the freshwater dissolved organic matter fluorescence–total organic carbon relationship
TRANSCRIPT
HYDROLOGICAL PROCESSESHydrol. Process. 21, 2093–2099 (2007)Published online 4 June 2007 in Wiley InterScience(www.interscience.wiley.com) DOI: 10.1002/hyp.6371
The freshwater dissolved organic matter fluorescence–totalorganic carbon relationship
Susan A. Cumberland and Andy Baker*School of Geography, Earth and Environmental Sciences, The University of Birmingham, Birmingham B15 2TT, UK
Abstract:
The fluorescent properties of dissolved organic matter (DOM) enable comparisons of humic-like (H-L) and fulvic-like (F-L)fluorescence intensities with dissolved organic carbon (DOC) in aquatic systems. The fluorescence-DOC relationship differedin gradient, i.e. the fluorescence per gram of carbon, and in the strength of the correlation coefficient. We compare thefluorescence intensity of the F-L and H-L fractions and DOC of freshwater DOM in north Shropshire, England, featuring ariver, wetland, spring, pond and sewage DOM sources. Correlations between fluorescence and DOC varied between samplesites. Wetland water samples for the F-L peak gave the best correlation, r D 0Ð756; the lowest correlation was from finaltreated sewage effluent, r D 0Ð167. The relationship between fluorescence and DOC of commercially available InternationalHumic Substances Society standards were also examined and they generally showed a lower fluorescence per gram of carbonfor the F-L peak than the natural samples, whereas peat wetland DOM gave a greater fluorescence per gram of carbon thanriver DOM. Here, we propose the strength of the fluorescence–DOC correlation to be a useful tool when discriminatingsources of DOM in fresh water. Copyright 2007 John Wiley & Sons, Ltd.
KEY WORDS dissolved organic carbon; dissolved organic matter; fluorescence; wetlands; rivers
Received 20 December 2005; Accepted 22 May 2006
INTRODUCTION
Humic substances occur in every natural water samplethat has ever been analysed for them (Malcolm, 1985).Dissolved humic substances can be defined as passingthrough a 0Ð45 µm membrane, and are made up of humicand fulvic substances, the latter being present in largerproportion and characterized as being soluble under allpH conditions (Aiken, 1985). Freshwater dissolved humicsubstances are derived from dead and decaying detritus,aquatic plants, animals and debris from overhanging veg-etation, and tend to be of a younger state than that ofsoil humic substances (Hongve, 1999). River aquatic ful-vic substances are derived from plant and tree residues,which contain more phenolic and lignin-derived organiccompounds than those found in soil (Chen et al., 2003).Historically, much work has gone into the determinationand separation of these humic and fulvic substances, andyet still relatively little is understood about the molec-ular distribution and fluorescent characteristics of thesemolecules (Alberts and Takacs, 2004). Analysis of freeze-dried fulvic acid using 13C nuclear magnetic resonancespectroscopy demonstrates that it contains a number ofgroups, such as aliphatic carbon atoms, carbohydrates,olefinic carbon atoms, aromatic carbons atom, carboxylester amide carbon atoms and ketonic and aldehydeiccarbon atoms (Klapper et al., 2002; Sierra et al., 2005).They are responsible for water colour (Hongve, 1999) and
* Correspondence to: Andy Baker, School of Geography, Earth andEnvironmental Sciences, The University of Birmingham, BirminghamB15 2TT, UK. E-mail: [email protected]
sometimes pH due to the carboxylic acid groups attachedto them (Klapper et al., 2002). The aromatic and car-boxylic fractions provide the fluorescence of humic andfulvic substances (Senesi et al., 1989); fulvic substancesare more highly fluorescent than humic substances andhumic fluorescence intensity is more pH dependent(Goslan et al., 2004; Sierra et al., 2005). Owing to theambiguous characteristics of these substances, we feelthat it is more appropriate to refer to them, hereafter, ashumic-like (H-L) and fulvic-like (F-L) substances.
In freshwaters, it has been suggested that dissolvedorganic carbon (DOC) is usually dominated by the H-Land F-L fractions, making up 50–75% of the total (ofwhich colloidal matter comprises up to 20% of the totalDOC; Hope et al., 1994). However, the proportion of H-Land F-L substances in the DOC is variable. For example,about 50% of uncoloured (USA) stream water consistsmostly of humic substances and 90% of these humicsubstances found in dissolved organic matter (DOM)are the F-L fraction (Malcolm, 1985). In the Amazonbasin, humic substances are attributed to as much as60% of the aquatic total organic carbon (TOC; Patel-Sorrentino et al., 2002). In stream/aquatic systems, DOClosses are possible due to biotic processes such as biofilmrespiration and absorption onto algae and abiotic uptakeonto mineral surfaces (Dawson et al., 2001). Losses ofCO2 from volatile DOC fractions by degassing to theatmosphere, as well as variable DOM inputs throughoutand between catchments, can also lead to this variability.
The presence of fluorescent DOM enables the compar-ison of H-L and F-L fluorescence intensity with DOC.
Copyright 2007 John Wiley & Sons, Ltd.
2094 S. A. CUMBERLAND AND A. BAKER
There are many potential benefits of using fluorescence asa tool for determining and measuring DOC in aquatic sys-tems. It is a relatively quick process, with a typical scantaking approximately 1 min; this is valuable, as labileDOM can degrade over time. The method is non-intrusiveand thus does not interfere with the molecular structure(Klabitz et al., 2000; Baker, 2002). However, given thevariable fraction of fluorescent DOM as part of the totalDOC pool, it is essential to investigate the fluorescenceintensity–DOC relationship. Therefore, we compare thefluorescence intensity of the F-L fraction and DOC offreshwater DOM. The study site that this paper focuseson, Norton in Hales, in north Shropshire, England, is asite that features river, wetland, spring, pond and sewageDOM sources all within a small location (approximately500 m), covering a wide range of variation in aquatic andsoil-derived water samples.
SITE DESCRIPTION AND METHODOLOGY
Norton in Hales [UK National Grid Reference SJ706 384], north Shropshire, UK, is situated along theRiver Tern, a tributary of the River Severn, 16 kmfrom its source. The River Tern is a lowland catchment,primarily groundwater fed from red sandstone overly-ing the Permo-Triassic Sherwood sandstone. Its catch-ment boundary, thus, is not defined by its topography(described by Clay et al. (2004)). This amounts to avery stable base flow in the River Tern, showing min-imal response to normal rainfall events. Local land useis generally under semi-intense mixed cereal/dairy/beeffarming.
Figure 1 illustrates the diversity of the sample siteswithin a small area. In-stream water samples werecollected (river sites A, C, E and H), as a transect downthe river starting above the wetland (A) and finishingbelow the sewage treatment outflow pipes (H). Betweenthese points, samples were also taken from the sewage
treatment works (STW) outlet itself (G), a groundwater-fed pond overflow, POND (B), a wetland dischargesite (D) (not always connected to the main river) anda groundwater spring diverted through small concreteconduit (F). Additionally, three samples were taken fromthe wetland: surface water (I), 0–45 cm depth (J), and50–100 cm depth (K). To obtain samples from the deeperpeat, two perforated drain pipes, about 50 cm apart, weresunk (to 50 cm deep and 1 m deep, with the latter beingperforated from 50 to 100 cm) and sealed at both ends, toreduce oxidation and contamination from other externalsources. The sample was obtained by unscrewing a lidand dipping for the sample, throwing away the first fewdips so that the sample was taken from the middle of theperforated column depth. The wells were then completelyemptied before resealing. Thus, water in tubes was notallowed to stagnate, but perforation allowed ‘flow’ ofthe wetland and was representative of (actual) bog water.These samples (D, I, J and K) were grouped together toform BOG.
Water samples were collected fortnightly from Nor-ton in Hales from March to October 2005. Plastic bot-tles (500 ml), pre acid washed (10% HCl, distilled-waterrinsed), were used to ensure carbon- and fluorescence-free blanks (Bolton, 2004). Bottles were also pre-rinsedwith sample before filling. On return to the laboratory(same day) the samples were then filtered through pre-rinsed Whatman GF/C filter papers under vacuum. Sam-ples were kept in the refrigerator overnight between 4 and6 °C they were then analysed for fluorescence and TOCthe next day, with DOC analysis completed with 36 h ofcollection (on two occasions this was a day later due tomachine downtime); this was critical, as the fluorescenceproperties of a sample can alter over time (Baker, 2002).
DOC was measured using a Shimadzu 5050 carbonanalyser from filtered and unfiltered (not reported here)samples. DOC was obtained by measuring the total car-bon and subtracting the measured inorganic carbon andcarbonates (dissolved CO2 and CO2�
3 ). As other studies
H GF
B
E
J K
I
C
A
Ponddrain
D
N
Sewage Works
Key
Telford BirminghamSJ 707 385
RiverTern
NORTON INHALES
Norton-in-Hales
0 50 100 metres
Bog
Water
Figure 1. Location of sample sites
Copyright 2007 John Wiley & Sons, Ltd. Hydrol. Process. 21, 2093–2099 (2007)DOI: 10.1002/hyp
FRESHWATER DOM FLUORESCENCE–TOC RELATIONSHIP 2095
have demonstrated, some methodologies suggest purgingwith acid to remove the inorganic fraction (Clark et al.,2002). This was not executed in this case because thetotal carbon and inorganic carbon concentrations are sig-nificantly different from each other, giving differencingerrors of <1 mg l�1 for DOC. Calibration was with cer-tified organic and inorganic standards from Reageacon,and samples were analysed in triplicate and the meantaken from the best two (CV <2%). Glass vials werewashed in 4% Decon 90 and 10% HCl and pre-rinsedwith ultra-pure water and sample.
A fluorescence scan for the organic fraction of the sam-ple was obtained from filtered samples within 48 h ofsample collection. DOM fluorescence was measured in4 ml capacity cuvettes using a Varian Cary Eclipse flu-orescence spectrophotometer equipped with a multicellholder with Peltier temperature controller to enable themeasurement of excitation–emission matrices (EEMs)at precisely controlled (š0Ð1 °C) temperatures at 20 °C.Each EEM was generated by scanning excitation wave-lengths from 200 to 400 nm at 5 nm steps and detectingthe emitted fluorescence between 280 and 500 nm at2 nm steps. Scan speed was 9600 nm min�1, permit-ting collection of a complete EEM in ¾60 s. For allEEMs, two fluorescence peaks were identified that werealways detectable. These were (1) fluorescence excitedbetween 300 and 340 nm excitation and emitted between400 and 460 nm, and (2) fluorescence excited between220 and 250 nm excitation and emitted between 400 and460 nm, both fluorophores attributed to F-L and H-L sub-stances (Baker and Inverarity, 2004). For each peak, theexcitation and emission wavelengths and the intensity ofemitted fluorescence were recorded. To calibrate the flu-orescence intensity, we also measured the strength of theRaman signal at excitation 348 nm (emitted between 395and 400 nm) and all results are standardized to a meanRaman peak of 20 intensity units. Our results can becompared with a quinine sulphate standard: 32Ð5 intensityunits are equivalent to one quinine sulphate unit (1 µg l�1
in 0Ð1 M H2SO4). Samples were not diluted, with theexception of one intensely fluorescent BOG sample, forwhich a ð2 dilution, to bring to scale, was used.
To be able to put our sample sites in a wider con-text, baseline geochemical and nutrient data have alsobeen collected (Table I). The pH was measured in thefield using a Jenway hand-held meter. Determinationof nutrients (nitrate, ammonium and phosphate) wasundertaken at the University of Reading using a Chem-Lab continuous flow autoanalyser. Samples were thenmicrowave digested (CEM Corporation) using the per-sulphate method as described by Johnes and Heathwaite,(1992) and then reanalysed to determine the organicfraction.
RESULTS
Table I presents summary fluorescence and DOC results,together with secondary geochemical and nutrient data.
Data from sites A, C, E, and H are combined to form‘RIVER’; sites D, I, J, and K are combined to form‘BOG’. The treated final effluent (STW), pond site B(POND) and groundwater site F (SPRG) were analysedwithout grouping.
For all groups, the excitation and emission wavelengthsand intensities of H-L and F-L fluorescence, as wellas DOC concentrations, fall within the range reportedfrom other rivers within the UK (Baker and Inverarity,2004; Baker and Spencer, 2004; Worrall et al., 2004),suggesting that DOM in the River Tern is typical ofrivers in the UK. Intensities for the river F-L peak rangedbetween 113 and 424 a.u., with a mean of 172 a.u. anda standard deviation of 69 a.u. Intensities for the BOGF-L peak ranged between 77 and 951 a.u., with a meanof 409 a.u. and a standard deviation of 208 a.u.
EEM fluorescence intensity values of the F-L peaks(fluorophores) for all samples were compared with mil-ligrams per litre DOC. The fluorescence per milligram perlitre of DOC is presented in Table I, and the correlationcoefficients between F-L fluorescence intensity and DOCin Figure 2, together with the fluorescence–DOC rela-tionship of commercially available International HumicSubstances Society (IHSS) standards: Nordic fulvic acid(referred to hereafter as Nordic FA), Suwanee River ful-vic acid (SRFA), Suwanee River humic acid (SRHA)and Suwanee River natural organic matter (SRNOM). Ingeneral, the Norton in Hales samples showed a higher flu-orescence per milligram of carbon for the fulvic peak thanthe IHSS standards, of which the Nordic FA showed themost fluorescence per gram of carbon of the IHSS stan-dards, then SRFA, then SRNOM, then SRHA displayingthe least intensity per unit carbon.
Correlations (product for normally distributed data;rank for non-normally distributed data) between fluores-cence and DOC varied between sample sites. BOG gavethe best correlation (r D 0Ð756), then POND (r D 0Ð657),SPRG (r D 0Ð631) and RIVER (r D 0Ð409), with theSTW giving the lowest correlation (r D 0Ð14). That BOGshowed less scatter than SPRG, POND and RIVER couldbe explained by its early stage of decomposition andrelative similarity in source and composition of the fluo-rescent molecules. The weak correlation observed in thesewage outlet could be attributed to other organic fluo-rescence chemicals, such as fluorescent whitening agentsthat are present in municipal sewage waste, and the lowgradient due to the relative insignificance of the F-Lfraction as a proportion of DOC compared with otherfluorophores.
Results for the H-L fluorophores for all samples werealso compared against DOC (Table I and Figure 3).H-L fluorophores generally had similar correlations withDOC than the F-L fluorescence. This is despite thefluorescence intensity for the H-L fluorophores beingmore dependent on pH (Patel-Sorrentino et al., 2002;Sierra et al., 2005), with fluorescence intensity increasingwith increasing pH (Mobed et al., 1996). The H-L peaksfor the BOG samples were harder to locate, as theyappear close or within the Rayleigh–Tyndall scatter line
Copyright 2007 John Wiley & Sons, Ltd. Hydrol. Process. 21, 2093–2099 (2007)DOI: 10.1002/hyp
2096 S. A. CUMBERLAND AND A. BAKER
Tabl
eI.
Fluo
resc
ence
,or
gani
can
din
orga
nic
carb
on,
nutr
ient
and
pHda
tafo
rth
efiv
esa
mpl
egr
oups
F-L
H-L
DO
C�m
gl�1
�In
t./D
OC
Org
anic
�mg
l�1�
C/N
pH
Ex.
(nm
)E
m.
(nm
)In
t.(a
.u.)
Ex.
(nm
)E
m.
(nm
)In
t.(a
.u.)
TC
ICD
OC
TO
Na
NO
2-N
NH
4-N
PO4-P
NP
All
data
Min
.30
040
626
219
408
7412
Ð960Ð3
00Ð2
90Ð0
00Ð0
00Ð0
00Ð0
00Ð4
9�0
Ð016Ð3
1M
ax.
344
450
951
255
461
1558
122Ð3
086
Ð7341
Ð6228
Ð920Ð3
25Ð6
54Ð6
74Ð8
90Ð7
18Ð1
5M
ean
331
425
220
237
428
366
48Ð47
37Ð53
10Ð95
20Ð14
8Ð05
0Ð05
0Ð51
0Ð52
2Ð10
0Ð09
1Ð28
7Ð30
SD8
816
47
722
517
Ð5713
Ð468Ð6
47Ð4
70Ð0
71Ð0
91Ð2
11Ð2
90Ð1
70Ð4
6N
o.29
229
229
229
129
129
114
814
814
829
2929
2929
2948
Riv
er Min
.31
541
711
323
541
421
723
Ð8318
Ð012Ð7
47Ð6
00Ð0
20Ð0
00Ð1
30Ð7
9�0
Ð016Ð8
2M
ax.
340
440
424
254
444
600
59Ð01
42Ð92
39Ð87
8Ð83
0Ð07
1Ð58
0Ð33
4Ð23
0Ð25
7Ð81
Mea
n33
042
717
223
942
830
942
Ð9434
Ð048Ð8
919
Ð387Ð6
40Ð0
40Ð4
10Ð1
82Ð0
70Ð0
71Ð1
17Ð5
4SD
75
695
688
6Ð23
5Ð51
7Ð54
2Ð33
0Ð02
0Ð65
0Ð09
1Ð42
0Ð10
0Ð27
No.
6464
6464
6464
6464
6412
1212
1212
1215
BO
G Min
.32
141
677
219
416
155
12Ð96
0Ð30
1Ð37
0Ð00
0Ð00
0Ð00
0Ð00
0Ð64
0Ð01
6Ð31
Max
.34
445
095
125
046
115
5812
2Ð386
Ð7341
Ð625Ð5
90Ð0
65Ð6
50Ð2
92Ð6
10Ð0
97Ð7
7M
ean
334
427
409
231
429
645
63Ð16
46Ð56
16Ð60
24Ð66
1Ð17
0Ð02
0Ð87
0Ð08
1Ð28
0Ð03
8Ð15
7Ð00
SD5
820
85
928
624
Ð4519
Ð899Ð9
02Ð7
50Ð0
31Ð8
80Ð1
11Ð2
40Ð0
80Ð4
6N
o.49
4949
4848
4949
4949
99
99
99
20
STW M
in.
320
406
177
240
420
284
22Ð42
15Ð91
4Ð79
14Ð87
0Ð18
0Ð17
2Ð21
3Ð72
0Ð25
7Ð01
Max
.34
042
631
125
043
640
746
Ð9735
Ð3317
Ð9528
Ð920Ð3
20Ð7
44Ð6
74Ð8
90Ð7
17Ð5
0M
ean
333
419
249
245
426
340
38Ð99
26Ð44
12Ð55
19Ð87
17Ð30
0Ð19
0Ð34
2Ð88
3Ð09
0Ð37
0Ð71
7Ð30
SD5
442
34
377Ð0
95Ð6
43Ð4
012
Ð990Ð1
40Ð3
22Ð2
32Ð1
30Ð3
10Ð2
3N
o.16
1616
1616
1616
1616
33
33
33
5
BM
in.
300
416
6823
041
615
929
Ð1625
Ð003Ð8
42Ð0
50Ð0
00Ð0
00Ð0
00Ð6
40Ð0
17Ð6
1M
ax.
330
436
112
245
434
255
51Ð08
40Ð68
14Ð65
2Ð06
0Ð08
0Ð04
0Ð01
0Ð89
0Ð00
8Ð15
Mea
n31
842
689
234
424
200
41Ð76
34Ð21
7Ð55
11Ð77
2Ð06
0Ð04
0Ð02
0Ð01
0Ð77
0Ð02
3Ð64
7Ð90
SD10
514
45
345Ð4
44Ð1
43Ð1
81Ð1
90Ð0
40Ð0
20Ð0
00Ð4
50Ð0
10Ð2
0N
o.14
1414
1414
1414
1414
22
22
22
5
FM
in.
315
412
4023
541
277
35Ð30
29Ð75
0Ð29
12Ð79
0Ð02
0Ð02
0Ð14
2Ð06
�0Ð01
6Ð82
Max
.34
042
816
025
143
826
059
Ð1146
Ð9016
Ð9618
Ð930Ð0
40Ð0
70Ð1
63Ð0
30Ð0
67Ð7
3M
ean
331
420
6524
442
411
241
Ð4736
Ð624Ð8
513
Ð4516
Ð790Ð0
30Ð0
40Ð1
52Ð6
00Ð0
20Ð2
97Ð2
1SD
65
285
847
6Ð20
5Ð14
4Ð31
8Ð86
0Ð02
0Ð03
0Ð07
1Ð36
0Ð03
0Ð38
No.
1616
1616
1616
1616
163
33
33
34
aTo
tal
oxid
ized
nitr
ogen
(nitr
ate
and
nitr
ite).
Copyright 2007 John Wiley & Sons, Ltd. Hydrol. Process. 21, 2093–2099 (2007)DOI: 10.1002/hyp
FRESHWATER DOM FLUORESCENCE–TOC RELATIONSHIP 2097
0 10 20 30 40 50 60
Inte
nsity
Inte
nsity
Inte
nsity
0
200
400
600
800
1000
0 10 20 30 40 50 600 10 20 30 40 50 600
200
400
600
800
1000
mgL−1 TOC
0 10 20 30 40 50 600
200
400
600
800
1000
0 10 20 30 40 50 60
r = 0.409 r = 0.756
r = 0.631 r = 0.657
r = 0.167
mgL−1 TOC
0 10 20 30 40 50 60
NORDIC r = 0.992SRFA r = 0.997SRHA r = 0.994SRNOM r = 0.998
(a) (b)
(c) (d)
(e) (f)
Figure 2. Comparisons of the fluorescence of F-L peak versus DOCfor: (a) RIVER, y D 7Ð92x C 103Ð15; (b) BOG, y D 15Ð87x C 121Ð03;(c) SPRG, y D 4Ð15x C 45Ð8; (d) POND, y D 2Ð89x C 67Ð77; (e) STW,
y D 2Ð09x C 219Ð82; (f) IHSS samples
(Ex/Em 230/460 nm). As for the F-L fluorescence, theNorton in Hales samples showed a higher fluorescenceper milligram of carbon for the H-L peak than the IHSSstandards.
DISCUSSION
Figures 2 and 3 demonstrate that H-L and F-L fluo-rophores have a correlation that varies with sample site.The strongest correlations are in the BOG and PONDsamples, where we hypothesize that natural H-L and F-Lmaterial dominates the DOM pool. In contrast, the weak-est relationship occurs in the final treated effluent, a watersource where H-L and F-L materials make up a relativelysmall component of DOM. These results match thoseof Baker (2002), investigating a small urban catchment.There, the F-L–DOC correlation for the whole catchmentwas r D 0Ð68, again with stronger correlations in the trib-utary containing greater proportions of natural DOM andpoorer correlations in the subcatchments where anthro-pogenic influences were stronger (Table II). We suggestthat the strength of the F-L–DOC correlation could beused as an indicator of anthropogenic influence on DOM.However, within-stream processing of DOM and mixingof DOM from different sources also has to be taken intoconsideration. For example, Clark et al. (2002) observedremarkable correlations (0Ð999) with DOM fluorescence
0 10 20 30 40 50 60 0 10 20 30 40 50 60
0 10 20 30 40 50 60 0 10 20 30 40 50 60
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Inte
nsity
Inte
nsity
Inte
nsity
0
200
400
600
800
1000
0
200
400
600
800
1000
mgL−1 TOC
0
200
400
600
800
1000
mgL−1 TOC
NORDIC r = 0.985SRFA r = 0.961SRHA r = 0.984SRNOM r = 0.956
r = 0.609
r = 0.463
r = 0.630
r = 0.504
r = 0.277
(a) (b)
(c) (d)
(e) (f)
Figure 3. Comparison of the fluorescence of H-L peak versus DOCfor: (a) RIVER, y D 9Ð42x C 224Ð62; (b) BOG, y D 14Ð42x C 377Ð3 (onedata point is off scale, DOC D 16Ð5, Int. D 1558, not shown tofacilitate equal scaling); (c) SPRG, y D 6Ð8621x � 80Ð69; (d) POND,y D 6Ð66x C 150Ð88; (e) STW, y D 2Ð919x C 299Ð69; (f) IHSS samples
and DOC when they went from a fresh water to a salinewater (Shark Head River to Florida Bay). Burdige et al.(2004) also reported good correlations from contrastingsites in the Chesapeake Bay, USA, with r D 0Ð92. Incontrast, the scatter observed in the BOG, SPRG andRIVER TOC–fluorescence intensity relationships indi-cates the variability and instability of natural DOM whensampled closer to its source, as originally hypothesized inthe river continuum concept (Vannote et al., 1980). Daw-son et al. (2001) showed changes in carbon fluxes in anupland stream (CO2 degassing) and that the source of thenatural organic matter (NOM) may be unstable due to itsstate of decomposition and tributary contribution. By thetime DOM reaches the estuary, within-stream process-ing and mixing of DOM from different sources leads toa final DOC–fluorescence relationship that is dominatedby the stable, recalcitrant fluorescent fraction, leading toa stronger correlation.
The F-L fluorescence per gram of carbon of our sam-ples (Table I and Figure 4) was shown to vary betweensample sites. The BOG samples, with arguably the fresh-est DOM, have a greater fluorescence per gram of carbonthan the river samples. F-L fluorescence per gram ofcarbon was also observed to be greater than the F-LIHSS standards, especially for the BOG samples site. Forthe H-L fluorescence per gram of carbon, this differencebetween our field samples and humic IHSS standards was
Copyright 2007 John Wiley & Sons, Ltd. Hydrol. Process. 21, 2093–2099 (2007)DOI: 10.1002/hyp
2098 S. A. CUMBERLAND AND A. BAKER
Table II. Comparison of the correlation coefficient between TOC and F-L fluorescence intensity with other studies
Study Site ID Description Correlation coefficient Comments
Ouseburn (Baker, 2002)Tributaries 4 Harey Burn 0Ð45 Impacted, sewerage failures
11 Brunton tributary 0Ð78 Unimpacted, agricultural land cover17 Gosforth Park tributary 0Ð53 Impacted from occasional CSO discharges21 The Letch 0Ð75 Unimpacted, suburban and agriculture
22a Town Moor Tributary 0Ð78 Unimpacted, grassland and suburban6 Abbotswood Burn �0Ð20 Impacted, cross connected sewers8 Airport tributary �0Ð30 Impacted, airport deicer
Main stream 8 Woolsington 0Ð82 Unimpacted, agricultural land cover10 Brunton Bridge �0Ð34 Downstream tributaries 6 & 813 Red House Farm �0Ð17 Downstream of tributary 1116 Great N Road �0Ð29 Downstream of tributary 1722 Castle Dean gauge 0Ð42 Downstream of tributaries 21 & 22
River Tern (this study) RIVER 0Ð87BOG 0Ð76SPRNG 0Ð63POND 0Ð66STW FTE 0Ð17
Clark et al. (2002) Florida Bay 0Ð99 EstuaryBurdige et al. (2004) Chesapeake Bay 0Ð92 Estuary
mgL−1 DOC
0 10 20 30 40 50 60
Ful
vic
- lik
e In
tens
ity
0
200
400
600
800
1000
RIVERPONDSPRGBOGSTWNordic FASRFASRHASRNOM
Figure 4. DOC versus F-L fluorescence intensity for all samples, showingthe different fluorescence intensities per gram of carbon for different
carbon sources
even greater. These results suggest that a fluorescent frac-tion is lost in the IHSS standard preparation procedure,especially the humic (SRHA) standard. This is in agree-ment with previous investigations that have suggestedthat no ideal system is available for isolating pure hypo-thetical humic substances from a water sample, and thatcertain changes in the structural composition of the DOMtake place during each isolation procedure based on achemically assisted sorption–desorption technique (Peu-ravuori et al., 2005). Studies that use IHSS standards (e.g.Mobed et al., 1996; Klapper et al., 2002; Chen et al.,2003; Alberts and Tackacs, 2004; Sierra et al., 2005), and
in particular humic standards to quantify carbon fluxesfrom DOM fluorescence properties, could be consider-ably overestimating carbon quantities.
CONCLUSIONS
We demonstrate different fluorescence intensity–DOCrelationships for H-L and F-L fluorescence centres,depending on the source of DOM. These relationshipsdiffer in both gradient, i.e. the fluorescence per gramof carbon, and in the strength of the correlation coef-ficient. Strongest correlations are observed at sites wherenatural DOM dominates the DOC pool, whereas greaterfluorescence per gram of carbon is observed in our peatwetlands than with within-river DOM. IHSS standards,in particular humic extracts, have significantly lower flu-orescence per gram of carbon than our field samples.Assuming that all catchments have sufficient variabilityin both DOC and fluorescence, we propose the strengthof the fluorescence–TOC correlation to be another usefultool when discriminating sources of NOM in freshwaters.Even when there is little enough variability of DOC andfluorescence, our results show significantly different flu-orescence per gram of carbon between sources. Analysesof both fluorescence and DOC potentially fill a gap inthe available data required to identify the mechanisticprocess, and magnitudes of, aquatic carbon fluxes to theestuary (Eatherall et al., 1998).
ACKNOWLEDGEMENTS
We would like to thank Andy Moss, Richard Johnson andIan Morrissey for excellent laboratory and field support,Kevin Burkhill (Birmingham University) for Figure 1and Martin Heaps and Geoff Warren for laboratoryanalysis at Reading University. The project was funded
Copyright 2007 John Wiley & Sons, Ltd. Hydrol. Process. 21, 2093–2099 (2007)DOI: 10.1002/hyp
FRESHWATER DOM FLUORESCENCE–TOC RELATIONSHIP 2099
by the NERC LOCAR programme (to SC) and fundingfrom a Philip Leverhulme Prize (to AB). The manuscriptbenefited from the helpful comments of three anonymousreferees.
REFERENCES
Aiken GR. 1985. Isolation and concentration techniques for aquatichumic substances. In Humic Substances in Soil, Sediment andWater: Geochemistry, Isolation, and Characterization , Aiken GR,McKnight DM, Wershaw RL, MacCarthy P (eds). Wiley: New York;363–385.
Alberts JJ, Takacs M. 2004. Comparison of the natural fluorescencedistribution among size fractions of terrestrial fulvic and humicacids and aquatic natural organic matter. Organic Geochemistry 35:1141–1149. DOI: 10Ð1016/j.orggeochem.2004Ð06010.
Baker A. 2002. Spectrophotometric discrimination of river dis-solved organic matter. Hydrological Processes 16: 3203–3213.DOI:10Ð1002/hyp.1097.
Baker A, Inverarity R. 2004. Protein-like fluorescence intensity asa possible tool for determining river water quality. HydrologicalProcesses 18: 2927–2945. DOI:10Ð1002/hyp.5597.
Baker A, Spencer RGM. 2004. Characterization of dissolved organicmatter from source to sea using fluorescence and absorbancespectroscopy. The Science of the Total Environment 333: 217–232.DOI:10Ð1016/j.scitotenv.2004Ð04Ð013.
Bolton L. 2004. The application of excitation–emission fluorescencespectrophotometry to the monitoring of dissolved organic matter inupland catchments in the United Kingdom . PhD thesis, The Universityof Newcastle (unpublished).
Burdige DJ, Kline SW, Chen W. 2004. Fluorescent dissolved organicmatter in marine sediment pore waters. Marine Chemistry 89:289–311. DOI: 10Ð1016/j.marchem.2004Ð02Ð015.
Chen J, LeBoeuf EJ, Dai S, Gu B. 2003. Fluorescence spectroscopicstudies of natural organic matter fractions. Chemosphere 50: 639–647.
Clark CD, Jimenez-Morais J, Jones II G, Zanardi-Lamardo E, MooreCA, Zika RG. 2002. A time resolved fluorescence study of dissolvedorganic matter in a riverine to marine transition zone. Marine Chemistry78: 121–135.
Clay A, Bradley C, Gerrard AJ, Leng MJ. 2004. Using stable isotopesof water to infer wetland hydrological dynamics. Hydrology and EarthSystems Science 8(6): 1164–1173.
Dawson JJC, Bakewell C, Billet MF. 2001. Is in-stream processing animportant control on spatial changes in carbon fluxes in headwatercatchments? The Science of the Total Environment 265: 153–167.
Eatherall A, Naden PS, Cooper DM. 1998. Stimulating carbon flux to theestuary: the first step. The Science of the Total Environment 210–211:519–533.
Goslan EH, Voros S, Banks J, Wilson D, Hillis P, Campbell AT,Parsons SA. 2004. A model for predicting dissolved organic carbon
distribution in a reservoir water using fluorescence spectroscopy. WaterResearch 38: 783–791. DOI: 10Ð1016/j.watres.2003Ð10Ð027.
Hongve D. 1999. Production of dissolved organic carbon in forestedcatchments. Journal of Hydrology 224: 91–99. PII: S0022-1694(99)00132-8.
Hope D, Billet MF, Cresser MS. 1994. A review of the export of carbonin river water fluxes and processes. Environmental Pollution 84:301–324.
Johnes PJ, Heathwaite AL. 1992. A procedure for the simultaneousdetermination of nitrogen and total phosphorus in freshwater samplesusing persulphate microwave digestion. Water Research 26(10):1281–1287.
Klabitz K, Geyer S, Geyer W. 2000. A comparative characterization ofdissolved organic matter by means of original aqueous samples andisolated humic substances. Chemosphere 40: 1305–1312. PII: S0045-6535(99)00238-6.
Klapper L, McKnight DM, Fulton JR, Blunt-Harris EL, Nevin KP, Lov-ley DR, Hatcher PG. 2002. Fulvic acid oxidation state detection usingfluorescence spectroscopy. Environmental Science and Technology 36:3170–3175. DOI: 10Ð1021/es0109702.
Malcolm RL. 1985. Geochemistry of stream fulvic and humic substances.In Humic Substances in Soil, Sediment and Water: Geochemistry, Isola-tion, and Characterization , Aiken GR, McKnight DM, Wershaw RL,MacCarthy P (eds). Wiley: New York; 181–209.
Mobed JJ, Hemmingsen SL, Autry JL, McGown LB. 1996. Fluorescencecharacterisation of IHSS humic substances: total luminescence spectrawith absorbance correction. Environmental Science and Technology 30:3061–3065. DOI: S0013-936X(96)00132-0.
Patel-Sorrentino N, Mounier S, Benaim JY. 2002. Excitation–emissionfluorescence matrix to study pH influence on organic matterfluorescence in the Amazon basin rivers. Water Research 36:2571–2581. PII: S0043-1354(01)00469-9.
Peuravuori J, Monteiro A, Eglite L, Pihlaja K. 2005. Comparative studyfor separation of aquatic humic-type organic constituents by DAX-8,PVP and DEAE sorbing solid and tangential ultrafiltration; elementalcomposition, size-exclusion chromatography, UV–vis and FT-IR.Talanta 65: 408–422. DOI: 10Ð1016/j.talanta.4004Ð06Ð042.
Senesi N, Miano TM, Provenzano MR, Brunetti G. 1989. Spectroscopicand compositional comparative characterization of IHSS references andstandard fulvic and humic acids of various origin. The Science of theTotal Environment 81–82: 143–156.
Sierra MMD, Giovanela M, Parlanti E, Soeiano-Sierra EJ. 2005. Fluo-rescence fingerprint of fulvic and humic acids from varied originsas viewed by single scan and excitation/emission matrix techniques.Chemosphere 58: 715–733. DOI: 10Ð1016/j.chemosphere.2004Ð09Ð038.
Vannote RL, Minshall GW, Cummings KW, Sedell JR, Cushing CE.1980. The river continuum concept. Canadian Journal of Fisheriesand Aquatic Sciences 37: 130–137.
Worrall F, Burt T, Adamson J. 2004. Can climate change explainincreases in DOC flux from upland peat catchments? The Science of theTotal Environment 326: 95–112. DOI:10Ð1016/j.scitotenv.2003Ð11Ð022.
Copyright 2007 John Wiley & Sons, Ltd. Hydrol. Process. 21, 2093–2099 (2007)DOI: 10.1002/hyp