absorption spectral slopes and slope ratios as indicators
TRANSCRIPT
Old Dominion UniversityODU Digital Commons
Chemistry & Biochemistry Faculty Publications Chemistry & Biochemistry
2008
Absorption Spectral Slopes and Slope Ratios asIndicators of Molecular Weight, Source, andPhotobleaching of Chromophoric DissolvedOrganic MatterJohn R. HelmsOld Dominion University
Aron StubbinsOld Dominion University
Jason D. RitchieOld Dominion University
Elizabeth C. MinorOld Dominion University
David J. Kieber
See next page for additional authors
Follow this and additional works at: https://digitalcommons.odu.edu/chemistry_fac_pubs
Part of the Chemistry Commons, Fresh Water Studies Commons, Marine Biology Commons,and the Oceanography Commons
This Article is brought to you for free and open access by the Chemistry & Biochemistry at ODU Digital Commons. It has been accepted for inclusionin Chemistry & Biochemistry Faculty Publications by an authorized administrator of ODU Digital Commons. For more information, please [email protected].
Repository CitationHelms, John R.; Stubbins, Aron; Ritchie, Jason D.; Minor, Elizabeth C.; Kieber, David J.; and Mopper, Kenneth, "Absorption SpectralSlopes and Slope Ratios as Indicators of Molecular Weight, Source, and Photobleaching of Chromophoric Dissolved Organic Matter"(2008). Chemistry & Biochemistry Faculty Publications. 120.https://digitalcommons.odu.edu/chemistry_fac_pubs/120
Original Publication CitationHelms, J. R., Stubbins, A., Ritchie, J. D., Minor, E. C., Kieber, D. J., & Mopper, K. (2008). Absorption spectral slopes and slope ratios asindicators of molecular weight, source, and photobleaching of chromophoric dissolved organic matter. Limnology and Oceanography,53(3), 955-969. doi:10.4319/lo.2008.53.3.0955
AuthorsJohn R. Helms, Aron Stubbins, Jason D. Ritchie, Elizabeth C. Minor, David J. Kieber, and Kenneth Mopper
This article is available at ODU Digital Commons: https://digitalcommons.odu.edu/chemistry_fac_pubs/120
Absorption spectral slopes and slope ratios as indicators of molecular weight, source,
and photobleaching of chromophoric dissolved organic matter
John R. Helms, Aron Stubbins, Jason D. Ritchie, and Elizabeth C. Minor1
Department of Chemistry and Biochemistry, Old Dominion University, Norfolk, Virginia 23529
David J. KieberChemistry Department, State University of New York, School of Environmental Science and Forestry, Syracuse,New York 13210
Kenneth Mopper2
Department of Chemistry and Biochemistry, Old Dominion University, Norfolk, Virginia 23529
Abstract
A new approach for parameterizing dissolved organic matter (DOM) ultraviolet-visible absorption spectra ispresented. Two distinct spectral slope regions (275–295 nm and 350–400 nm) within log-transformed absorptionspectra were used to compare DOM from contrasting water types, ranging from wetlands (Great Dismal Swampand Suwannee River) to photobleached oceanic water (Atlantic Ocean). On the basis of DOM size-fractionationstudies (ultrafiltration and gel filtration chromatography), the slope of the 275–295-nm region and the ratio ofthese slopes (SR; 275–295-nm slope : 350–400-nm slope) were related to DOM molecular weight (MW) and tophotochemically induced shifts in MW. Dark aerobic microbial alteration of chromophoric DOM (CDOM)resulted in spectral slope changes opposite of those caused by photochemistry. Along an axial transect in theDelaware Estuary, large variations in SR were measured, probably due to mixing, photodegradation, andmicrobial alteration of CDOM as terrestrially derived DOM transited through the estuary. Further, SR varied byover a factor of 13 between DOM-rich wetland waters and Sargasso Sea surface waters. Currently, there is noconsensus on a wavelength range for log-transformed absorption spectra. We propose that the 275–295-nm slopebe routinely reported in future DOM studies, as it can be measured with high precision, it facilitates comparisonamong dissimilar water types including CDOM-rich wetland and CDOM-poor marine waters, and it appears tobe a good proxy for DOM MW.
The properties of dissolved organic matter (DOM) arediverse and depend on its source (e.g., terrestrial vs.aquatic) and diagenetic state. The fraction of DOM thatabsorbs ultraviolet (UV) and visible light is referred to aschromophoric or colored dissolved organic matter(CDOM). CDOM is largely responsible for the opticalproperties of most natural waters and plays key roles inshielding biota from harmful UV radiation (Walsh et al.2003) and in several biogeochemical and photochemicalprocesses (e.g., Mopper and Kieber 2002).
UV-visible absorption spectra for CDOM increaseapproximately exponentially with decreasing wavelength(Twardowski et al. 2004). To extract information fromthese spectra about CDOM properties, several spectralparameters, primarily absorption ratios, have been defined.These ratios are largely independent of CDOM concentra-tion, which is important in regimes such as estuaries, whereCDOM concentrations often vary by a factor of five ormore. De Haan and De Boer (1987) used the ratio ofabsorption at 250 to 365 nm (called E2 : E3) to trackchanges in the relative size of DOM molecules (see alsoPeuravouri and Pihlaja 1997). As molecular size increased,E2 : E3 decreased because of stronger light absorption byhigh-molecular-weight (HMW) CDOM at longer wave-lengths. The ratio of absorption at 465 to 665 nm (E4 : E6)was reported to be inversely related to CDOM aromaticity(Summers et al. 1987; Piccolo et al. 1992; Chin et al. 1994).However, E4 : E6 ratios have been shown to be bettercorrelated with molecular size, O : C and C : N atom ratios,carboxyl content, and total acidity than to aromaticity(Chen et al. 1977; Senesi et al. 1989) and, therefore, may bebetter suited as a general tracer of humification. In manynatural waters (as opposed to XAD resin extracts and soilextracts), where there is often little or no measurableabsorption at 665 nm, absorption at 254 nm (or sometimes280 nm) has been used in lieu of the E4 : E6 ratio as anindicator of humification or aromaticity (Summers et al.
1 Present address: Large Lakes Observatory, University ofMinnesota Duluth, Duluth, Minnesota 55812.
2 Corresponding author.
AcknowledgmentsWe thank Tim Brown, Brent Dalzell, Hussain Abdulla,
Nianhong Chen, and Joy Davis for UV-visible and DOCmeasurements as well as ultrafiltration of Elizabeth River andChesapeake Bay samples. We thank E. M. Perdue for providingthe SEC data, and George Westby and the crew of the RVEndeavor for their assistance in collecting samples and data in theDelaware Estuary. We acknowledge Robert F. Dias and twoanonymous reviewers for their helpful comments.
This research was supported by the following NSF grants:OCE-0096426, OCE-0096413, OCE-0241946, OCE-0327423,OCE-0555245, and OPP-0230499.
Limnol. Oceanogr., 53(3), 2008, 955–969
E 2008, by the American Society of Limnology and Oceanography, Inc.
955
1987; Weishaar et al. 2003). Weishaar et al. (2003) showedthat UV absorption at 254 nm, when normalized todissolved organic carbon (DOC) concentration, a param-eter called specific UV absorbance (SUVA254), correlatedstrongly (r2 . 0.97) with DOM aromaticity, as determinedby 13C-nuclear magnetic resonance (13C-NMR), for a largenumber of humic substance isolates.
In addition to the above parameters, the spectral slope(S, nm21) has been derived from CDOM absorptionspectra by fitting the absorption data to the equation:
al ~ alrefe{S l { lrefð Þ ð1Þ
where a 5 Napierian absorption coefficient (m21), l 5wavelength (nm), and lref 5 reference wavelength (nm)(e.g., Twardowski et al. 2004).
Spectral slopes provide further insights into the averagecharacteristics (chemistry, source, diagenesis) of CDOMthan absorption values alone, and like the E2 : E3 andE4 : E6 ratios, they are largely independent of CDOMconcentration (Brown 1977). S has been used for correctingremote sensing (bio-optic) data (Schwarz et al. 2002) andmonitoring CDOM degradative processes (Vahatalo andWetzel 2004). Carder et al. (1989) showed that S may beused to semiquantitatively describe the ratio of fulvic acids(FA) to humic acids (HA) in a sample. Similarly, it wasnoted that S correlates strongly with molecular weight(MW) of isolates of fulvic acids but not humic acids(Hayase and Tsubota 1985; Carder et al. 1989).
From the above, it is apparent that S is potentially usefulfor characterizing DOM. However, its usefulness is limitedby the fact that the value obtained for S depends on thewavelength interval over which it is calculated (Carder et al.1989; Stedmon et al. 2000). In fact, using a narrowwavelength range to calculate S often provides a differentresult from that obtained with a broader range (Twar-dowski et al. 2004). The use of narrow wavelength intervalsis advantageous as they minimize variations in S caused bydilution (Brown 1977). Sarpal et al. (1995) found that theslopes of two narrow wavelength intervals, 260–330 nmand 330–410 nm, described log-transformed absorptionspectra of seawater (Antarctic waters) far better than asingle slope determined over a broader wavelength interval.Further, these investigators suggested that the use of S overnarrow spectral regions may be exploited to reveal subtledifferences in the shape of spectra, which, in turn, mayprovide compositional or diagenetic insights.
In addition to the wavelength range chosen, the valueobtained for S also depends on the method used for itscalculation. For example, Stedmon et al. (2000) and DelVecchio and Blough (2004a) demonstrated that spectra areoften better described by nonlinear regression fittingroutines, especially over broad wavelength ranges andwhere absorption approaches the detection limit of theinstrument. As a result of the lack of a standardizedmathematical or operational definition of S (i.e., linear vs.nonlinear regression) or a specific wavelength range for itscalculation, the literature contains a wide range in valuesfor S even for similar sample types (for a review seeTwardowski et al. 2004). This lack of standardization has
made comparisons of published S values difficult orimpossible, and, in fact, has resulted in contradictoryconclusions. For example, Miller (1994), Morris andHargraeves (1997), and Gao and Zepp (1998) reportedthat S decreased during irradiation of natural watersamples, which runs counter to the findings of severalother studies (e.g., Whitehead et al. 2000; Del Vecchio andBlough 2002; Vahatalo and Wetzel 2004). However, theapparent contradictory results were internally consistentgiven the wavelength ranges and fitting routines that werechosen.
Like the E2 : E3 ratio, the spectral slope has been used asa proxy for MW in a broad range of samples. However, Shas only been shown to covary with MW for CDOM thatwas isolated by solid-phase extraction (e.g., XAD resins orC-18 silica) or fractionated in relatively harsh acid–baseconditions (Zepp and Schlotzhauer 1981; Hayase andTsubota 1985; Carder et al. 1989). Consequently, observeddifferences in MW between FA and HA fractions, to someextent, may have resulted from artifacts related to theisolation methods (Schmit and Wells 2002).
In this study, these potential artifacts and the depen-dence of S on wavelength range are addressed byexamining absorption spectra of sterile-filtered wholewater, ultrafiltered samples (,1-kDa permeates andretentates), and gel filtration chromatography (GFC) sizefractions; and by determining S over two narrow wave-length ranges. By calculating the ratio of the slope of theshorter wavelength region (275–295 nm) to that of thelonger wavelength region (350–400 nm), a dimensionlessparameter called ‘‘slope ratio’’ or SR is defined. Thisapproach avoids the use of spectral data near the detectionlimit of the instruments used, and focuses on absorbancevalues that shift dramatically during estuarine transit andphotochemical alteration of CDOM. We investigate herethe dependence of spectral shape parameters (primarily Sand SR) on the MW distribution of aquatic CDOMsamples from a broad range of environments.
Methods and materials
Cleaning procedures—All glassware was cleaned withdilute (5–10%) HCl and rinsed with MilliQ ultrapure-gradewater (Millipore). Glassware was either combusted at450uC or autoclaved before use. Plasticware was cleanedwith 5–10% HCl and rinsed with MilliQ water. Stainlesssteel equipment was cleaned with mild detergent andcopiously rinsed with MilliQ. All containers were rinsedseveral times with sample before filling. Filter capsules(0.1 mm or 0.2 mm pore size, Whatman PolyCap) wererinsed with acetonitrile, flushed with .20 liters of MilliQwater, and conditioned with approximately 1 liter ofsample.
Elizabeth River, Great Dismal Swamp, and lowerChesapeake Bay—Water samples were taken from threesites within the Elizabeth River and Chesapeake Bayestuary system during May 2004, October 2004, May2005, and October 2005. The Elizabeth River estuary istidally influenced over its entire length and receives
956 Helms et al.
substantial input of CDOM-rich water from the GreatDismal Swamp canal system (by way of the Deep CreekLock, Portsmouth Ditch, and several other small creeks).At our sampling times, the salinity at the upriver siteranged between 8 and 12 and contained high concentra-tions of DOC (typically 10–20 mg L21). Our high-salinity(marine) station (the mouth of the Chesapeake Bay; salinity5 23–26) is at the confluence of several estuarine systems(typical DOC concentrations were 1–2 mg L21).
Sampling was always performed at low tide using astainless steel or polypropylene bucket. The samples weretransported in polypropylene carboys back to the lab,where they were gravity filtered using high-flow filtercapsules within 12 h of collection. The filtrate was collectedand stored in a glass carboy, which was wrapped in blackplastic to minimize light exposure. The filtered watersamples were placed in 500-mL round-bottom quartz glassflasks and irradiated using the solar simulator describedbelow. In 2004, UV-visible absorption spectra (200–800 nm) were measured for irradiated samples and darkcontrols using a double-beam scanning spectrophotometer(Bio300 UV-Vis Cary, Varian) with 1-cm quartz cuvettesand ultrapure water (Elga) as the blank. In May 2005,measurements were made using a Hitachi U-1000 double-beam spectrophotometer with either 1-cm or 10-cm quartzcuvettes and ultrapure water (Elga) as the blank. InOctober 2005, spectra were taken with an Agilent 8453diode array spectrophotometer using either a 1-cm or 5-cmquartz cuvette.
Irradiated and dark control 0.1-mm- or 0.2-mm-filteredsamples were also size fractionated by ultrafiltration (UF)using Amicon 8400 stirred cells fitted with 1,000-Daregenerated cellulose membranes (Millipore YM1, 76-mmdiameter) and pressurized using ultrapure nitrogen. Sterilefiltered (non-UF fractionated) samples, UF permeates, andretentates were analyzed by high-temperature combustionDOC analysis (Shimadzu TC-5000) to measure the non-purgeable organic carbon concentration and by UV-visiblespectroscopy to measure absorption spectra as describedabove. UV-visible mass balances were calculated using thetotal (volume-corrected) absorption integrated from 250 to450 nm and are presented as a percentage of the absorptionmeasured for the two size fractions (i.e., permeate andretentate) divided by the absorption measured for the non-UF sample. DOC mass balances were calculated from theconcentration of DOC (measured via high-temperaturecombustion on acidified samples) for both size fractions(volume corrected) divided by the DOC concentration ofthe non-UF sample.
In June 2004 and February 2006, samples were collectedfrom the Portsmouth Ditch and Northwest Canal sections,respectively, of the Great Dismal Swamp canal system.Samples were collected and filtered as above. TheNorthwest Canal sample was irradiated using the solarsimulator described below and ultrafiltered as describedabove. One aliquot of the sample was sparged withultrapure helium to significantly decrease the concentrationof dissolved molecular oxygen during light exposure.Before UV-visible absorption measurement, samples werediluted by a factor of five in MilliQ water to bring the
absorbance into the linear response range of the spectro-photometer (1-cm pathlength).
In July 2005 the Old Dominion University (ODU)Sailing Basin site on the Elizabeth River (salinity 5 19) wassampled to investigate microbial (community) effects onthe shape of the CDOM UV-visible spectrum. Theunfiltered, unamended water sample was stirred constantly,kept at room temperature, in the dark, and maintained atequilibrium with lab air for 2 weeks. Duplicate 20-mLsubsamples were collected at intervals and immediatelypoisoned with 100 mmol L21 potassium cyanide (finalconcentration). Samples were allowed to settle for at least7 d, after which they were analyzed using the Agilent 8453spectrophotometer with a 1-cm quartz cell. Blanks consist-ing of an exact aliquot of poison in MilliQ water wereprepared at the time of each poisoning.
Suwannee River natural organic matter (SRNOM)—Awell-characterized aquatic humic substance sample wasseparated by size using preparative gal filtration chroma-tography (GFC) according to the procedure in Ritchie(2005). Briefly, 1 g of SRNOM (International HumicSubstances Society, 1R101N) was separated into six sizefractions on a column packed with Superdex-30 (AmershamPharmacia Biotech-GE Healthcare) that was calibratedusing polystyrene sulfonate salts (Polysciences), methylthy-mol blue, and p-hydroxybenzoic acid. The mobile phaseconsisted of a NaHCO3/Na2CO3 mixture (pH 5 9.2, ionicstrength 5 0.1 mol L21). Size fraction 1 (greatest apparentMW) through fraction 6 (smallest apparent MW) representfractions collected sequentially from the Gaussian-like humpbetween the total permeation volume and total exclusionvolume of the column. Each size fraction was subsequentlydesalted and freeze dried. The average MW of each sizefraction was determined by analytical size-exclusion chro-matography (SEC). SEC was performed using Superdex-30stationary phase packed in a 0.6-cm-diameter Pharmaciacolumn with UV detection at 254 nm. The mobile phaseconsisted of 0.1 mol L21 NaCl buffered with 2 mmol L21
phosphate adjusted to pH 6.8 (1 mL min21). The GFCfractions and whole SRNOM were prepared at 15 mg C L21
with the same electrolyte concentrations as the mobile phase,and polystyrene sulfonate salts were used as MW standards.For UV-visible absorbance measurements, the size fractionswere prepared as for SEC. Spectra were recorded using a 1-cm quartz cuvette at 25uC on a HP 8451 diode arrayspectrophotometer with 2-nm resolution.
Delaware Estuary and Atlantic Ocean—Surface watersamples were collected during two transects along the mainaxis of the Delaware Estuary in the summers of 2000 and2002 on board the RV Endeavor and again in the summerof 2005 aboard the RV Cape Henlopen, from the ships’clean seawater supply systems, which were fed from an inletat ,2–3-m depth. Surface samples in the coastal watersnear the mouth of the bay and along the mid-Atlantic Bightwere also obtained from the underway system. In a thirdEndeavor cruise during summer 2003, waters were sampledat two coastal sites in the Georgia Bight (GAB). Sampleswere collected at depths from 4 to 10 m using 20-liter GO-
CDOM molecular weight and sources 957
FLOTM bottles. All water samples were filtered and storedas described above. The UV-visible measurements weremade at sea using a Hewlett Packard 8453 diode arrayspectrophotometer equipped with a 5-cm flow cell withMilliQ water as the blank.
Samples from the near-shore GAB sample site, Sta.B, were exposed to sunlight on the ship’s deck, as wellas in a shipboard solar simulator containing three UVA340(Q-Panel) bulbs and three full-spectrum fluorescentbulbs (GE Spectra Rays F40T12/SR). Sample UV-visibleabsorption spectra were measured onboard using theHewlett Packard 8453 spectrophotometer, as describedabove.
In June 2006, surface waters were collected from theDelaware River at the Yardley, Pennsylvania public boatramp by completely submerging several precleaned poly-propylene carboys. Samples were 0.1 mm filtered andpoisoned with mercuric chloride (0.1 mL saturated solutionper liter of sample) to inhibit microbial regrowth. Controlexperiments showed that mercuric chloride had no discern-able effect on either photoreactivity or UV-visible spectra.Samples were exposed to sunlight for 3 weeks in August onthe roof of the chemistry building on the campus of ODUin Norfolk, Virginia. The containers used during theirradiation were 80-liter polycarbonate tubs (Rubbermaid)with lids modified to include UV-transparent fluorinatedethylene–propylene windows (American Durafilm). Sub-samples were collected at 2- to 4-d intervals using cleansilicon tubing and analyzed using the Agilent 8453 diodearray spectrophotometer.
Sample irradiation system—Unless indicated otherwise,both light-exposed samples and dark controls were placedon a rotating table (1 revolution per minute) within a solarsimulator containing 12 Q-Panel UV340 bulbs, whichprovided a spectral shape similar to that of natural sunlightfrom ,295 to ,365 nm (Q-Panel), but underrepresentedsolar radiation at wavelengths greater than about 365 nm.The light output from the solar simulator was measuredusing an International Light IL-1700 radiometer, equippedwith an SUD240 sensor and a W#6931/300#14571 filter/diffuser (UVA), and was comparable with springtime, noonsunlight intensity at 40uN (Leifer 1988). To directlycompare our irradiations with sunlight, an Elizabeth Riverwater sample was irradiated for 5 h under both natural andsimulated sunlight, and the extent of photobleaching(integrated over 250 to 400 nm) was determined. In thiscomparison, the solar simulator provided 127% of thephotobleaching occurring under winter midday sunlight at36.89uN.
Spectral corrections and S determination—CDOM ab-sorbance was assumed to be zero above 700 nm; therefore,the average sample absorbance between 700 and 800 nmwas subtracted from the spectrum to correct for offsets dueto instrument baseline drift, temperature, scattering, andrefractive effects (Green and Blough 1994). Absorbanceunits were converted to absorption coefficients as follows:
a ~ 2:303A=l ð2Þ
where a 5 absorption coefficient (m21), A 5 absorbance,and l 5 path length (m).
The spectral slope for the interval of 300–700 nm (S300–
700) was determined by fitting the absorption spectra to asingle exponential decay function (Eq. 1) by nonlinearregression (SigmaPlot). Spectral slopes reported here forthe intervals of 275–295 nm (S275–295) and 350–400 nm(S350–400) were calculated using linear regression of the log-transformed a spectra. S275–295 and S350–400 were calculat-ed, for a subset of sample spectra, using both the log-transform linear regression and nonlinear regressionmethods. Variation between the two methods was less than1%. Slopes are reported as positive numbers to follow themathematical convention of fitting to an exponential decay(Eq. 1). Thus, higher (or steeper) slopes indicate a morerapid decrease in absorption with increasing wavelength.The ranges, 275–295 nm and 350–400 nm, were chosenbecause the first derivative of natural-log spectra indicatedthat the greatest variations in S from a variety of samples(marsh, riverine, estuarine, coastal, and open ocean)occurred within the narrow bands of 275–295 nm and350–400 nm. The slope ratio or SR was calculated as theratio of S275–295 to S350–400. The E2 : E3 and E4 : E6 ratioswere calculated using absorption coefficients at theappropriate wavelengths. SUVA254 was calculated by
Fig. 1. (A) Absorption spectra obtained for 0.2 mm filteredwater samples from six sites: Great Dismal Swamp, Great Bridgeand Town Point in the Elizabeth River estuary, Chesapeake BayBridge on the lower Chesapeake Bay (all collected in spring 2004),a coastal site on the Georgia Bight (coastal), and Sargasso Sea(ocean). (B) Natural log-transformed absorption spectra for theabove samples with best-fit regression lines for two regions (275–295 nm and 350–400 nm); slope ratio (SR) values are given foreach log-transformed spectrum.
958 Helms et al.
dividing a254 (m21) by the DOC concentration (mg L21) asin Weishaar et al. (2003). The total CDOM absorption wascalculated as the integrated absorption from 250 to 450 nm(1-nm resolution).
Results
MW and spectral shape—Figure 1A shows absorptionspectra obtained for a variety of freshwater, estuarine, and
Table 1. Mass balances on the basis of total absorption (250–450 nm) and dissolved organic carbon measurements for each of theultrafiltration size fractions. Chesapeake Bay Bridge (CBB), Town Point (TP), Great Bridge (GB), and Dismal Swamp (DS); init5initial;con5dark control; T15irradiated 24 h; T25 irradiated 48 h; ox5air saturated; anox5He sparged, low oxygen samples.
Samplename Salinity
Irr t(h)
%LMWCDOM
%HMWCDOM
%CDOMrecovery
%LMWDOC
%HMWDOC
%DOCrecovery
May 2004 GB init 12 0 18.9 72.9 91.8 41.7 68.2 109.9May 2004 GB con 12 0 16.5 74.5 91.0 39.2 71.0 110.2May 2004 GB T1 12 24 22.8 72.4 95.2 40.9 60.6 101.5May 2004 GB T2 12 48 22.2 65.7 87.9 31.5 54.6 86.1May 2004 TP init 19 0 29.8 67.6 97.4 49.7 61.8 111.5May 2004 TP con 19 0 25.3 65.0 90.3 43.5 56.3 99.8May 2004 TP T1 19 24 36.0 65.2 101.1 63.0 62.4 125.4May 2004 TP T2 19 48 41.3 52.3 93.6 62.3 54.6 116.9May 2004 CBB init 23 0 57.4 43.9 101.3 59.1 38.5 97.6May 2004 CBB con 23 0 49.8 51.6 101.4 54.9 47.2 102.1May 2004 CBB T1 23 24 70.7 47.9 118.6 67.3 48.2 115.5May 2004 CBB T2 23 48 60.0 40.6 100.6 51.1 35.5 89.6Oct 2004 GB init 10 0 17.5 89.4 106.9 26.0 67.6 93.6Oct 2004 GB con 10 0 19.6 77.1 96.7 31.0 69.3 100.3Oct 2004 GB T1 10 24 30.3 78.8 109.1 33.6 59.2 92.8Oct 2004 GB T2 10 48 26.4 74.4 100.8 57.7 59.1 116.8Oct 2004 TP init 17 0 24.4 70.6 95.0 36.1 59.2 95.3Oct 2004 TP con 17 0 24.3 78.5 102.8 38.8 62.1 100.9Oct 2004 TP T1 17 24 32.1 67.7 99.8 37.2 52.4 89.6Oct 2004 TP T2 17 48 31.4 57.4 88.8 49.0 47.0 96.0Oct 2004 CBB init 26 0 35.1 78.3 113.4 88.0 35.2 123.3Oct 2004 CBB con 26 0 18.4 51.8 70.2 70.9 20.8 91.7Oct 2004 CBB T1 26 24 No data No data No data 61.7 27.9 89.6Oct 2004 CBB T2 26 48 No data No data No data 82.5 24.0 106.5May 2005 GB init 8 0 26.8 65.7 92.5 40.5 60.5 101.0May 2005 GB con 8 0 29.1 65.4 94.5 43.2 63.4 106.6May 2005 GB T1 8 24 29.7 79.9 109.6 66.4 56.1 122.5May 2005 GB T2 8 48 32.3 58.7 91.0 46.1 48.0 94.0May 2005 TP init 18 0 22.5 71.4 93.9 36.3 51.3 87.6May 2005 TP con 18 0 24.6 73.6 98.3 45.4 60.7 106.2May 2005 TP T1 18 24 36.9 55.5 92.4 60.5 51.6 112.1May 2005 TP T2 18 48 38.5 63.0 101.5 103.8 45.9 149.7May 2005 CBB init 25 0 46.0 No data No data 55.6 22.8 78.5May 2005 CBB con 25 0 45.9 No data No data 52.1 20.9 73.1May 2005 CBB T1 25 24 43.7 No data No data No data No data No dataMay 2005 CBB T2 25 48 45.0 No data No data 64.5 17.9 82.5Oct 2005 GB init 11 0 19.3 71.6 90.8 No data No data No dataOct 2005 GB con 11 0 21.5 73.2 94.6 No data No data No dataOct 2005 GB T1 11 24 25.3 67.0 92.3 No data No data No dataOct 2005 GB T2 11 48 27.6 60.8 88.4 No data No data No dataOct 2005 TP init 18 0 23.4 67.3 90.7 No data No data No dataOct 2005 TP con 18 0 23.5 62.0 85.5 No data No data No dataOct 2005 TP T1 18 24 28.4 63.9 92.3 No data No data No dataOct 2005 TP T2 18 48 34.2 55.3 89.5 No data No data No dataOct 2005 CBB init 23 0 49.4 54.8 104.2 No data No data No dataOct 2005 CBB con 23 0 No data 31.8 No data No data No data No dataOct 2005 CBB T1 23 24 60.3 40.6 110.9 No data No data No dataOct 2005 CBB T2 23 48 68.4 37.6 106.0 No data No data No dataFeb 2006 DS ox con 0 0 9.4 88.7 98.1 No data No data No dataFeb 2006 DS ox T1 0 144 24.3 77.3 101.6 No data No data No dataFeb 2006 DS anox con 0 0 7.5 88.0 95.5 No data No data No dataFeb 2006 DS anox T1 0 144 12.0 85.6 97.6 No data No data No data
CDOM molecular weight and sources 959
marine samples. Figure 1B illustrates the linearization andleast-squares fit of these spectra, with the linear fit shownfor the spectral regions 275–295 nm and 350–400 nm.There are visible distinctions between the slopes for thesetwo regions, depending on whether the sample is mainlyterrestrial or marine in character; e.g., for marine samplesS275–295 . S350–400, whereas the opposite trend holds forhigh-CDOM, terrestrially dominated samples.
Absorption and DOC mass balances for the ultrafilteredElizabeth River samples are given in Table 1 and Fig. 2.These results show two general trends. First, the lower-salinity upriver samples contained more HMW CDOM(.1,000 Da) relative to low-MW (LMW) CDOM(,1,000 Da) than the high-salinity Chesapeake Bay sam-ples. Second, the percentage of CDOM in the LMWfraction in all three sampling sites increased upon lightexposure, whereas in the HMW fraction, it decreased. Thelatter trend is especially pronounced for the upriversamples (Fig. 2; Table 1). These results indicate that thereis a shift from HMW to LMW CDOM both withirradiation and with increasing salinity across the freshwa-ter–marine continuum.
The UV-visible absorption (a), spectral slope (S), andslope ratio (SR) obtained for the 0.2-mm-filtered samplesshowed several consistent trends. Upriver samples hadhigher absorption coefficients (Fig. 1), lower S275–295
values, and lower SR values (Figs. 3, 4; Table 2) thandownriver and Chesapeake Bay samples. Irradiation
consistently increased S275–295, SR, and E2 : E3, anddecreased UV absorption and DOC (Figs. 3, 4; Table 2)for samples from all three sites (e.g., compare the foursamples from any single site in Table 2). Furthermore,S275–295 and SR values of the HMW (.1,000-Da UFretentate) fractions were generally lower than those for thecorresponding LMW (UF permeate) fractions (Table 3). Inaddition, spectral slopes S275–295 and S350–400 for both MWfractions showed consistent shifts upon irradiation, withS275–295 generally increasing upon irradiation and S350–400
values generally decreasing (e.g., compare treatments froma single site and season in Table 3). These results provideevidence that shifts in SR and S275–295 are related to shifts inMW and photobleaching. Figure 4A shows that SR
correlates well with CDOM MW, as determined from theUF experiments. In this figure, SR values of 0.2-mm-filteredsamples were plotted against the ratio of LMW to HMWabsorption (integrated over 250–450 nm), whereas Fig. 4Bshows the plot on a DOC basis. The SR values correlatedwith MW shifts in the estuary (i.e., for dark controls andinitial samples) and with shifts in MW that occurred duringthe 48-h irradiations on both a CDOM and DOC basis.The latter implies that estimates of MW on the basis ofCDOM measurements likely apply to total DOM. TheE2 : E3 ratio (Table 2) also showed a linear correlation withMW in irradiated samples. However, this correlation wasless robust (Helms 2006) and was generally not applicableto marine waters where 365-nm CDOM absorption oftenapproached the detection limit of traditional spectropho-tometers with 10-cm quartz cuvettes (Stubbins 2006).
In agreement with the above Elizabeth River–Chesa-peake Bay results, we found that the spectral slopes of size-fractionated SRNOM were highly correlated to weight-average MW over the size range analyzed (Fig. 5A,B),whereas the SR parameter (Fig. 5C) was only linearlyrelated to MW less than 3,000 Da. In each plot, theunfractionated SRNOM sample plotted near the scatter ofthe size-fractioned results. Using broader wavelengthranges such as 300–700 nm (data not shown) also yieldedlinear trends (slope 5 1.81 3 1026 nm21 Da21; r2 5 0.903).
Long-term (6 d) irradiation of the Dismal Swampsamples caused significant decreases in overall CDOMabsorption and a large transfer of colored material fromthe HMW fraction to the LMW fraction (Tables 1, 3),whose absolute absorption actually increased. SR increasedfrom a terrestrial value (,0.71) to one more typical ofestuarine or coastal samples (,1.1) (Table 2). Partialremoval of oxygen (initial oxygen concentration reducedto 21% saturation, resulting in 88% less photochemicaloxygen consumption than in oxygenated samples) beforethe irradiation of the Dismal Swamp samples decreased theextent to which CDOM was photobleached (Table 2), aswell as the extent to which CDOM shifted from HMW toLMW pool (Tables 1, 3). However, this partial removal ofoxygen did not significantly affect the light-induced shift inS and SR for this sample (Table 2). This anomalous resultwill be discussed below.
Interestingly, changes in spectral shape (and consequent-ly S and SR values) caused by microbial activity differedsignificantly from those observed for photochemical
Fig. 2. (A) CDOM and (B) DOC mass balance percentagesaveraged over all four sampling seasons. Great Bridge (GB),Town Point (TP), and Chesapeake Bay Bridge (CBB); Dark, darkcontrol (no irradiation); 24, irradiated for 24 h; 48, irradiated for48 h. Percentages are given above each symbol.
960 Helms et al.
degradation. Aerobic microbial incubations in the darkresulted in a statistically significant decrease in SR overtimescales of days to weeks (from SR 5 1.02 to SR 5 0.94in 2 weeks) (Fig. 6), either due to the microbial productionor selective preservation of long-wavelength-absorbingsubstances. This result is consistent with previous studiesthat found that microbial processes shifted spectral slopesopposite, but of a lower magnitude, to those caused byphotochemistry over timescales of several weeks to severalmonths (Moran et al. 2000; Vahatalo and Wetzel 2004).Our results therefore indicate that the down-estuarychanges in optical properties are likely the result of acombination of processes, including mixing of low-SR
freshwater with high-SR marine water, photochemicallyinduced increases in SR values, and microbially relateddecreases in SR values.
Spectral changes in the Delaware Estuary—SR valueswithin the Delaware estuary ranged from 0.88 in the riverto 1.32 at the bay mouth. Figure 7C illustrates that SR
showed remarkably similar trends with salinity in summer2000, 2002, and 2005. These trends were likely primarilydue to mixing of river water with marine water in theestuary, and to progressive photobleaching during thedownstream transit, especially near the mouth of the bay.Figure 8 shows that S275–295 and SR values throughout theestuary are somewhat higher than would be predicted bysimple two end-member mixing, whereas S350–400 values arelower than predicted by mixing. These deviations are mostlikely due to photochemical bleaching of CDOM, as inputsof marsh- (i.e., terrestrially) derived DOM in the lowerestuary would have resulted in opposite deviations (Fig. 1).
The down-estuary increase in SR was mainly due to anincrease in S275–295 (Fig. 7A, 8A), which is much moresensitive to DOM spectral changes within the estuary thanS300–700 (Fig. 7A). However, the sensitivity of SR as anindicative parameter is enhanced by the decrease of slope inthe long-wavelength region (Fig. 7A, 8B).
Irradiating surface water collected in the Delaware Riverwith sunlight caused changes in S that were similar in
Fig. 3. Spectral slope parameters averaged over all foursampling seasons. (A) S275–295, (B) S350–400, and (C) ratio ofspectral slopes, SR, increase down-estuary and with increasingirradiation dose during an experiment. The vertical bars representseasonal and interannual variations, not analytical error (analyt-ical error was usually ,1%).
Fig. 4. Regression plots of the ratio of spectral slopes (SR)versus molecular weight distribution, quantified by (A) UV-visibleabsorption and by (B) DOC concentration, for samples from theGreat Dismal Swamp (absorbance data only), Elizabeth River(Town Point, Great Bridge), and Chesapeake Bay Bridge Tunnel,as delineated by the boxes. SR values were based on unfraction-ated samples, whereas MW fractions were calculated indepen-dently of these spectra (see Methods and materials). (A) Slope 50.594, intercept 5 0.794, r2 5 0.858; (B) slope 5 0.333, intercept 50.7663, r2 5 0.531.
CDOM molecular weight and sources 961
Table
2.
Sp
ectr
al
pa
ram
eter
sm
easu
red
for
each
of
the
cart
rid
ge-
filt
ered
(,0
.2o
r,
0.1
mm
)sa
mp
les
fro
mth
eC
hes
ap
eak
eB
ay
Bri
dg
e(C
BB
),T
ow
nP
oin
t(T
P),
Gre
at
Bri
dg
e(G
B),
an
dD
ism
al
Sw
am
p(D
S);
init
5in
itia
l;co
n5
da
rkco
ntr
ol;
T15
irra
dia
ted
24
h;
T25
48
h.
Sa
mp
len
am
eS
ali
nit
yIr
rt
(h)
S3
00
–7
00
(nm
21)
S2
75
–2
95
(nm
21)
S3
50
–4
00
(nm
21)
SR
To
tal
a(2
50
–4
50
nm
)D
OC
(mm
ol
L2
1)
E2
:E3
a2
54
(m2
1)
a3
00
(m2
1)
SU
VA
25
4(m
gC
L2
1
m2
1)
Ma
y2
00
4G
Bin
it1
20
0.0
15
70
.01
44
0.0
16
70
.85
97
,09
48
20
5.0
01
02
.75
7.2
10
.40
Ma
y2
00
4G
Bco
n1
20
0.0
15
80
.01
39
0.0
16
50
.84
17
,17
18
57
4.9
71
03
.15
8.7
10
.00
Ma
y2
00
4G
BT
11
22
40
.01
53
0.0
16
50
.01
53
1.0
86
,00
78
59
5.4
29
1.0
48
.08
.84
Ma
y2
00
4G
BT
21
24
80
.01
54
0.0
17
50
.01
51
1.1
55
,20
49
11
5.8
88
2.8
41
.17
.59
Ma
y2
00
4T
Pin
it1
90
0.0
15
00
.01
45
0.0
16
10
.90
13
,17
14
28
4.8
04
5.7
25
.58
.91
Ma
y2
00
4T
Pco
n1
90
0.0
14
90
.01
46
0.0
16
00
.91
43
,12
64
07
4.8
54
5.1
25
.69
.24
Ma
y2
00
4T
PT
11
92
40
.01
47
0.0
18
60
.01
50
1.2
42
,46
93
99
5.7
13
8.9
28
.88
.13
Ma
y2
00
4T
PT
21
94
80
.01
46
0.0
20
70
.01
50
1.3
82
,05
23
90
6.2
33
4.5
25
.77
.39
Ma
y2
00
4C
BB
init
23
00
.01
74
0.0
19
10
.01
56
1.2
25
42
19
18
.07
9.4
4.0
4.1
1M
ay
20
04
CB
Bco
n2
30
0.0
17
40
.02
16
0.0
17
81
.21
50
91
92
7.7
19
.04
.53
.94
Ma
y2
00
4C
BB
T1
23
24
0.0
14
90
.03
10
0.0
17
11
.81
37
01
77
8.7
47
.32
.83
.47
Ma
y2
00
4C
BB
T2
23
48
0.0
17
00
.03
08
0.0
20
11
.53
31
52
24
10
.74
6.8
2.2
2.5
6O
ct2
00
4G
Bin
it1
00
0.0
14
60
.01
31
0.0
15
50
.84
51
0,2
64
12
79
4.5
61
43
.28
3.0
9.3
3O
ct2
00
4G
Bco
n1
00
0.0
14
10
.01
31
0.0
14
90
.87
91
0,6
79
12
37
4.5
11
48
.98
5.8
10
.00
Oct
20
04
GB
T1
10
24
0.0
13
90
.01
50
0.0
13
91
.08
9,1
38
12
16
4.8
31
33
.27
1.8
9.1
3O
ct2
00
4G
BT
21
04
80
.01
32
0.0
16
30
.01
34
1.2
18
,03
01
10
65
.13
12
1.9
62
.59
.19
Oct
20
04
TP
init
17
00
.01
30
0.0
13
00
.01
40
0.9
26
6,0
96
77
44
.32
84
.44
8.5
9.1
0O
ct2
00
4T
Pco
n1
70
0.0
13
40
.01
30
0.0
14
50
.89
76
,21
37
54
4.3
38
5.8
49
.49
.48
Oct
20
04
TP
T1
17
24
0.0
12
60
.01
58
0.0
12
81
.23
5,0
06
73
84
.79
73
.83
8.5
8.3
4O
ct2
00
4T
PT
21
74
80
.01
19
0.0
16
70
.01
24
1.3
44
,62
46
87
4.9
37
0.1
35
.18
.51
Oct
20
04
CB
Bin
it2
60
0.0
13
80
.01
96
0.0
14
21
.79
53
81
62
9.0
08
.64
.04
.43
Oct
20
04
CB
Bco
n2
60
0.0
11
60
.02
51
0.0
10
42
.52
55
81
74
6.1
78
.64
.04
.15
Oct
20
04
CB
BT
12
62
4N
od
ata
No
da
taN
od
ata
No
da
taN
od
ata
17
4N
od
ata
No
da
taN
od
ata
No
da
taO
ct2
00
4C
BB
T2
26
48
No
da
taN
od
ata
No
da
taN
od
ata
No
da
ta1
64
No
da
taN
od
ata
No
da
taN
od
ata
Ma
y2
00
5G
Bin
it8
00
.01
48
0.0
14
20
.01
88
0.7
60
6,6
93
92
95
.06
92
.95
1.5
8.3
4M
ay
20
05
GB
con
80
0.0
14
70
.01
43
0.0
19
00
.75
56
,73
28
96
4.9
89
2.9
51
.78
.65
Ma
y2
00
5G
BT
18
24
0.0
15
70
.01
70
0.0
18
10
.93
95
,65
19
13
5.6
08
3.3
42
.37
.61
Ma
y2
00
5G
BT
28
48
0.0
16
60
.01
85
0.0
18
50
.99
74
,81
98
52
6.1
27
4.7
35
.57
.31
Ma
y2
00
5T
Pin
it1
80
0.0
16
40
.01
41
0.0
15
80
.88
84
,46
16
23
4.9
66
0.6
33
.98
.11
Ma
y2
00
5T
Pco
n1
80
0.0
16
10
.01
37
0.0
14
50
.94
84
,59
55
38
4.8
96
1.5
34
.79
.54
Ma
y2
00
5T
PT
11
82
40
.01
58
0.0
17
10
.01
24
1.3
83
,75
34
26
5.6
75
4.3
27
.51
0.6
0M
ay
20
05
TP
T2
18
48
0.0
16
80
.01
93
0.0
12
71
.52
3,0
21
51
46
.71
47
.52
1.7
7.7
1M
ay
20
05
CB
Bin
it2
50
0.0
20
40
.02
27
0.0
26
51
.02
05
05
16
21
0.1
49
.03
.84
.66
Ma
y2
00
5C
BB
con
25
00
.01
92
0.0
22
00
.02
14
0.9
96
55
81
74
7.8
29
.44
.04
.51
Ma
y2
00
5C
BB
T1
25
24
0.0
20
80
.03
09
0.0
31
51
.24
35
0N
od
ata
14
.36
7.6
2.3
No
da
taM
ay
20
05
CB
BT
22
54
80
.02
11
0.0
36
10
.03
06
1.8
72
61
17
62
6.8
96
.71
.63
.17
Oct
20
05
GB
init
11
00
.01
54
0.0
13
70
.01
66
0.8
24
8,4
18
No
da
ta4
.79
11
3.1
64
.1N
od
ata
Oct
20
05
GB
con
11
00
.01
56
0.0
13
70
.01
69
0.8
13
8,2
77
No
da
ta4
.85
11
1.8
63
.2N
od
ata
Oct
20
05
GB
T1
11
24
0.0
14
90
.01
56
0.0
15
21
.02
7,4
99
No
da
ta5
.12
10
4.9
55
.8N
od
ata
Oct
20
05
GB
T2
11
48
0.0
14
20
.01
63
0.0
14
41
.13
7,0
77
No
da
ta5
.09
99
.35
1.7
No
da
taO
ct2
00
5T
Pin
it1
80
0.0
15
60
.01
43
0.0
16
80
.85
04
,47
9N
od
ata
4.9
66
1.5
33
.9N
od
ata
Oct
20
05
TP
con
18
00
.01
56
0.0
14
20
.01
67
0.8
50
4,4
84
No
da
ta4
.94
61
.43
3.9
No
da
taO
ct2
00
5T
PT
11
82
40
.01
50
0.0
17
10
.01
49
1.1
53
,74
1N
od
ata
5.5
35
4.5
27
.4N
od
ata
Oct
20
05
TP
T2
18
48
0.0
15
90
.01
95
0.0
15
51
.26
2,9
97
No
da
ta6
.56
47
.52
1.6
No
da
taO
ct2
00
5C
BB
init
23
00
.01
95
0.0
22
10
.01
84
1.2
05
91
No
da
ta8
.85
10
.14
.3N
od
ata
962 Helms et al.
magnitude and direction to those observed along theestuary (Fig. 7B,D). The change in absorption observedduring photobleaching was considerably smaller thanobserved along the estuary, but does not incorporatemixing of low-absorbing water as occurs in the estuary. It isinteresting to note that the duration of this experiment isconsiderably shorter than the freshwater residence time(,3.3 months) in the Delaware estuary (Dettmann 2001).We can therefore qualitatively attribute a considerable rolein the spectral alteration of CDOM to photochemicaldegradation. This result is similar to those from morequantitative modeling studies for temperate UK estuaries,where photobleaching within the estuaries was estimated toaccount for up to 90% of the observed spectral slope shiftfrom the head to the mouth (Stubbins 2001). The DelawareEstuary results in Fig. 8 show a strong influence frommixing, but the Stubbins result and our results shown inFig. 7B,D raise an interesting question of whether marineDOM S and SR are higher because the source is different orbecause river DOM is substantially photobleached duringits transit to the ocean.
Coastal ocean transects—Near-surface (,2 m) measure-ments of SR were 1.3 at the mouth of the Delaware Bay(38.86uN, 75.09uW), 1.5 at the shelf break (38.71uN,73.77uW), 3.9 midway across the continental slope(40.12uN, 70.60uW), and 9.4 in the Sargasso Sea(36.43uN, 71.41uW). Comparing two locations in theGAB (Table 4), the near-shore site (about 50 km off thecoast; 32.17uN, 79.99uW) showed average SR values of 1.7(60.1; n 5 6) in surface water samples in summer 2003.Farther offshore, near the shelf break (about 140 kmoffshore; 31.76uN, 79.28uW), SR values in surface watersaveraged 4.6 (60.1; n 5 2). The markedly higher SR valuesmeasured at the offshore site are consistent with ourmeasurements in the Sargasso Sea (Helms 2006) and withirradiation experiments.
Irradiation of the near-shore GAB sample (Sta. B) byeither natural sunlight or a solar simulator resulted in anincrease in S275–295, SR, and E2 : E3 and a decrease in S350–
400, S300–700, a254, a300, and E4 : E6, whereas no changes wereobserved in the dark control. These photochemicallyinduced shifts in the spectral parameters for the near-shoresample resulted in water with spectral properties approach-ing those of the offshore samples (Sta. C) (Tables 4, 5). Thechanges in spectral parameters are consistent with thoseobtained for the irradiated Elizabeth River and ChesapeakeBay samples, which showed that S275–295 values increased,S350–400 values decreased, SR and E2 : E3 increased, andUV-visible absorption decreased upon irradiation. Inaddition, irradiation of terrestrially dominated low-salinitysamples resulted in CDOM with optical properties charac-teristic of the nonirradiated marine-dominated high-salinitysamples (Fig. 3; Tables 2, 5).
Discussion
Comparison of the UF data and CDOM a spectraobtained from Elizabeth River and Chesapeake Baysamples showed that S275–295 and SR are inversely related
Sa
mp
len
am
eS
ali
nit
yIr
rt
(h)
S3
00
–7
00
(nm
21)
S2
75
–2
95
(nm
21)
S3
50
–4
00
(nm
21)
SR
To
tal
a(2
50
–4
50
nm
)D
OC
(mm
ol
L2
1)
E2
:E3
a2
54
(m2
1)
a3
00
(m2
1)
SU
VA
25
4(m
gC
L2
1
m2
1)
Oct
20
05
CB
Bco
n2
30
0.0
18
10
.02
17
0.0
18
71
.16
59
4N
od
ata
8.1
51
0.0
4.3
No
da
taO
ct2
00
5C
BB
T1
23
24
0.0
17
30
.02
65
0.0
14
41
.84
46
1N
od
ata
10
.18
8.6
3.1
No
da
taO
ct2
00
5C
BB
T2
23
48
0.0
16
80
.02
90
0.0
14
12
.06
39
0N
od
ata
11
.39
7.8
2.5
No
da
taF
eb2
00
6D
So
xco
n0
00
.01
58
0.0
13
10
.01
86
0.7
06
5,4
71
No
da
ta4
.64
90
8.1
52
0.5
No
da
taF
eb2
00
6D
So
xT
10
14
40
.01
45
0.0
17
00
.01
58
1.0
73
8,9
75
No
da
ta4
.97
55
4.5
27
5.3
No
da
taF
eb2
00
6D
Sa
no
xco
n0
00
.01
55
0.0
13
10
.01
80
0.7
26
7,7
02
No
da
ta4
.57
90
0.7
518
.2N
od
ata
Feb
20
06
DS
an
ox
T1
01
44
0.0
14
90
.01
53
0.0
17
01
.11
57
,25
8N
od
ata
4.7
17
86
.542
1.5
No
da
ta
Table
2.
Co
nti
nu
ed.
CDOM molecular weight and sources 963
Table
3.
Sp
ectr
al
slo
pe
(S),
tota
la
bso
rba
nce
(a),
an
dd
isso
lved
org
an
icca
rbo
n(D
OC
)m
easu
red
for
each
ult
rafi
ltra
tio
nsi
zefr
act
ion
of
the
Ch
esa
pea
ke
Ba
yB
rid
ge
(CB
B),
To
wn
Po
int
(TP
),G
rea
tB
rid
ge
(GB
),a
nd
Dis
ma
lS
wa
mp
(DS
);in
it5
init
ial;
con
5d
ark
con
tro
l;T
15
irra
dia
ted
24
h;
T2
54
8h
.
Sa
mp
len
am
eS
ali
nit
yIr
rt
(h)
UF
filt
rate
UF
rete
nta
te
S2
75
–2
95
(nm
21)
S3
50
–4
00
(nm
21)
SR
To
tal
a(2
50
–4
50
nm
)
DO
C(m
mo
lL
21)
S2
75
–2
95
(nm
21)
S3
50
–4
00
(nm
21)
SR
To
tal
a(2
50
–4
50
nm
)v
olu
me
corr
ecte
d
DO
Cv
olu
me
corr
ecte
d(m
mo
lL
21)
Ma
y2
00
4G
Bin
it1
20
0.0
18
80
.02
08
0.9
04
1,3
39
34
20
.01
33
0.0
16
30
.81
85
,16
85
59
Ma
y2
00
4G
Bco
n1
20
0.0
19
00
.02
06
0.9
20
1,1
84
33
60
.01
33
0.0
16
30
.81
35
,34
36
08
Ma
y2
00
4G
BT
11
22
40
.02
05
0.0
19
51
.05
1,3
68
35
10
.01
47
0.0
14
90
.98
34
,34
85
21
Ma
y2
00
4G
BT
21
24
80
.02
79
0.0
14
41
.94
1,1
55
28
70
.01
56
0.0
14
91
.05
3,4
17
49
7M
ay
20
04
TP
init
19
00
.01
75
0.0
17
70
.98
89
44
21
30
.01
29
0.0
15
50
.82
82
,14
32
64
Ma
y2
00
4T
Pco
n1
90
0.0
16
50
.01
65
1.0
07
89
17
70
.01
29
0.0
15
70
.82
22
,03
22
29
Ma
y2
00
4T
PT
11
92
40
.02
46
0.0
15
41
.60
88
72
51
0.0
15
80
.01
43
1.1
11
,60
92
49
Ma
y2
00
4T
PT
21
94
80
.02
74
0.0
15
61
.76
84
82
43
0.0
16
70
.01
43
1.1
71
,07
22
13
Ma
y2
00
4C
BB
init
23
00
.02
21
0.0
31
20
.70
83
11
11
30
.01
90
0.0
16
61
.14
23
87
4M
ay
20
04
CB
Bco
n2
30
0.0
18
00
.02
92
0.6
15
25
31
05
0.0
19
40
.01
64
1.1
82
62
90
Ma
y2
00
4C
BB
T1
23
24
0.0
27
80
.01
69
1.6
52
61
11
90
.02
22
0.0
15
91
.40
17
78
5M
ay
20
04
CB
BT
22
34
80
.02
63
0.0
02
11
2.8
18
91
15
0.0
25
40
.01
66
1.5
31
28
80
Oct
20
04
GB
init
10
00
.01
28
0.0
12
81
.08
1,7
62
33
30
.01
23
0.0
15
30
.80
29
,14
78
65
Oct
20
04
GB
con
10
00
.01
48
0.0
12
91
.14
2,0
66
38
30
.01
23
0.0
15
30
.80
78
,23
98
57
Oct
20
04
GB
T1
10
24
0.0
17
20
.01
29
1.3
42
,76
46
38
0.0
13
50
.01
37
0.9
88
7,1
69
65
4O
ct2
00
4G
BT
21
04
80
.01
96
0.0
11
61
.69
2,1
06
40
90
.01
50
0.0
14
21
.06
5,9
80
72
0O
ct2
00
4T
Pin
it1
70
0.0
13
40
.01
25
1.0
71
,48
22
79
0.0
12
50
.01
50
0.8
30
4,3
24
45
8O
ct2
00
4T
Pco
n1
70
0.0
14
80
.01
34
1.1
11
,51
62
92
0.0
12
20
.01
45
0.8
39
4,8
96
46
9O
ct2
00
4T
PT
11
72
40
.01
87
0.0
11
51
.63
1,6
15
27
40
.01
40
0.0
13
51
.04
3,3
98
38
6O
ct2
00
4T
PT
21
74
80
.02
11
0.0
11
51
.83
1,4
64
33
70
.01
54
0.0
13
51
.15
2,6
70
32
3O
ct2
00
4C
BB
init
26
00
.02
55
0.0
24
91
.02
33
41
43
0.0
20
50
.01
76
1.1
71
53
57
Oct
20
04
CB
Bco
n2
60
0.0
19
20
.02
30
0.8
34
28
31
16
0.0
19
90
.01
47
1.3
51
22
36
Oct
20
04
CB
BT
12
62
4N
od
ata
No
da
taN
od
ata
No
da
ta1
07
No
da
taN
od
ata
No
da
taN
od
ata
48
Oct
20
04
CB
BT
22
64
8N
od
ata
No
da
taN
od
ata
No
da
ta1
36
No
da
taN
od
ata
No
da
taN
od
ata
39
Ma
y2
00
5G
Bin
it8
00
.01
59
0.0
20
00
.79
51
,79
53
76
0.0
16
10
.01
50
1.0
74
,39
65
62
Ma
y2
00
5G
Bco
n8
00
.02
35
0.0
17
01
.38
1,9
58
38
70
.01
55
0.0
15
41
.00
4,4
03
56
8M
ay
20
05
GB
T1
82
40
.01
66
0.0
18
60
.88
91
,67
66
07
0.0
13
30
.01
68
0.7
89
4,5
15
51
2M
ay
20
05
GB
T2
84
80
.02
12
0.0
11
11
.90
1,5
56
39
20
.01
32
0.0
16
50
.80
12
,82
94
09
Ma
y2
00
5T
Pin
it1
80
0.0
18
30
.02
14
0.8
56
1,0
05
22
60
.01
30
0.0
16
60
.78
33
,18
53
20
Ma
y2
00
5T
Pco
n1
80
0.0
22
40
.01
83
1.2
21
,13
22
44
0.0
17
00
.01
69
1.0
03
,38
33
27
Ma
y2
00
5T
PT
11
82
40
.01
73
0.0
17
30
.99
71
,38
52
57
0.0
13
00
.01
61
0.8
09
2,0
84
22
0M
ay
20
05
TP
T2
18
48
0.0
20
70
.01
64
1.2
31
,16
25
33
0.0
14
70
.01
50
0.9
80
1,9
04
23
6M
ay
20
05
CB
Bin
it2
50
0.0
23
30
.02
09
1.1
22
32
90
No
da
taN
od
ata
No
da
taN
od
ata
37
Ma
y2
00
5C
BB
con
25
00
.02
34
0.0
22
01
.07
25
69
1N
od
ata
No
da
taN
od
ata
No
da
ta3
6M
ay
20
05
CB
BT
12
52
40
.02
83
0.0
16
51
.71
15
3N
od
ata
No
da
taN
od
ata
No
da
taN
od
ata
No
da
taM
ay
20
05
CB
BT
22
54
80
.02
95
0.0
18
31
.62
11
71
13
.8N
od
ata
No
da
taN
od
ata
No
da
ta3
2O
ct2
00
5G
Bin
it1
10
0.0
16
90
.02
11
0.8
01
1,6
20
No
da
ta0
.01
30
0.0
16
70
.77
86
,02
5N
od
ata
Oct
20
05
GB
con
11
00
.01
67
0.0
21
00
.79
51
,77
7N
od
ata
0.0
13
00
.01
67
0.7
78
6,0
56
No
da
taO
ct2
00
5G
BT
11
12
40
.01
95
0.0
18
31
.06
1,8
99
No
da
ta0
.01
45
0.0
15
30
.94
85
,021
No
da
taO
ct2
00
5G
BT
21
14
80
.02
11
0.0
19
01
.11
1,9
53
No
da
ta0
.01
52
0.0
15
11
.01
4,3
01
No
da
taO
ct2
00
5T
Pin
it1
80
0.0
17
40
.01
97
0.8
83
1,0
47
No
da
ta0
.01
32
0.0
16
80
.78
63
,01
5N
od
ata
Oct
20
05
TP
con
18
00
.01
76
0.0
20
60
.85
41
,05
2N
od
ata
0.0
13
30
.01
70
0.7
82
2,7
80
No
da
taO
ct2
00
5T
PT
11
82
40
.02
20
0.0
19
61
.12
1,0
63
No
da
ta0
.01
60
0.0
15
31
.05
2,3
88
No
da
ta
964 Helms et al.
to the MW of the CDOM in a water sample. A similartrend was also observed for SRNOM that was fractionatedby GFC. Using flow field–flow fractionation, Floge andWells (2007) demonstrated that the spectral shape covariedwith size for colloidal CDOM. That these relationships areobserved over a broad and continuous MW range of size-fractionated natural organic matter samples indicates thatvariations in spectral shape parameters with MW have aphysicochemical basis. The above results, plus the fact thatDOC-based MWs generally reflect the CDOM-based MWs(Fig. 4), indicate that UV spectral slopes have considerablepotential for semiquantitative assessment of DOM MW.However, further research is needed to determine whethersuch relationships between SR and average MW are systemdependent or more widely applicable.
Our UF data clearly indicate that spectral shifts in S andSR are accompanied by variations in both DOC and CDOMMW (e.g., Fig. 4, Tables 2, 3, 5). The decrease in CDOMMW with irradiation implies that chromophores associatedwith HMW CDOM are destroyed during photobleaching,resulting in a significant portion of the CDOM shifting fromthe HMW pool to the LMW pool (e.g., due to bond cleavageor disaggregation [or both]). These results are consistent withbut cannot distinguish between the two current conceptualmodels for CDOM spectral shape, i.e., a polydispersedmixture of discrete chromophores versus intramolecularcharge transfer. If CDOM behaves as a polydispersedmixture of conjugated organic chromophores, then theLMW material should absorb shorter-wavelength radiation(Pavia et al. 1979); therefore, when the MW distribution isshifted to lower MW by photodegradation, the shape of theabsorption spectrum should shift to greater relative absorp-tion at shorter wavelengths. On the other hand, if theprimary factor controlling CDOM optical properties is theintramolecular charge transfer capability of CDOM (DelVecchio and Blough 2004b), then a decrease in MW wouldalter (and probably diminish) the potential for intramolec-ular charge transfer interactions, which again would shift theabsorption spectrum toward shorter wavelengths, causingsteeper S275–295, shallower S350–400, and increased SR.
The large change in the spectral shape during irradiationof Dismal Swamp water at low oxygen levels (4% largerincrease in SR in the low-oxygen sample) concomitant witha smaller shift in MW (70% smaller increase in %LMW inthe low-oxygen sample) than occurred during air-saturatedirradiations (Table 2) suggests that (1) the CDOM chargetransfer system (Del Vecchio and Blough 2004b) wassignificantly altered by photochemical reactions, even atlow oxygen concentrations, and (2) oxidative cleavage ofcovalent bonds, probably by reactive oxygen species(Andrews et al. 2000), appears to be more impeded undersuboxic conditions than the disruption of DOM donor–acceptor charge transfer groups. Thus the decoupling ofMW, S, and SR in this experiment is more consistent withthe charge transfer model. Further study of the wavelengthdependence of spectral slope shifts and the compositionalchanges accompanying these photochemically inducedshifts is clearly needed.
Estuarine changes in spectra resulted in increases in SR
by nearly a factor of two between the DOM-rich terrestrial
Sa
mp
len
am
eS
ali
nit
yIr
rt
(h)
UF
filt
rate
UF
rete
nta
te
S2
75
–2
95
(nm
21)
S3
50
–4
00
(nm
21)
SR
To
tal
a(2
50
–4
50
nm
)
DO
C(m
mo
lL
21)
S2
75
–2
95
(nm
21)
S3
50
–4
00
(nm
21)
SR
To
tal
a(2
50
–4
50
nm
)v
olu
me
corr
ecte
d
DO
Cv
olu
me
corr
ecte
d(m
mo
lL
21)
Oct
20
05
TP
T2
18
48
0.0
24
10
.01
83
1.3
21
,02
4N
od
ata
0.0
17
70
.01
62
1.0
91
,65
8N
od
ata
Oct
20
05
CB
Bin
it2
30
0.0
22
40
.01
90
1.1
82
92
No
da
ta0
.02
06
0.0
18
51
.11
32
3N
od
ata
Oct
20
05
CB
Bco
n2
30
0.0
22
90
.02
03
1.1
34
02
No
da
ta0
.01
87
0.0
17
11
.09
18
9N
od
ata
Oct
20
05
CB
BT
12
32
40
.02
74
0.0
19
71
.39
27
8N
od
ata
0.0
24
30
.01
46
1.6
61
87
No
da
taO
ct2
00
5C
BB
T2
23
48
0.0
28
60
.01
35
2.1
22
66
No
da
ta0
.02
70
0.0
15
71
.72
14
6N
od
ata
Feb
20
06
DS
ox
00
0.0
18
00
.02
67
0.6
76
6,1
41
No
da
ta0
.01
27
0.0
17
20
.73
95
8,0
74
No
da
taF
eb2
00
6D
So
xT
10
14
40
.02
09
0.0
20
61
.01
9,4
66
No
da
ta0
.01
42
0.0
13
51
.05
30
,13
4N
od
ata
Feb
20
06
DS
an
ox
00
0.0
17
80
.02
52
0.7
06
5,0
73
No
da
ta0
.01
28
0.0
17
40
.73
55
9,6
03
No
da
taF
eb2
00
6D
Sa
no
xT
10
14
40
.01
93
0.0
20
00
.96
36
,85
8N
od
ata
0.0
13
80
.01
50
0.9
24
49
,03
1N
od
ata
Table
3.
Co
nti
nu
ed.
CDOM molecular weight and sources 965
freshwater end member and the surface waters from nearthe mouth of the Delaware Estuary. These shifts areprobably related to changes in DOM molecular size andcomposition during transit through the estuary. In addi-tion, CDOM spectral properties are affected by ionicstrength and pH shifts during estuarine transit (Minor et al.2006, unpubl. data). However, on the basis of previousphotochemical modeling studies in other midlatitudeestuaries (Stubbins 2001), the spectral changes that weobserved in the Delaware were probably caused byphotobleaching of DOM (e.g., photo-oxidative degrada-tion), mixing, and, to a lesser extent, flocculation, microbial
alteration or production of CDOM, and inputs frommarshes and sediments (especially in the bay). SR istherefore a potentially useful integrative indicator ofCDOM history (source and transformations) in naturalwaters.
The ratio of spectral slopes (SR) is a fast andreproducible method for characterizing CDOM in naturalwaters. Spectral corrections and calculations used to obtainSR are considerably simpler than those needed to determineexcitation–emission matrices (EEMs). Furthermore, SR
values can be measured across diverse aquatic regimes(e.g., from the black waters of the Great Dismal Swamp tohighly photobleached Sargasso Sea surface waters) andprovide strong differentiation between open ocean, coastal
Fig. 5. Regression plot of (A and B) spectral slopes and (C)ratio of spectral slopes, SR, versus apparent MW (weight average)of size-fractionated Suwannee River natural organic matter(SRNOM) as determined by size exculsion chromatography.Horizontal bars represent standard deviation. The open triangleindicates the unfractionated SRNOM datum. (A) Slope 5 21.573 1026, intercept 5 0.0147, r2 5 0.896; (B) slope 5 22.12 3 1026,intercept 5 0.0206, r2 5 0.9195; (C) slope 5 21.32 3 1025,intercept 5 0.7331, r2 5 0.9141.
Fig. 6. Regression plot of (A and B) spectral slopes and (C)ratio of spectral slopes, SR, versus incubation (aerobic) time ofunfiltered Elizabeth River (ODU Sailing Basin) water. Error barsrepresent standard deviation from replicate subsamples (n 5 3)that were poisoned and stored in separate vials. (A) Slope 5 26.03 1025, intercept 5 0.021, r2 5 0.807; (B) slope 5 6.0 3 1025,intercept 5 0.021, r2 5 0.997; (C) slope 5 25.5 3 1023, intercept5 1.0, r2 5 0.940.
966 Helms et al.
ocean, estuarine, and riverine surface waters. For example,between the Delaware River and the Sargasso Sea, SR
varies by more than a factor of 11 (and by over a factor of13 between DOM-rich Dismal Swamp water and SargassoSea surface water). The SR term, because of its indepen-dence from DOC concentration, may therefore be usefulfor differentiating open ocean waters from those of near-shore coastal or estuarine origin.
A potentially important application of SR is in theanalysis of ballast water. To limit the proliferation ofinvasive aquatic species (Choi et al. 2005), seafaring vesselsare required to exchange their ballast water at least200 miles from shore (Murphy et al. 2003). Our spectralanalysis provides a rapid analytical tool for differentiatinginland waters from oceanic regimes with considerableaccuracy and reproducibility, and it may be a simplealternative or supplement to using EEMs. However, furtherresearch is needed to determine whether our approach isrobust enough to be effective in an environment likely tocontain large amounts of petroleum and metal contamina-tion. Other potential applications for the two-slopeapproach include detecting zones of enhanced microbialheterotrophy and tracing recently subducted, photo-bleached surface water, e.g., 18u subtropical-mode waterformation (Helms 2006).
A consistent set of definitions for describing the shapeand wavelength ranges of log-transformed CDOM spectrais clearly needed to facilitate comparison of field data and
experimental results in the literature. Currently, noconsensus exists. In this study, we have identified twolinear wavelength regions that appear applicable to a widevariety of water types, including DOM-rich terrestrialwater, estuarine waters, coastal waters, and highly photo-bleached, DOM-poor open oceanic water, as well as size-fractionated samples. Reliable measurement of absorptionat wavelengths longer than ,400 nm are problematic formany instruments. Several approaches have been used todeal with this problem when calculating CDOM spectralslopes in the visible region. The use of variable wavelengthcutoffs has resulted in data that are very difficult tocompare (Twardowski et al. 2004). The use of nonlinearregression to calculate S300–700 has improved this situation.However, it requires considerably more computing powerand data processing than our approach for calculatingS350–400. The slope of the short wavelength range (275–295 nm), which can be measured with high precision (evenin highly photobleached open ocean waters), appears to beparticularly sensitive to, and perhaps also indicative of,shifts in MW or DOM sources (or both), whereas the use ofbroader wavelength ranges (e.g., several hundred nanome-ters), as is commonly practiced, is not sensitive enough toidentify these shifts. Therefore, we strongly advocate theinclusion of this short wavelength slope in future DOMstudies, as it would not only provide insights into DOMMW and source, but also greatly facilitate comparisonsamong investigations and water types.
Fig. 7. Spectral parameters measured within the Delaware Estuary and in rooftop-irradiated samples of Delaware River water(Yardley, Pennsylvania Public Boat Ramp, 80 liters, summer 2006): (A) spectral slope coefficients along the estuary (summer 2002) and(B) in irradiated river sample, (C) ratio of spectral slopes, SR (summer 2000, 2002, and 2005), and (D) in irradiated river sample.
CDOM molecular weight and sources 967
References
ANDREWS, S. S., S. CARON, AND O. C. ZAFIRIOU. 2000.Photochemical oxygen consumption in marine waters: Amajor sink for colored dissolved organic matter? Limnol.Oceanogr. 45: 267–277.
BROWN, M. 1977. Transmission spectroscopy examinations ofnatural waters. Estuar. Coast. Mar. Sci. 5: 309–317.
CARDER, K. L. R. G., STEWARD, G. R. HARVEY, AND P. B. ORTNER.1989. Marine humic and fulvic acids: Their effects on remotesensing of ocean chlorophyll. Limnol. Oceanogr. 34: 68–81.
CHEN, Y., N. SENESI, AND M. SCHNITZER. 1977. Informationprovided on humic substances by E4:E6 ratios. Soil Sci. Soc.Am. J. 41: 352–358.
CHIN, Y. P., G. AIKEN, AND E. O’LOUGHLIN. 1994. Molecularweight, polydispersity, and spectroscopic properties of aquat-ic humic substances. Environ. Sci. Technol. 26: 1853–1858.
CHOI, K. H., W. KIMMERER, G. SMITH, G. RUIZ, AND K. LION.2005. Post-exchange zooplanktion in ballast water of shipsentering the San Francisco Estuary. J. Plankton Res. 27:707–714.
DE HAAN, H., AND T. DE BOER. 1987. Applicability of lightabsorbance and fluorescence as measures of concentrationand molecular size of dissolved organic carbon in humicLaken Tjeukemeer. Water Res. 21: 731–734.
DEL VECCHIO, R., AND N. V. BLOUGH. 2002. Photobleaching ofchromophoric dissolved organic matter in natural waters:Kinetics and modeling. Mar. Chem. 78: 231–253.
———, AND ———. 2004a. Spatial and seasonal distribution ofchromophoric dissolved organic matter and dissolved organiccarbon in the Middle Atlantic Bight. Mar. Chem. 89:169–187.
———, AND ———. 2004b. On the origin of the optical propertiesof humic substances. Environ. Sci. Technol. 38: 3885–3891.
DETTMANN, E. H. 2001. Effect of water residence time on annualexport and denitrification of nitrogen in estuaries: A modelanalysis. Estuaries 24: 481–490.
FLOGE, S. A., AND M. L. WELLS. 2007. Variation in colloidalchromophoric dissolved organic matter in the DamariscottaEstuary, Maine. Limnol. Oceanogr. 52: 32–45.
GAO, H. Z., AND R. G. ZEPP. 1998. Factors influencingphotoreactions of dissolved organic matter in coastal riverof the southern United States. Environ. Sci. Technol. 32:2940–2946.
GREEN, S. A., AND N. V. BLOUGH. 1994. Optical absorption andfluorescence properties of chromophoric dissolved organicmatter in natural waters. Limnol. Oceanogr. 39: 1903–1916.
Fig. 8. Measured values and ideal mixing lines for (A) S275–295,(B) S350–400, and (C) SR in the Delaware Estuary (summer 2002).Salinity 5 1 was used for freshwater end member because offlocculation, particle adsorption, or apparent third end member atlower salinities.
Table 5. Spectral parameters of Georgia Bight (Sta. B)surface water samples altered by irradiation with natural andartificial sunlight.
Darkcontrol
8 h of naturalsunlight
18 h of artificialsunlight
S300–700 (nm21) 0.0150 0.0132 0.0139S275–295 (nm21) 0.025 0.028 0.032S350–400 (nm21) 0.015 0.013 0.013SR 1.705 2.082 2.42E2 : E3 8.621 8.967 11.2E4 : E6 4.3 4.095 3.372a254 (m21) 3.24 3.228 2.913a300 (m21) 1.23 1.11 0.840
Table 4. Average spectral parameters (61 SD) for two sitesin the Georgia Bight (GAB; June 2003). For Sta. B, n 5 6; for Sta.C, n 5 2. Replicate measurements were from separate samplestaken from the same station and depth.
Nearshore GAB(B) Offshore GAB(C)
S300–700 (nm21) 0.0148 6 0.0006 0.0086 6 0.0009S275–295 (nm21) 0.024 6 0.002 0.036 6 0.001S350–400 (nm21) 0.0135 6 0.0003 0.0078 6 0.0001SR 1.74 6 0.15 4.56 6 0.13E2 : E3 8.7 6 1.4 13.5 6 1.6E4 : E6 4.63 6 1.0 2.71 6 1.0a254 (m21) 3.368 6 0.001 1.78 6 0.06a300 (m21) 1.33 6 0.15 0.400 6 0.034
968 Helms et al.
HAYASE, K., AND H. TSUBOTA. 1985. Sedimentary humic-acid andfulvic-acid as fluorescent organic materials. Geochim. Cos-mochim. Acta 49: 159–163.
HELMS, J. R. 2006. Spectral shape as an indicator of molecularweight in chromophoric dissolved organic matter. M.S. thesis,Old Dominion Univ.
LEIFER, A. 1988. The kinetics of environmental aquaticphotochemistry: Theory and practice. American ChemicalSociety.
MILLER, W. L. 1994. Recent advances in the photochemistry ofnatural dissolved organic matter p. 111–128. In G. R. Helz,R. G. Zepp, and D. G. Crosby [eds.], Aquatic and surfacephotochemistry. CRC Press.
MINOR, E. C., J. POTHEN, B. J. DALZELL, H. ABDULLA, AND K.MOPPER. 2006. Effects of salinity changes on the photodeg-radation and UV-visible absorbance of terrestrial dissolvedorganic matter. Limnol. Oceanogr. 51: 2181–2186.
MOPPER, K., AND D. J. KIEBER. 2002. Photochemistry and thecycling of carbon, sulfur, nitrogen and phosphorusp. 455–489. In D. Hansell and C. Carlson [eds.], Biogeo-chemistry of marine organic matter. Academic Press.
MORAN, M. A., W. M. SHELDON, JR., AND R. G. ZEPP. 2000.Carbon loss and optical property changes during long-termphotochemical and biological degradation of estuarineorganic matter. Limnol. Oceanogr. 45: 1254–1264.
MORRIS, D. P., AND B. R. HARGRAEVES. 1997. The role ofphotochemical degradation of dissolved organic carbon inregulating the UV transparency of three lakes on the PoconoPlateau. Limnol. Oceanogr. 42: 239–249.
MURPHY, K., AND oTHERS. 2003. Verification of mid-ocean ballastwater exchanging using naturally occurring coastal tracers.Mar. Poll. Bull. 48: 711–730.
PAVIA, D. L., G. M. LAMPMAN, AND G. S. KRIZ. 1979. Introductionto spectroscopy: A guide for students of organic chemistry.Saunders.
PEURAVOURI, J., AND K. PIHLAJA. 1997. Molecular size distributionand spectroscopic properties of aquatic humic substances.Anal. Chim. Acta 337: 133–149.
PICCOLO, A., P. ZACCHEO, AND P. G. GENEVINI. 1992. Chemicalcharacterization of humic substances extracted from organic-waste-amended soils. Biores. Technol. 40: 275–282.
RITCHIE, J. D. 2005. The distribution of charge and acidicfunctional groups in the Suwannee River natural organicmatter: The dependence on molecular weight and pH. Ph.D.thesis, Georgia Institute of Technology.
SARPAL, R. S., K. MOPPER, AND D. J. KEIBER. 1995. Absorbanceproperties of dissolved organic matter in Antarctic sea water.Antarc. J. 30: 139–140.
SCHMIT, K. H., AND M. J. M. WELLS. 2002. Preferential adsorptionof fluorescing fulvic and humic acid components on activatedcarbon using flow field-flow fractionation analysis. J.Environ. Monit. 4: 75–84.
SCHWARZ, J. N., AND oTHERS. 2002. Two models for absorption bycoloured dissolved organic matter (CDOM). Oceanologia 44:209–241.
SENESI, N., T. M. MIANO, M. R. PROVENZANO, AND G. BRUNETTI.1989. Spectroscopic and compositional comaparative charac-terization of I.H.S.S. reference and standard fulvic and humicacids of various origin. Sci. Tot. Environ. 81/82: 143–156.
STEDMON, C. A., S. MARKAGER, AND H. KAAS. 2000. Opticalproperties and signatures of chromophoric dissolved organicmatter (CDOM) in Danish coastal waters. Estuar. Coast.Shelf Sci. 51: 267–278.
STUBBINS, A. P. 2001. Aspects of aquatic CO photoproductionfrom CDOM. PhD. thesis, New Castle Upon Tyne Univ.
———. 2006. Open-ocean carbon monoxide photoproduction.Deep Sea Res. II. 53: 1695–1705.
SUMMERS, R. S., P. K. CORNEL, AND P. V. ROBERTS. 1987.Molecular size distribution and spectroscopic characterizationof humic substances. Sci. Tot. Environ. 62: 27–37.
TWARDOWSKI, M. S., E. BOSS, J. M. SULLIVAN, AND P. L.DONAGHAY. 2004. Modeling the spectral shape of absorbingchromophoric dissolved organic matter. Mar. Chem. 89:69–88.
VAHATALO, A. V., AND R. G. WETZEL. 2004. Photochemical andmicrobial decomposition of chromophoric dissolved organicmatter during long (months-years) exposures. Mar. Chem. 89:313–326.
WALSH, J. J., AND oTHERS. 2003. Phytoplankton response tointrusions of slope water on the West Florida Shelf: Modelsand observations. J. Geophys. Res. Oceans 108: 21–31.
WEISHAAR, J. L., G. R. AIKEN, B. A. BERGAMASCHI, M. S. FRAM,R. FUGII, AND K. MOPPER. 2003. Evaluation of specificultraviolet absorbance as an indicator of the chemicalcomposition and reactivity of dissolved organic carbon.Environ. Sci. Technol. 37: 4702–4708.
WHITEHEAD, R. F., S. DE MORA, S. DEMERS, M. GOSSELIN, P.MONFORT, AND B. MOSTAJIR. 2000. Interactions of ultraviolet-B radiation, mixing, and biological activity on photobleachingof natural chromophoric dissolved organic matter: A meso-cosm study. Limnol. Oceanogr. 45: 278–291.
ZEPP, R. G., AND P. F. SCHLOTZHAUER. 1981. Comparison ofphotochemical behavior of various humic substances in water:3. Spectroscopic properties of humic substances. Chemo-sphere 10: 479–486.
Received: 14 May 2007Accepted: 22 December 2007
Amended: 7 January 2008
CDOM molecular weight and sources 969