrill Publishing Co., Columbus, Ohio. 503 p.

TIPPIE, V. K. 1984. An environmental characterization of Chesapeake Bay and a framework for action, p. 467-488. In V. S. Kennedy (ed.), The Estuary as a Filter. Academic Press, New York.

WEISBERG, R. H. AND W. STURGES. 1976. Velocity observations in the West Passage of Narragansett Bay: A partially mixed estuary. Journal of Physical Oceanography 6:345-354.

WOLLAST, R. 1983. Interactions in estuaries and coastal waters, p. 385-407. In B. Bolin and R. B. Cook (eds.), The Major Biogeochemical Cycles and Their Interactions. John Wiley and Sons, New York.

Received for consideration, July 20, 1989 Accepted for publication, December 18, 1989

control of nutrient concentrations in the seekonk ......(qm + e,,) lower layer: predicted c ’ , =...

Documents