Estuarine, Coastal and Shelf Science (2002) 55, 705–717doi:10.1006/ecss.2001.0922, available online at http://www.idealibrary.com on

Spatial and Temporal Distribution of ColouredDissolved Organic Matter (CDOM) in NarragansettBay, Rhode Island: Implications for Phytoplankton inCoastal Waters

D. J. Keitha, J. A. Yoderb and S. A. Freemanb

aUnited States Environmental Protection Agency, Atlantic Ecology Division, Narragansett, RI 02882, U.S.A.bGraduate School of Oceanography, University of Rhode Island, Narragansett, RI 02882, U.S.A.

Received 7 June 2001 and accepted in revised form 31 August 2001

One indicator of health in estuarine and coastal ecosystems is the ability of local waters to transmit sunlight to planktonic,macrophytic, and other submerged vegetation for photosynthesis. The concentration of coloured dissolved organic matter(CDOM) is a primary factor affecting the absorption of incident sunlight in coastal and estuarine waters. In estuaries,CDOM concentrations vary due to changes in salinity gradients, inflows of industrial and domestic effluents, and theproduction of new dissolved organic matter from marine biologic activity. CDOM absorption data have been collectedfrom a variety of waters. However, there are a limited number of measurements along the US east coast and a general lackof data from New England waters.

This study characterized the temporal and spatial variability of CDOM absorption over an annual cycle in NarragansettBay and Block Island Sound (Rhode Island). Results suggested that, in Narragansett Bay, the magnitude of CDOMabsorption is related to the seasonal variability of freshwater input from surrounding watersheds and new CDOMproduction from in situ biologic activity. The data show that the average CDOM absorption coefficient at 412 nm was0·45 m�1 and the average spectral slope was 0·020 nm�1. � 2002 Elsevier Science Ltd. All rights reserved.

Keywords: coloured dissolved organic matter; remote sensing; phytoplankton; chlorophyll a

Introduction

Coastal and estuarine waters are optically complexowing to high concentrations of aquatic constituentsaffecting the inherent properties of the water column(i.e. absorption and backscattering processes) whichchange the intensity and spectral character of incidentsunlight.

The ability of coastal and estuarine waters to trans-mit sunlight affects phytoplankton productivity as wellas submerged aquatic vegetation (SAV). Changes inlight transmissivity may be due to natural and anthro-pogenic factors. Nutrient enrichment from point andnon-point sources can stimulate water column pro-duction which reduces light transmission to SAV.Monitoring the optical properties of local waters atseveral spatial and temporal scales could providediagnostic information on natural and anthropogenicfactors affecting the capacity of these waters to providesufficient sunlight to planktonic and macrophyticvegetation for photosynthesis and growth.

The biological and optical characteristics of coastaland estuarine waters are complicated by temporal and

0272–7714/02/110705+13 $35.00/0

spatial variability in the concentration of coloureddissolved organic matter (CDOM) and suspendedparticulate matter (SPM), the spectral properties ofpure water, and the composition of phytoplanktonphotosynthetic pigments. CDOM (Gelbstoff or gilvin)of terrestrial origin is primarily composed of humicacids produced from the decomposition of plantlitter and organically rich soils within coastal water-sheds and upland areas. Humic acids are moderatemolecular weight, yellow-coloured organic acids thatdominate the absorption of visible and UV light at theblue end of the spectrum (Kirk, 1996; McKnightet al., 2001). From a beneficial standpoint, the strongabsorption of CDOM in the UV portion of thespectrum protects phytoplankton and other biotafrom damaging UVB radiation (Blough & Zepp,1990; Blough & Green, 1995). However at increasinglevels, CDOM absorption can affect primary produc-tivity and ecosystem structure by reducing the amountand quality of photosynthtically active radiation tophytoplankton (Bidigare et al., 1993).

CDOM concentrations also increase in coastalwaters due to the in situ creation of fulvic acids

� 2002 Elsevier Science Ltd. All rights reserved.

706 D. J. Keith et al.

produced from the seaweed decomposition (Sieburth& Jensen, 1969), as a by-product of primary pro-duction stimulated by nutrients (Carder et al., 1989;Del Castillo et al., 2000), and the anthropogenic inputof industrial or domestic effluents from populatedareas (Bricaud et al., 1981). In the coastal environ-ment, the optical properties of CDOM change owingto seawater mixing and photodegradation (Bricaudet al., 1981; Morel, 1988; Carder et al., 1989;Vodacek et al., 1997; Del Castillo et al., 2000). Theeffects of CDOM on the accuracy of remote sensingalgorithms to predict chlorophyll a distribution andphytoplankton biomass estimates in coastal areas hasbeen previously discussed (e.g. Carder et al., 1989;IOCCG, 2000).

A review of the published literature indicates thatCDOM absorption data have been globally col-lected from a variety of geographically diverse waters(Kirk, 1996). Along the US east coast, CDOMmeasurements have been made from continental shelfwaters of the Middle Atlantic Bight (Vodacek et al.,1997) and the Gulf of Maine (Yentsch & Phinney,1997). However, there is a lack of CDOM absorptiondata from estuaries in New England. NarragansettBay, a well-mixed mesotidal estuary, is an excellentsite to study the variation of CDOM in the coastalenvironment. Nutrients are input into the upper bayby freshwater streams and rivers which drain theBlackstone Valley and Narragansett watersheds. Inaddition, a distinct salinity gradient exists fromthe brackish waters of the Providence River to themore oceanic waters of Block Island Sound (Hess,1976).

In this manuscript, we examined the behaviour ofCDOM along coastal New England to answer thefollowing questions:

(1) What is the temporal and spatial variation in theabsorption of light by CDOM in the NarragansettBay (Rhode Island) system?

(2) What is the relationship between the variability inCDOM absorption, seasonal freshwater inflowfrom surrounding watersheds, and salinity differ-ences from upper Narragansett Bay into BlockIsland Sound?

(3) What are possible ecological implications for phy-toplankton photosynthesis in CDOM-dominatedwaters?

To answer these questions, we relied on laboratoryanalyses of CDOM absorption coefficients andchlorophyll a concentrations as well as field measure-ments of freshwater flow and salinity variation inNarragansett Bay.

Materials and methods

Surface measurements of CDOM absorption coef-ficients and chlorophyll a concentrations were madefrom water samples collected during May 1999through June 2000. Samples were partitioned forlaboratory analysis from a 5 l Go-Flo bottle lowered to4 m depth at four stations along the Western Passageof Narragansett Bay, Rhode Island, U.S.A. (Figure 1;Table 1). Salinity measurements were made at eachstation using a Seabird Sealogger CTD SBE 25 pro-filer also lowered to a depth of 4 m averaged from thesurface to depth.

The stations were located along the salinity gradientthat is present from Rhode Island Sound into theProvidence River. During the sampling period, 32weekly to biweekly cruises were primarily madeonboard the RV Bay Challenger, RV Coastal Explorer,and RV Arbacia which are based at the USEPA/Atlantic Ecology Division Laboratory (Narragansett,RI). On two of these cruises, data were collected fromUS Coast Guard (USCG) patrol boats based atUSCG Station–Castle Hill (Newport, RI) . The upperbay stations were chosen due to their location relativeto river and stream sources.

Daily freshwater discharge data (May 1999–July2000) for 12 streams and rivers which flow intoNarragansett Bay from central Massachusetts andnorthern Rhode Island were obtained from theUSGS Water Resources Division in Northborough,Massachusetts and averaged into weekly (7 day)estimates of freshwater discharge.

Water for CDOM absorption analysis was collectedin 250 ml bottles and placed on ice while in thefield. In the laboratory, these samples were held inrefrigerated storage at 4 �C until analysed.

Water samples, reference samples, and sampleblanks were twice filtered using a 50 ml syringe filterholder and 0·22 micron Millipore filters. The refer-ence sample was stored in a 10 cm (0·1 m) cuvettethat was refrigerated at 4 �C until needed. A sampleblank of deionized water was included during eachanalysis and its spectral values substracted fromthe sample measurements to correct the absorbancevalues for offsets between the deionized water refer-ence and the seawater sample. CDOM absorptioncoefficients were determined from laboratorymeasurements of the optical density (OD) of the watersamples using a Perkin Elmer Lambda 3 Double-Beam UV-Visible Spectrophotometer with a spectralresolution of 0·5 nm using 10 cm pathlength cells.The OD was determined using:

OD=log10(Io/I) (1)

Distribution of CDOM in Narragansett Bay 707

41.5°

71.2°

41.7°

Longitude (W)

Lat

itu

de (

N)

71.4°

41.6°

71.3°

Rhode Island Sound7

0 42Kilometres

N

4

USEPAAtlanticEcologyDiv.Lab

BeavertailPoint

BonnetPoint

QuonsetPoint

40

CO

NA

NIC

UT

ISL

AN

D

PR

UD

EN

CE

ISL

AN

D

AQUIDNECKISLAND

Newport

36Mt. Hope

Bay

NARRAGANSETTBAY

Providence R

iver

RockyPoint

Greenwich Bay

West P

assage

East Passage

F 1. Map of sampling locations along the West Passage of Narragansett Bay and Block Island Sound. Absorptionspectra were collected for the surface waters at each location over a 36 week period from May 1999–June 2000. Samples werecollected during cruises of the RV Bay Challenger, RV Coastal Explorer, and RV Arbacia.

708 D. J. Keith et al.

T 1. Sampling locations with average CDOM absorption coefficients, with standard deviations, and spectral slopes

Station Latitude Longitudea412

(m�1) (average)S

(m�1) (average) Location

7 41�26.38N 71�22.99W 0·286 (�0·132) 0·020 (�0·012) Block Island Sound (SW of Brenton Reef)4 41�28.80N 71�24.40W 0·336 (�0·134) 0·021 (�0·014) Lower West Passage (NE of Bonnet Point)40 41�35.00N 71�23.30W 0·500 (�0·199) 0·021 (�0·011) Middle West Passage (E of Quonset Point)36 41�40.90N 71�20.70W 0·690 (�0·223) 0·017 (�0·007) Upper Narragansett Bay (SE of Rocky Point)

where Io is the absorbance of the water sample and I isthe absorbance of a reference sample, in this casedeionized water. CDOM absorption was calculatedusing Equation 2:

a(�)=(2·303) [(ODs(�)�ODb(�))/l](Mitchell et al., 2000) (2)

where ODs(�)=sample absorbance, ODb(�)=sampleblank absorbance, and l=cuvette pathlength.

Absorption coefficients were corrected for back-scattering of small particles and colloids which passthrough filters using Equation 3 (Bricaud et al., 1981;Green & Blough, 1994).

a(�)corr=a(�)�a700 (�/700) (Bricaud et al., 1981) (3)

where a(�)corr=absorption at a given wavelength (�)corrected for scattering.

a(�)=measured absorption at a given �

a700=measured absorption at 700 nm

Corrected absorption coefficients were calculatedat 400, 412, 440, 443,490, 510, 555, 615, 670, 675,and 700 nm in units of m�1. The corrected absorp-tion coefficients were plotted to yield a spectralabsorption curve for the range of 400 to 700 nm.The exponential (or spectral) slope (S), of eachspectral absorption curve was determined by takingthe natural logarithm (LN) of the corrected absorp-tion coefficients (a) and regressing it against wave-length between 400–550 nm. The a412 coefficient isconsistently referred to in this paper due to itscorrespondance to Band 1 of the SeaWiFS andMODIS ocean colour sensors.

The general CDOM spectrum for Narragansett Baywas produced by calculating an average absorptioncoefficient per wavelength using data from all sam-pling stations. The spectral slope of the generalabsorption curve was derived by regressing the naturallogarithm of the corrected absorption against wave-length from 400–550 nm. The spatial variation in

CDOM absorption was determined by calculating anaverage absorption coefficient per wavelength usingdata from individual sampling stations. The spectralslope data were averaged (Savg) for all stations toexamine temporal variations in S.

Because upper Narragansett Bay is heavily influ-enced by the freshwater discharge and organic input ofthe Providence River, the amount of ambient lightabsorbed in the upper bay by dissolved organic matter(DOM) was determined from the CDOM absorptionmeasurements. Using a Biospherical InstrumentsPRR 600 Series profiling reflectance radiometer, thetotal downwelling irradiance (Ed) was collected fromthe upper 0·5 m of the water column at Station 36 at412, 443, 490, 510, 555, and 665 nm.

Chlorophyll concentrations were determinedusing Equation 4 based on phytoplankton pig-ment fluorescence using the Welschmeyer protocol(Welschmeyer, 1994). The Welschmeyer protocolallows for the non-acidic determination of chl a in thepresence of chl b and pheopigments. The proceduresfrom EPA Method 445.0 were used during the prep-aration of standard solutions, sample collection,storag, instrument performance demonstrations, andquality assurance (Arar & Collins, 1992). Chlorophyllpigments were extracted in a 90% acetone solution inaccordance with Prezioso et al. (1999) which involvesthe immersion of sample filters for 12–24 h in glasscuvettes as compared with the destruction of filtersusing a tissue grinder as required in Method 445.0.Seawater was collected in the 5 l Go-Flo bottle andtransferred to 1 l darkened glass bottles. In the field,water samples (50 ml volume) were immediately pro-cessed after collection and filtered, in subdued light,using 47 mm Whatman GF/F filters using a vacuumpump. If the samples could not be processed in thefield, water samples were placed on ice and filtered ina laboratory within 4 h of collection. Filters wereplaced in aluminum foil pouches and stored at�80 �C until needed for analysis. The fluorescencesignature of the samples was determined with aTurner Designs Model AU-10 Digital Fluorometerequipped with the Non-Acidification Optical Kit (P/N

Distribution of CDOM in Narragansett Bay 709

10-040R). The Non-Acidification Optical Kit consistsof a blue mercury vapour lamp which exciteschlorophyll a at 436 nm and narrow band interferencefilters which only allow the specific excitation at436 nm and emission at 680 nm wavelengths of chl ato pass. Chlorophyll concentrations were determinedusing:

Chl a (�g/l)=KFov/Vf (UNESCO, 1997) (4)

where K=0·998=fluorescence sensitivity coefficientin extraction solvent; Fo=fluorescence reading;v=volume of acetone used for extraction; Vf=volumeof seawater filtered.

Chlorophyll-specific absorptions (achl) inNarragansett Bay were estimated at 400, 412, 440,443, 465, 490, 510, 550, 555, 615, 670, 675, and700 nm using the measured chlorophyll concen-trations; and the spectral absorption coefficients(ac*�(�)) of Prieur and Sathyendranath (1981)(Equation 5).

achl (�) (m�1)=(a*(�))(Chl a concentration)0·65 (5)

where a* (�)=0·06 (ac*�(�)) (Mobley, 1998, 1999).The functional relationships between CDOM

absorption and several parameters (e.g., salinity, tem-perature) were statistically examined using correlationand the Major Axis method of Model II linear regres-sion analysis (Kermack & Haldane, 1950; Laws,1997).

Results

0.0700

1.0

Wavelength (nm)

Abs

orpt

ion

(m

–1)

400

0.9

650600550500450

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

F 2. Average CDOM and chlorophyll-specific absorption spectra for Narragansett Bay. Average CDOM absorption;Average chlorophyll-specific absorption.

0.0700

1.0

Wavelength (nm)

CD

OM

Abs

orpt

ion

(m

–1)

400

0.8

650600550500450

0.6

0.4

0.2

F 3. Spatial variability in CDOM absorption alongWest Passage–Narragansett Bay, RI. Station 36; Station40; � Station 4; � Station 7.

Spectral character of CDOM in Narragansett Bay

The general spectral curve for CDOM in NarrgansettBay shows near zero absorption values, at the redportion of the visible spectrum (700 nm), whichexponentially increase to maximum values in near-ultraviolet (UV) wavelength regimes (350–400 nm)(Figure 2). The average CDOM absorption coef-ficient at 412 nm was 0·45 m�1 (Figure 2). AverageCDOM absorption (a412) coefficients for each stationare found in Table 1 and Figure 3. Individual CDOM

710 D. J. Keith et al.

T 2. CDOM absorption spectral slopes between May 1999–June 2000

Cruise date Station 7 Station 4 Station 40 Station 36 Savg

21 May 1999 0·022 0·025 0·015 0·013 0·01927 May 1999 0·010 0·012 0·012 0·011 0·0118 June 1999 0·024 0·027 no data 0·019 0·02316 June 1999 0·013 0·011 0·016 0·013 0·01324 June 1999 0·012 0·010 0·015 0·014 0·01330 June 1999 0·027 0·011 0·023 0·015 0·01915 July 1999 0·012 0·010 0·011 0·013 0·01221 July 1999 0·020 0·027 0·013 0·016 0·01929 July 1999 0·008 0·013 0·014 0·018 0·0134 August 1999 0·016 0·029 0·011 0·015 0·01820 August 1999 0·031 0·014 0·034 0·022 0·0258 October 1999 0·009 0·024 0·027 0·014 0·01915 October 1999 0·013 0·015 0·014 0·012 0·01421 October 1999 0·011 0·013 0·010 0·013 0·01227 October 1999 0·011 0·028 0·012 0·014 0·01629 November 1999 0·036 0·028 0·015 0·022 0·0258 December 1999 0·028 0·011 0·022 0·034 0·02413 December 1999 0·033 0·034 0·021 0·016 0·0266 January 2000 0·018 0·018 0·021 0·039 0·0249 February 2000 no data 0·063 0·017 0·024 0·03517 February 2000 0·013 0·019 0·020 0·020 0·0186 March 2000 0·028 0·023 0·020 0·028 0·02513 March 2000 0·027 0·068 0·037 0·022 0·03824 March 2000 0·012 no data 0·012 0·012 0·01230 March 2000 no data 0·014 0·028 0·015 0·01913 April 2000 no data 0·044 no data 0·024 0·0343 May 2000 0·006 0·018 0·008 0·008 0·01016 May 2000 0·014 0·008 0·013 0·013 0·01226 May 2000 0·013 0·011 0·014 0·013 0·01331 May 2000 no data no data 0·014 0·013 0·0148 June 2000 0·006 0·007 0·009 0·011 0·00821 June 2000 0·046 0·014 0·041 0·020 0·019

absorption values (at 412 nm) in the entire datasetranged from a minimum of 0·09 to a maximum of1·4 m�1. Spectral slopes (R2 values of 0·74 to 0·99)for each station/sampling date varied from 0·006 to0·068 nm�1 (Table 2). The average value of S alongWest Passage during the study period was 0·020(�0·011) nm�1 (Table 1). Correlation analysisindicated that a moderately strong, inverse relation-ship existed between S and CDOM absorption(r= �0·42). This inverse relationship with CDOMhas also been observed by D’Sa et al. (1999). Vodaceket al. (1997) indicated that when terrestrial-derivedCDOM is present in surface waters (under conser-vative mixing conditions), S�0·020 nm�1. Increasedvalues of S are thought to arise from the transfor-mation of terrestrial-derived CDOM and (or) itsreplacement by CDOM generated in situ as wellas changes in the salinity and stratification ofNarragansett Bay at the time of sample collection(Carder et al., 1989; Vodacek et al., 1997). Carderet al. (1989) indicated that, in Gulf of Mexico waters,

CDOM samples composed primarily of fulvic acidshad S values that were nearly twice as large as thosecomposed primarily of humic acids.

The highest absorption values (Figure 3) consist-ently occur along the middle reach of West Passage[Station 40 (Quonset Pt.)] into upper NarragansettBay [Station 36 (Rocky Pt.)]. The amount of ambientlight absorbed by CDOM at Station 36 was highest(0·004 �W m3 nm�1) in the blue portion of the spec-trum and steadily decreased with increased wave-length to zero at the red range. However, in theblue-green part of the spectrum, CDOM absorptionvalues noticibly decreased suggesting increased avail-ability of light energy between 500–540 nm (Figure4). The lowest absorption values (Figure 3) consist-ently occur along the lower West Passage [Station 4(Bonnet Pt.)] into Block Island Sound [Station 7(Brenton Pt.)].

CDOM absorption coefficients (a412) showedseasonal cycles (Figure 5). At all stations, highestabsorptions usually occurred during late spring and

Distribution of CDOM in Narragansett Bay 711

lowest absorptions occurred during the winter/earlyspring season. In upper Narragansett Bay, highestabsorption values occurred between May and June.Lowest values occurred in the lower bay and RhodeIsland Sound between February–April (Figure 5).

Temporal and spatial variation in CDOM absorptionand its relation to salinity

In Narragansett Bay, CDOM absorption showed aninverse relationship (r= �0·72) with surface salinity(Figure 6). This relationship has been observed atother coastal locations and reported by others (seeMonahan & Pybus, 1978; Bricaud et al., 1981; Fox,1983; Pegau, 1997; Yentsch & Phinney, 1997). CTDmeasurements indicate that lowest salinities (<27–28 ppt) consistently occur in upper Narragansett Bayand steadily increased to approximately 33 ppt inRhode Island Sound.

Del Castillo et al.(1999) indicated that for OrinocoRiver discharge, S was independent of salinity for

waters with values lower than 30 ppt. They suggestedthat this independence implies there is no significantchange in the optical properties and chemical com-position of the riverine organic matter within thissalinity range. For Narragansett Bay, plotting S versussalinity indicated that there was no dependence of Son salinity from 26–32 ppt (Figure 7).

0.000700

0.004

Wavelength (nm)

Ed*

a CD

OM

(µW

m–3

nm

–1)

400

0.002

650600550500450

F 4. Ambient light absorption by CDOM in upperNarragansett Bay at Station 36.

1.40

1.20

0.00

Month

Spring

a 412

nm

(m

–1)

May

99

Jun

99

Jul 9

9

Au

g 99

Oct

99

Nov

99

Jan

00

Feb

00

Jun

00

Mar

00

Apr

00

1.00

0.80

0.60

0.40

0.20

Summer Fall Winter Spring

Sep

99

Dec

99

May

00

F 5. Seasonal variation in CDOM absorption at 412 nm for each sampling station based on monthly averages.Station 7; Station 4; Station 40; Station 36.

0.034.0

1.6

Salinity (ppt)

a 412

(m

–1)

26.0

1.4

32.030.028.0

1.2

1.0

0.8

0.6

0.4

0.2

r = –0.72n = 100

F 6. Relationship between salinity and CDOMabsorption in Narragansett Bay based on Model IIregression analysis.

Variation in CDOM absorption and freshwater inputinto Narragansett Bay

The highest average CDOM absorptions consistentlyoccur during the Spring 1999 and 2000 samplingseasons (Figures 3 and 5). To examine the relation-ship between these seasonal increases and the springfreshet, correlation and Model II regression analyseswere conducted using the weekly discharge of fresh-water into the bay system and average CDOMabsorption coefficients from upper Narragansett Bay(Stations 36 and 40).

712 D. J. Keith et al.

Freshwater is delivered into the bay system fromthe Narragansett Basin watershed, a mainly forestedwatershed which covers approximately 3520 km2

in north-central Rhode Island and easternMassachusetts, and the Blackstone Valley watershed,a forested watershed which covers approximately1170 km2 in central Massachusetts and northernmostRhode Island). From May 1999–July 2000, thecombined flow of twelve rivers and streams fromthese watersheds varied from approximately 60–80 million m3 over a 7 day period, during early-midspring, to less than 10 million m3 per 7 days duringthe summer months (Figure 8). CDOM absorption inthe upper bay varied from 0·3–1·2 m�1 during thisperiod. Regression analysis indicated that CDOMabsorption was strongly correlated with freshwaterdischarge during mid-spring through summer (r=0·88and 0·68) and in the autumn (r=0·79). Weaker

correlations with CDOM absorption were foundduring winter and early spring seasons (r=0·22,r= �0·19, respectively). Correlation analysis indi-cated a weak relationship between S and freshwaterinflow (r=0·20).

0.00034.0

0.080

Salinity (ppt)

S (

m–1

) (4

00–5

50 n

m)

26.0

0.070

33.032.031.030.029.028.027.0

0.060

0.050

0.040

0.030

0.020

0.010

F 7. Relationship between salinity and CDOMspectral slope in Narragansett Bay.

0

100 000 000

Time (week of)

Fre

shw

ater

inpu

t (m

3 wee

k–1)

0.0

1.0

CD

OM

abs

orpt

ion

(m

–1)

80 000 000

22 M

ay 9

9

13 J

un

99

11 J

ul 9

9

01 A

ug

99

31 O

ct 9

9

19 J

un

00

06 D

ec 9

9

60 000 000

40 000 000

20 000 000

0.90.80.70.60.50.40.30.20.1

22 M

ay 0

0

10 A

pr 0

0

13 M

ar 0

0

07 F

eb 0

0

10 O

ct 9

9

F 8. Freshwater input and CDOM variability in Narragansett Bay during May 1999–June 2000.

Relationship between CDOM and chlorophyll a

The distribution and variability in chlorophyll aconcentration is often used as an index of seasonal andregional abundance and bloom dynamics (Li &Smayda, 1998). In Narragansett Bay, chlorophyll aseasonally varies with the highest concentrationscharacteristically occuring in the upper reaches of thesystem (Pratt, 1959; Farmer et al., 1982; Li &Smayda, 1998). One main feature of the chlorophylldistribution in Narragansett Bay is the occurrence ofintense winter–spring diatom blooms (>20�106

cells l�1) which usually begin in December and end inlate March (Pratt, 1959, 1965; Hitchcock & Smayda,1977; Smayda, 1998). During this study, in the upperbay (at Stations 36 and 40) peak concentrationsranged from 19–32 �g l�1 (Table 3). In the lower bay(Stations 4 and 7), peak chlorophyll concentrationsranged from 8–11 �g l�1 (Table 3).

The relationship between chlorophyll a concen-tration and S has been used in the Middle AtlanticBight to determine the link between increased chl alevels and the production of in situ CDOM (Vodaceket al., 1997). For Narragansett Bay, this link wastested by plotting the average chlorophyll a concen-tration against S(avg) for each cruise date (Figure 9).These results showed that between May 1999–June2000, there were several periods during whichincreases in S directly coincided with increases in

Distribution of CDOM in Narragansett Bay 713

T 3. Chlorophyll a concentrations (�g l�1) between May 1999–June 2000

Cruise date Station 7 Station 4 Station 40 Station 36 Chlavg

21 May 1999 0·94 2·20 2·22 20·30 6·4127 May 1999 1·45 2·10 3·24 25·00 7·958 June 1999 2·12 3·80 no data 32·80 12·8916 June 1999 1·10 2·40 7·20 24·20 8·7324 June 1999 2·46 5·10 3·61 3·58 3·7030 June 1999 2·02 4·10 15·00 6·29 6·8615 July 1999 1·39 2·50 7·11 8·62 4·8921 July 1999 4·28 6·20 8·45 14·90 8·4529 July 1999 2·46 6·10 8·34 13·74 7·654 August 1999 2·58 5·20 4·35 10·09 5·5620 August 1999 1·43 8·50 10·74 26·13 11·698 October 1999 3·40 4·40 5·20 4·30 4·3315 October 1999 4·80 7·30 19·20 4·40 8·9321 October 1999 4·20 11·0 7·10 6·40 7·1827 October 1999 2·60 7·30 16·70 14·20 10·2029 November 1999 3·30 4·30 4·90 2·40 3·738 December 1999 3·70 5·90 8·40 2·80 5·2013 December 1999 4·50 5·30 8·70 5·30 5·956 January 2000 3·90 6·30 15·30 4·40 7·489 February 2000 no data 6·30 11·1 22·50 13·3017 February 2000 7·10 4·50 9·00 31·00 12·906 March 2000 5·50 5·90 9·60 16·00 9·2513 March 2000 3·30 5·80 12·40 28·80 12·5824 March 2000 1·60 no data 11·20 12·30 8·3730 March 2000 no data 1·30 1·20 6·90 3·1313 April 2000 no data 4·40 no data 17·70 11·053 May 2000 3·50 6·00 14·70 6·70 7·7316 May 2000 8·30 7·10 14·40 31·70 15·3826 May 2000 2·50 3·20 7·90 14·90 7·1331 May 2000 no data no data 8·60 4·60 6·608 June 2000 2·60 3·50 11·30 9·50 6·7321 June 2000 2·50 6·60 17·50 10·5 9·27

Chl a concentration. Correlation analysis of these dataindicated a moderate relationship (r=0·32) betweenchl a and S.

The chlorophyll a concentration data collected inour study were used to derive a chlorophyll-specificabsorption curve based on spectral coefficients (a*) ofPrieur and Sathydranath (1981) at 13 wavelengths.The mean chlorophyll-specific absorption curve forNarragansett Bay is shown in Figure 2.

Phytoplankton absorptions were then compared toCDOM absorptions at 443 and 490 nm using theratio of aCDOM (�)/(aCDOM (�)+aChl (�)). 443and 490 nm were chosen because of their locationrelative to the Chl a absorption maximum and theirroles as the primary wavebands used to calculate Chla in remote sensing algorithms (O’Reilly et al., 1998).490 nm also is a wavelength associated with absorp-tion by caroteniods (Stuart et al., 1998). Since theupper bay consistently has the highest absorptionvalues, a time series of normalized CDOM absorptionis presented only for Station 36 (Figure 10).

In general, these data showed that the competitionfor photons at these wavelengths was highly variablethroughout the year. In Narragansett Bay, the highlyvariable nature of chlorophyll and possibly, CDOMconcentrations may be the product of in situ regener-ation of dissolved organic nitrogen by marine bottomcommunities (Nixon et al., 1976).

During late spring–early summer, our results indi-cate that CDOM is responsible for as much as 70–80% of the total absorption at 443 and 490 nm in thewater column (Figure 10). These absorption estimatesare consistent with an earlier study by Yentsch andPhinney (1997) in which they suggest that 80–90% ofthe total absorption at short wavelengths duringspringtime in Gulf of Maine waters was due to absorp-tion by CDOM. These estimates also indicate thatCDOM absorption is reduced to 20–40% of lightenergy during diatom blooms during the winter/springtransition and summer months (Figure 10).

Using spectral data of Puget Sound phytoplanktonspecies (Culver & Perry, 1999), we attempted to

714 D. J. Keith et al.

0

18

Ch

l (av

g) (

µg l–1

)

0.000

0.045

S (

avg)

16

21 M

ay 9

9

16 J

un

99

15 J

ul 9

9

04 A

ug

99

15 O

ct 9

9

29 N

ov 9

9

06 J

an 0

0

06 M

ar 0

0

08 J

un

00

30 M

ar 0

0

16 M

ay 0

0

0.040

0.035

0.030

0.025

0.020

0.015

0.010

0.005

14

12

10

8

6

4

2

F 9. Relationship between the average CDOM spectral slope (Savg) and average Chl a concentration duringMay 1999–June 2000. Chl average; Average slope.

0.0

1.0

a CD

OM

/aC

DO

M +

aC

hl 0.8

21 M

ay 9

9

21 J

un

99

21 J

ul 9

9

21 A

ug

99

21 O

ct 9

9

21 N

ov 9

9

21 F

eb 0

0

21 M

ar 0

0

21 J

un

00

21 A

pr 0

0

21 M

ay 0

0

21 S

ep 9

9

21 D

ec 9

9

21 J

an 0

0

0.6

0.4

0.2

F 10. Normalized CDOM absorption time series at 443 and 490 nm for upper Narragansett Bay (Station 36) using a*values of Prieur and Sathyendranath, 1981. 443 nm; 490 nm.

understand how specific species of diatoms anddinoflagellates compete for light in estuarine watersdominated by CDOM absorption. These analyses didnot consider size and packaging effects. Applyingthe mean absorption spectra of Pseudonitzschia spp.,Ditylum spp., and Chaetoceros spp. (coastal diatoms)and the dinoflagellate Alexandrium spp. in PugetSound waters (Figure 6 from Culver & Perry, 1999)combined with aCDOM443 and aCDOM532 fromStation 36, we calculated normalized CDOM absorp-tion coefficients from the aCDOM (�)/(aCDOM(�)+aChl (�)) ratios for Narragansett Bay. We suggestthat the results are applicable to New England watersas the diatom and dinoflagellate species used in thisanalysis commonly occur in Narragansett Bay (M.McFarland, pers. comm.). The wavelength used was

532 nm because this represents the region of peakabsorption by total carotenoid pigments (Stuart et al.,1998).

Results indicate that CDOM absorption and thecompetition for light continues to be highly variablethroughout the year (Figure 11) at 443 and 532 nm.Using the aCDOM443/(aCDOM443+aChl443) ratio,these results indicate that CDOM continued toaccount for more than 80% of light absorbed duringmid-late spring and autumn (Figure 11). CDOMabsorptions were reduced to approximately 60%of the total absorption during late winter/spring.aCDOM532 results indicate that during mid–latespring and autumn CDOM absorptions follow thetrend at 443 nm but are lower. In contrast, CDOMabsorptions at 532 nm are substantially reduced

Distribution of CDOM in Narragansett Bay 715

(between 20–90%) during the winter/spring diatombloom and isolated periods during the summermonths (Figure 11).

0.0

1.0

a CD

OM

/aC

DO

M +

aC

hl 0.8

21 M

ay 9

9

21 J

un

99

21 J

ul 9

9

21 A

ug

99

21 O

ct 9

9

21 N

ov 9

9

21 F

eb 0

0

21 M

ar 0

0

21 J

un

00

21 A

pr 0

0

21 M

ay 0

0

21 S

ep 9

9

21 D

ec 9

9

21 J

an 0

0

0.6

0.4

0.2

F 11. Normalized CDOM absorption time series at 443 and 532 nm based on a* values for coastal diatoms(Pseudonitzschia spp., Ditylum spp., and Chaetoceros spp.) and the dinoflagellate (Alexandrium spp.) from Culver and Perry,1999. 443 nm; 532 nm.

Discussion

In the North Atlantic Ocean, variability in CDOMabsorption has been strongly linked to seasonal cyclesof water column mixing (Garver & Siegel, 1997).During times of deep mixing, coloured dissolved anddetrital materials rise to the surface creating con-ditions for enhanced CDOM absorption. Periods ofshallow mixing result in low CDOM absorptions ascoloured dissolved material undergoes photobleach-ing effects in the stratified waters (Vodacek et al.,1997). In the equatorial Pacific, Pegau (1997) sug-gests that the magnitude CDOM absorption is relatedto variations in the salinity of the Equatorial Under-current. Our results show that the temporal andspatial variation in the absorption of light by CDOMin estuarine waters is strongly related to the salinitygradient produced by mixing processes. This obser-vation is consistent with coastal observations ofBricaud et al. (1981), Green and Blough (1994),Yentsch and Phinney (1997), Del Castillo et al.(1999) and (2000). In Narragansett Bay, this relation-ship results in the highest CDOM absorptions occur-ring in the brackish waters of the upper bay. Wesuggest there are no significant changes in the opticaland chemical properties of CDOM based on salinityvariability in Narragansett Bay based on the indepen-dence of salinity and S. In comparison, the relation-ship between S and chlorophyll indicates that changesin the optical and chemical properties of CDOM will

be controlled by increased chlorophyll concentrationsduring phytoplankton blooms.

We speculate that the late winter/early springfreshwater inflow from watersheds which surroundNarragansett Bay lowers surface water salinities andappears to coincide with episodes of high chlorophyllconcentration. Between late summer–early spring,CDOM absorptions in the middle–lower portionsof the bay rapidly decline to minimum values asNarragansett Bay surface waters respond to intrusionsof higher salinity waters from Rhode Island Sound(Figure 7). During most of the year, freshwater inputand the magnitude of CDOM absorption is highlycorrelated. However, this relationship becomes veryweak during the winter/early spring bloom whenfreshwater streams bring in sufficient concentrationsof nitrate and silicate (Pratt, 1965) to fuel phytoplank-ton blooms which briefly overwhelm the effects ofCDOM absorption. This association has also beenobserved by Yentsch and Phinney (1997).

The results suggest that variability in S may be usedas an indicator of the conditions under which acces-sory pigments become the primary absorbers of lightin estuarine and coastal waters. Our data show thatCDOM is the major absorber of light at 443 nm and amajor competitor at 490 nm (Figure 2). In waters withhigh CDOM absorptions and values of S�0·020,phytoplankton must utilize accessory photosyntheticpigments at longer wavelengths to collect sufficientlevels of light energy to assure survival. The dataindicate that during periods when the aCDOM/(aCDOM+aChl) ratio is at a minimum (i.e. lowCDOM absorption and S values �0·025), there isa concurrent rise in chlorophyll absorption. Forexample, at 532 nm, chlorophyll absorption increases

716 D. J. Keith et al.

by 40–90% in comparison to chl a absorption at443 nm (Figure 11). These periods of increasedchlorophyll absorption concur with times of increasedchlorophyll concentrations in the upper and lowerbay.

The ecological implications for phytoplankton arethat coastal diatoms and dinoflagellates appear toutilize photosynthetic carotenoid pigments, overthe spectral range of approximately 470–550 nm(UNESCO, 1997), to carry out the task of collectinglight in waters dominated by CDOM absorption.

Within algae cells, photosynthetic carotenoidsserve to harvest light energy while photo-protectivecarotenoids protect cells from the damaging effects ofphoto-oxidation (Roman, 1989). In coastal waters,photosynthetic carotenoids can account for as muchas 90% of the total carotenoids (Stuart et al., 1998).Specifically, the presence of carotenoid pigments suchas fucoxanthin (found in diatoms) and peridinin(found in dinoflagellates) can produce shifts in theabsorption spectra from the blue portion of the spectrato longer wavelengths (Kirk, 1996).

Remote sensing from satellites and aircraft offersthe environmental community a tool that cansynoptically monitor coastal waters at regional andsmaller spatial scales. The difficulty in developingocean colour algorithms to predict chlorophyll aconcentrations is underscored by the influence thatCDOM has as a major absorber of light over the samewavelengths favoured by phytoplankton.

Acknowledgements

This work was accomplished with fundingprovided by the U.S. EPA’s Office of Research andDevelopment under the Integrated AssessmentsResearch Theme at the National Health and Environ-mental Effects Research Laboratory/Atlantic EcologyDivision. We sincerely thank M. Berman (NOAA/NMFS) for additional chlorophyll data and T.Shepard (USGS-Northborough) for river and stream-flow data. Additionally, we thank Chief D. Brown andBosun’s Mate B. Miller (USCG-Castle Hill) for logis-tical support and ship time during winter operations.Special thanks go to Malcolm McFarland (URI/GSO), Don Cobb (USEPA), Colleen Beckmann(Western Michigan Univ.), and R. Ahlgren (USEPA)for their invaluable participation during data collec-tion and analysis. We sincerely appreciate commentsfrom Dan Campbell (USEPA), Hal Walker (USEPA),Antelmo Santos (USEPA) and the journal reviewersduring their critical review of this manuscript.

This report is contribution AED-01-042 of theAtlantic Ecology Division. Approval for publication

does not signify that the contents necessarily reflectthe views and policies of the U.S. EnvironmentalProtection Agency, nor does mention of trade namesor commercial products constitute endorsement orrecommendation.

References

Arar, E. & Collins, G. 1992 Method 445.0: In Vitro determinationof chlorophyll a and phaeophytin a in marine and freshwaterphytoplankton by fluorescense-version 1.1. EnvironmentalMonitoring Systems Laboratory, U.S. Environmental ProtectionAgency, EPA/600/R-92/121, 14 pp.

Bidigare, R. R., Ondrusek, M. E. & Brooks, J. M. 1993 Influenceof the Orinoco River outflow on distributions of algal pigmentsin the Caribbean Sea. Journal of Geophysical Research 98,2259–2269.

Blough, N. V. & Green, S. A. 1995 Spectroscopic characterizationand remote sensing of non-living organic matter. In The role ofnon-living organic matter in the earth’s carbon cycle (Zepp, R. G. &Sonntag, C., eds). John Wiley and Sons, pp. 23–45.

Blough, N. V. & Zepp, R. G. 1990 Effects of solar ultravioletradiation on biogeochemical dynamics in aquatic environ-ments. Woods Hole Oceanographic Institution Technical Report,WHOI-90-09, 194 pp.

Bricaud, A., Morel, A. & Prieur, L. 1981 Absorption by dissolvedorganic matter of the sea (yellow substance) in the UV and visibledomains. Limnology and Oceanography 26, 43–53.

Carder, K. L., Steward, R., Harvey, G. & Ortner, P. 1989 Marinehumic and fulvic acids: Their effects on remote sensing of oceanchlorophyll. Limnology and Oceanography 34 (1), 68–81.

Culver, M. & Perry, M. 1999 The response of photosyntheticabsorption coefficients to irradiance in culture and in estuarinewaters. Limnology and Oceanography 44 (1), 24–36.

D’Sa, E. J., Steward, R. G., Vodacek, A., Blough, N. V. & Phinney,D. 1999 Determining optical absorption of colored dissolvedorganis matter in seawater with a liquid capillary waveguide.Limnology and Oceanography 44 (4), 1142–1148.

Del Castillo, C., Coble, P., Morell, J. M., Lopez, J. M. & Corredor,J. E. 1999 Analysis of the optical properties of the Orinoco Riverplume by absorption and fluorescence spectroscopy. MarineChemistry 66, 35–51.

Del Castillo, C., Gilbes, F., Coble, P. & Muller-Karger, F. 2000 Onthe dispersal of riverine colored dissolved organic matter overthe West Florida Shelf. Limnology and Oceanography 45 (6),1425–1432.

Farmer, F., Vargo, G., Brown, C. Jr & Jarrett, O. Jr 1982 Spatialdistributions of major phytoplankton community components inNarragansett Bay at the peak of the winter-spring bloom. Journalof Marine Research 40 (3), 593–614.

Fox, L. E. 1983 The removal of dissolved humic acid duringestuarine mixing. Estuarine, Coastal and Shelf Science 16,431–440.

Garver, S. A. & Siegel, D. A. 1997 Inherent optical propertyinversion of ocean color spectra and its biochemical interpret-ation 1. Time series from the Sargasso Sea. Journal of GeophysicalResearch 102 (C8), 18,607–18,625.

Green, S. & Blough, N. 1994 Optical absorption and fluorescenceproperties of chomophoric dissolved organic matter in naturalwaters. Limnology and Oceanography 39 (8), 1903–1916.

Hess, K. 1976 A Three-dimensional numerical model of the estuarycirculation and salinity in Narragansett Bay. Estuarine and CoastalMarine Science 4, 325–338.

Hitchcock, G. & Smayda, T. 1977 The importance of light inthe initiation of the 1972–1973 winter-spring diatom bloom inNarragansett Bay. Limnology and Oceanography 22, 126–131.

International Ocean-Colour Coordinating Group (IOCCG) 2000Remote Sensing of Ocean Colour in Coastal, and Other

Distribution of CDOM in Narragansett Bay 717

Optically-Complex, Waters, edited by V. Stuart. IOCCG ReportNumber 3, 140 pp.

Kermack, K. A. & Haldane, J. B. S. 1950. Organic correlation andallometry. Biometrika 37, 30–41.

Kirk, J. 1996 Light and Photosynthesis in Aquatic Ecosystems, 2ndedn. Cambridge University Press, UK, 509 pp.

Laws, E. 1997 Mathematical Methods for Oceanographers-AnIntroduction. John Wiley and Sons, Inc., 343 pp.

Li, Y. & Smayda, T. 1998 Temporal variability of chlorophyll inNarragansett Bay, 1973–1990. ICES Journal of Marine Science 55,661–667.

Mitchell, B., Bricaud, A., Carder, K., Cleveland, J., Ferrari, G.,Gould, R., Kahru, M., Kishino, M., Maske, H., Moisan, T.,Moore, L., Nelson, N., Phinney, D., Reynolds, R., Sosik, H.,Stramski, D., Tassan, S., Trees, C., Weidemann, A., Wieland, J.& Vodacek, A. 2000 Determination of spectral absorptioncoefficients of particles, dissolved material and phytoplanktonfor discrete water samples. Ocean Optics Protocols for SatelliteOcean Color Sensor Validation-Revision 2 (Fragion, G. &Mueller, J., ed.). NASA Technical Memorandum, NASA/TM-2000-209966, pp. 125–153.

McKnight, D., Boyer, E., Westerhoff, P., Doran, P., Kulbe, T. &Andersen, D. 2001 Spectrofluorometric characterization of dis-solved matter for indication of precursor organic material andaromaticity. Limnology and Oceanography 46 (1), 38–48.

Mobley, C. D. 1998 Light and Water-Radiative Transfer in NaturalWaters. Academic Press, San Diego, 592 pp.

Mobley, C. D. 1999 Hydrolight 4.0 Users Guide. Sequoia Scientific,Inc., Mercer Island, WA, 106 pp.

Monahan, E. & Pybus, M. 1978 Colour, ultraviolet absorbance andsalinity of the surface waters off the west coast of Ireland. Nature274, 782–784. August 24.

Morel, A. 1988 Optical modeling of the upper ocean in relationto its biogenous matter content (Case I Waters). Journal ofGeophysical Research 93 (C9), 10,749–10,768. September.

Nixon, S. W., Oviatt, C. & Hale, S. S. 1976 Chapter 11: Nitrogenregeneration and the metabolism of coastal marine bottom com-munities. In The Role of Terrestrial and Aquatic Organisms inDecomposition Processes (Anderson, J. M. & Macfadyen, A., eds).Blackwell Scientific Publications, pp. 269–282.

O’Reilly, J., Maritorena, S., Mitchell, B. G., Siegel, D. A.,Carder,K. L. ,Garver, S., Kahru, M. & McClain, C 1998 Oceancolor chlorophyll algorithms for SeaWiFS. Journal of GeophysicalResearch 103, 24,937–24,953.

Pegau, S. 1997 The distribution of colored dissolved organicmatter (CDOM) in the Equatorial Pacific. In Ocean Optics XIII(Ackleson, S. G., ed.). SPIE Proceedings 2963, 508–513.

Pratt, D. 1965 The winter-spring diatom flowering in NarragansettBay, Limnology and Oceanography 10 (2), 173–184.

Pratt, D. 1959 The phytoplankton of Narragansett Bay. Limnologyand Oceanography 4 (4), 425–440.

Prieur, L. & Sathyendranath, S. 1981 An optical classification ofcoastal and oceanic waters based on the specific absorptioncurves of phytoplankton pigments, dissolved organic matter, andother particulate materials. Limnology and Oceanography 24 (4),671–689.

Prezioso, J., Goulet, J., O’Reilly, J., Holzwarth, T. & Benway, R. inreview Research Vessel Survey Sampling-3. 4. 4. Flow throughsampling operations. In MARMAP Ecosystem Surveys:Operational Manual, Jossi, J. (ed.). NOAA TechnicalMemoradum, NMFS-F/NEC, 125 pp.

Roman, K. S. 1989 Photosynthetic Pigments of Algae. CambridgeUniversity Press, Cambridge, 339 pp.

Sieburth, J. & Jensen, A. 1969 Studies on Algal Substances in theSea. II. The formation of gelbstoff (humic material) by exudatesof phaeophyta. Journal of Experimental Marine Biology and Ecology3, 275–289.

Smayda, T. 1998 Patterns of variability characterizing marinephytoplankton, with examples from Narragansett Bay. ICESJournal of Marine Science 55, 562–573.

Stuart, V., Sathyendranath, S., Platt, T., Maass, H. & Irwin, B.1998 Pigments and species composition of natural phytoplanktonpopulations: effect on the absorption spectra. Journal of PlanktonResearch 20, 187–217.

UNESCO 1997 Phytoplankton Pigments in Oceanography: Guidelinesto Modern Methods (Jeffrey, S. W., Mantoura, R. F. C. & Wright,S. W., eds). UNESCO Publishing, 661 pp.

Vodacek, A., Blough, N. V., DeGrandpre, M. D., Peltzer, E. T. &Nelson, R. K. 1997 Seasonal variation of CDOM and DOC inthe Middle Atlantic Bight: Terrestrial inputs and photooxidation.Limnology and Oceanography 42 (4), 674–686.

Welschmeyer, N. 1994 Flurorometric analysis of chlorophyll a inthe presence of chlorophyll b and pheopigments. Limnology andOceanography 39 (8), 1985–1992.

Yentsch, C. & Phinney, D. 1997 Yellow substances in the coastalwaters of the Gulf of Maine: Implications for ocean coloralgorithms. In Ocean Optics XIII (Ackleson, S. G., ed.). SPIEProceedings 2963, 120–131.