Environmental Toxicology and Chemistry, Vol. 23, No. 12, pp.

2889–2899, 2004
q 2004 SETAC
Printed in the USA
0730-7268/04 $12.00 1 .00

INFLUENCE OF NATURAL ORGANIC MATTER SOURCE ON ACUTE COPPER, LEAD,

AND CADMIUM TOXICITY TO RAINBOW TROUT (ONCORHYNCHUS MYKISS)

MELISSA L. SCHWARTZ,*† P. JEFFERSON CURTIS,‡ and RICHARD C. PLAYLE§

†CANMET—Mining and Mineral Sciences Laboratories, Natural Resources Canada, Ottawa, Ontario K1A 0G1
‡Earth and Environmental Sciences, Okanagan University College, Kelowna, British Columbia V1V 1V7, Canada
§Department of Biology, Wilfrid Laurier University, Waterloo, Ontario N2L 3C5, Canada

( Received 17 October 2003; Accepted 1 May 2004)

Abstract—Natural organic matter (NOM) was concentrated from various sites across Canada using a portable reverse-osmosis unit
to obtain a range of NOM types, from mainly allochthonous (terrestrially derived) to mainly autochthonous (aquatically derived)
NOM. The addition of NOM to Cu exposures in ion-poor water always decreased Cu toxicity to rainbow trout (Oncorhynchus
mykiss, ;1 g) over a 96-h period, and the degree of protection varied with respect to NOM source. A good correlation was found
between the specific absorbance coefficient (SAC) and time to reach 50% mortality (LT50; p , 0.001), indicating that more optically
dark, allochthonous-like NOM decreases Cu toxicity better than does optically light, more autochthonous-like NOM. A similar,
good relationship between NOM source and Pb toxicity was seen ( p , 0.001), once confounding effects of Ca binding to NOM
were accounted for. No significant relationship between Cd toxicity and NOM optical quality was seen ( p 5 0.082), and in toxicity
tests with Cd the presence of some of the NOM sources increased Cd toxicity compared to Cd-only controls. Specific absorbance
coefficients were used as a proxy measurement of NOM aromaticity in our study, and fluorescence indices were run on some NOM
samples to obtain percent aromaticity for each sample. A good correlation was found between SAC and percent aromaticity,
indicating that the simple SAC measurement is a reasonable indication of NOM aromaticity and of metal binding by NOM.

Keywords—Natural organic matter Copper Lead Cadmium Fish

INTRODUCTION
It has been well established that variations in water chem-
istry can alter metal toxicity by affecting the amount of metal
available to bind to fish gills [1–4]. Characterizing the binding
affinities of various natural ligands in surface waters, as well
as understanding how the freshwater fish gill interacts with
cations present in the water, furthers the understanding of how
metals exert their toxic effects and how these effects are related
to water chemistry. In 1983 the basis for a gill-metal interaction
model was published by both Pagenkopf [5] and Morel [6]
and since then has been expanded and is now termed the biotic
ligand model (BLM) [7–11].
The unique aspect of metal-gill modeling and the BLM in
general is that the approach considers the external gill surface
as a surface with receptors for metal binding. Essentially, the
fish gill can be thought of as a negatively charged ligand that
binds metals because of its tendency to carry a net negative
charge at environmental pHs, and the gill has the potential to
strip metals from weaker ligands. The metal-gill modeling
framework has been developed for Cu, Ag, Cd, Co, and Pb
[4,12–15]. The gills of freshwater fish are the primary target
for Cu, Pb, and Cd toxicity, where an inhibition of ion transport
mechanisms takes place, impairing the fish’s ability to maintain
proper ionic balance and eventually causing death [16–18].
Although substantial evidence exists that organic matter pres-
ent in natural waters decreases metal toxicity and accumulation
[19–21], little work exists that examines the relative protective
effects of different organic matters on the effects of metals.
How to properly deal with the complexity of natural organic
matter (NOM) in these models remains a challenge. The chem-
* To whom correspondence may be addressed
(mschwart@nrcan.gc.ca).
2889
ical nature of NOM is poorly defined but can be broadly clas-
sified as humic and fulvic material. Humic acid is defined as
the fraction of NOM that precipitates at low pH, is yellow to
black, is refractory, and is high in molecular weight, while
fulvic acid is the portion of NOM that is soluble under all pH
conditions and is lower in molecular weight [22,23]. Given
the complicated nature of NOM, assigning a standard structure
and functionality is impossible. Attempts at elemental analysis,
molecular weight measurements, acid-base titration, and spec-
troscopic analysis tend to reveal only average properties and
are often ambiguous since these methods are meant to be ap-
plied to pure substances [24]. Generally, it can be assumed
that NOM contains two discrete types of metal binding sites,
mainly carboxylate and phenolic hydroxyl groups [24]. Sul-
fydryl and amino groups within organic matter also provide
strong binding sites for certain metals [25–27]. One method
to deal with the differences in NOM properties and the sub-
sequent differences in binding affinities for cations is to use
simple measurements, such as optical characteristics, to assess
the metal binding ability of NOM and to use these measures
to weight the concentration of NOM for use in modeling (F
in Fig. 1).
Richards et al. [28] showed that the source of organic matter
affects the accumulation of Cu, Pb, and to some extent Hg on
the gills of rainbow trout and that these differences were cor-
related with optical properties of the NOM. One simple optical
measurement is the specific absorbance coefficient (SAC),
where the SAC is the absorbance at a given wavelength (usu-
ally 300–350 nM) normalized by the organic carbon concen-
tration [28]. Optically darker, more allochthonous NOM has
a higher SAC, and optically lighter, more autochthonous-like
NOM has a lower SAC, which can therefore indicate the gen-
2890 Environ. Toxicol. Chem. 23, 2004
Fig. 1. Metal-gill binding models for Cu, Pb, and Cd. This approach
to understanding toxic metal interactions with fish considers the gill
as a negatively charged ligand that complexes metal cations, similar
to naturally occurring ligands like natural organic matter (NOM),
carbonates, and chloride. Other cations, like Ca or Na, are also in-
corporated into the models since they compete with metals for binding
sites on NOM or the gill. Conditional equilibrium stability constants
(log K ) represent binding strength between a cation and a ligand.
Copper binds relatively strongly to NOM, Pb binds with intermediate
strength, and Cd has the weakest affinity for NOM. The F represents
NOM metal binding quality factors. Models are from Playle et al.
[12,13] and MacDonald et al. [4].
eral source of the NOM and its aromaticity [29]. The objectives
of the present study were to investigate the effect of NOM
source on the bioavailability to fish of Cu, Pb, and Cd and to
determine whether the effect of NOM source, if any, can be
modeled by including a NOM metal binding quality factor, F,
into existing metal-gill interaction models.
MATERIALS AND METHODS
Experimental animals
Juvenile rainbow trout (Oncorhynchus mykiss, ;1 g) were
purchased from Rainbow Springs Hatchery (Thamesford, ON,
Canada). Fish were initially held in 80-L tanks supplied with
a 1:1 mixture of well water and ion-poor water produced by
reverse osmosis (Culligan MP1000, series E reverse-osmosis
system, Culligan, Kitchener, ON, Canada). After 2 d, fish were
acclimated to reverse-osmosis water to which a low concen-
tration of Ca was added for fish survival (as CaCl2·2H2O;
Sigma-Aldrich, St. Louis, MO, USA). Water chemistry of the
ion-poor acclimation water was approximately 350 mM Na,
60 mM Ca, pH 7.5, 11 to 138C. During acclimation, fish were
fed every other day with ground three-point Hi-Pro commer-
cial fish food (Corey Feed Mills, Fredericton, NB, Canada).
Feeding was stopped 3 d before each experiment, and fish
were not fed during an experiment to eliminate metal binding
to food or feces during metal exposures.
M.L. Schwartz et al.
Experimental protocols
Ten trout were assigned to randomly arranged polyethylene
buckets containing 10 L of reverse-osmosis water to which 30
mM Ca (for Cu and Cd experiments) or 100 mM Ca (for Pb
experiments) were added to ensure survival of control fish.
Exposure buckets (seven to nine) were placed in a water bath
containing flowing well water to keep water temperatures sta-
ble. Before fish were added, each exposure was brought to pH
6.5 to 7.0 with diluted HNO3 (Trace Metal Grade, Fisher Sci-
entific, Unionville, ON, Canada) or diluted KOH (Fluka-Gar-
antie, St. Gallen, Switzerland) as needed. During the 96-h
exposure, no water replacements were made; however, KOH
or HNO3 was added as needed to keep water pH between 6.5
and 7.0, and 1 mM (nominal) Cu or Cd or 2 mM Pb were
added, Cu as CuSO4 (AnalaR, BDH Chemicals, Toronto, ON,
Canada), Pb as PbCl2 (Mallinckrodt Analytical Reagent, Mal-
linckrodt Specialty Chemicals, Paris, KY, USA), and Cd as
CdCl2 (Mallinckrodt Analytical Reagent). A nominal concen-
tration of 5.0 mg C/L was added in the NOM treatments with
Cu and Cd and 2.5 mg C/L for Pb exposures; NOM was limited
so exposures were not replicated. In addition to NOM collected
by us from natural sources (see the following discussion),
Suwannee River NOM (N101; reverse osmosis isolation; In-
ternational Humic Substances Society [St. Paul, MN, USA])
and Aldrich humic acid (Sigma) were also used. Containers
were checked for fish mortality every 6 to 12 h.
Water samples were taken at 0 h and every 12 h thereafter
for chemical analysis. Samples for Ca, Na, and Cl analysis
were collected in 7-ml scintillation vials, while samples for
metal analysis were collected and acidified with two drops of
16 N HNO3. Samples for total organic carbon (TOC) mea-
surement were collected in 6-ml previously combusted boro-
silicate vials (4 h at 4508C; Canadawide Scientific, Ottawa,
ON, Canada) that were filled without leaving any headspace,
sealed with Teflont-lined caps, and stored in the dark at 48C
until analysis. Water pH was measured in each bucket at these
sampling times with a Radiometer PHM82 pH meter and
GK2401C combination electrode (London Scientific, London,
ON, Canada).
Analytical techniques
Gill and water metal concentrations were analyzed on a
graphite furnace atomic absorption spectrophotometer (Varian
SpectrAA-880 with a GTA 100 atomizer [Mulgrave, Victoria,
Australia]) with 10-ml injections for Cd and Pb and 20-ml
injections for Cu. Operating conditions were as specified by
Varian. One percent ammonium phosphate (monobasic) was
added as a chemical modifier for Pb and Cd analysis to allow
the use of a higher ashing temperature. Aqueous Na and Ca
concentrations were measured on a flame atomic absorption
spectrophotometer (Varian SpectAA-880, Varian Australia),
and Cl was measured using a colorimetric assay (mercuric
thiocyanate method; Sigma-Aldrich).
Total carbon and inorganic carbon concentrations were
measured with a total carbon analyzer (Shimadzu TOC-5050A,
Mandel Scientific, Guelph, ON, Canada). Total organic carbon
concentration of each sample was calculated automatically by
subtracting inorganic carbon from total carbon. Standard so-
lutions were prepared according to Shimadzu specifications.
For each analysis, 65 ml of sample was injected two or three
times giving a coefficient of variation of ,5%. Absorbance of
each NOM sample was read in polystyrene cuvettes (1 cm path
length; Fisher) at 340 nm on a Milton Roy 301 spectropho-
Effect of NOM source on acute metal toxicity to trout
tometer (Milton Roy, Rochester, NY, USA) zeroed against ul-
trapure water, and optical density per mg C was calculated as
a specific absorbance coefficient using the equation SAC 5
2,303 3 absorbance340nm 4 TOC [4,28]. Natural organic matter
fluorescence was measured on a Varian Cary Eclipse fluores-
cence spectrophotometer (Varian, Mulgrave, Australia).
Collection of natural organic matter
To investigate the influence of variability in NOM on metal
bioavailability and subsequent toxicity to fish, NOM was col-
lected from various sources across Canada. The objective was
to obtain NOM that ranged in metal binding quality, depending
on its source and exposure to processes that are known to
degrade OM in natural waters (e.g., ultraviolet light and bac-
teria [30]). Approximately 200 L of water were pumped
through silicon tubing by a Masterflex peristaltic pump (Cole
Parmer Instrument, Vernon Hills, IL, USA) and passed through
1-mm glass-fiber filters (144-mm-diameter filters; Geotech En-
vironmental Equipment, Denver, CO, USA) before being con-
centrated using a stainless-steel portable reverse-osmosis unit
(Limnological Research, Kelowna, BC, Canada) with a mo-
lecular weight cutoff of about 400 Da (FilmTec FT30 U.S.
Filters thin composite RO membrane, Minneapolis, MN,
USA). Once concentrated to approximately 5 L, the NOM was
treated in a cation exchange resin (Amberlite IR-118H; Sigma),
brought to a pH of around 2.0 with 4 N HCl to remove metal
contaminants, and stored at 48C. The acidic NOM concentrate
was neutralized with NaOH before an exposure.
The first set of NOM was concentrated from Kalamalka
Lake (508109N, 119829W), Wood Lake (50859N, 1198239W),
Duck Lake (508 00 9 N, 119 8 24 9 W), and Mission Creek
(498519N, 1198299W) near Kelowna, British Columbia, in the
fall of 1999. Natural organic matter from Duck Lake flows
into Wood Lake and eventually into Kalamalka Lake, therefore
providing a source of NOM ranging from allochthonous-like
to autochthonous-like. Mission Creek is not connected to these
lakes but is nearby and drains similar terrain.
To include completely algal-derived NOM in this set, NOM
was generated within a mesocosm. The 1,000-L Plexiglas me-
socosm was outdoors under natural lighting conditions. De-
ionized water had salts added to approximate global average
river water [31], had nitrogen and phosphorus added to attain
an N:P ratio of approximately 15:1 (1,500 mg/L N, 100 mg/
L P), and then was inoculated with algae from Okanagan Lake
(BC, Canada [498509N, 1198309W]). The mesocosm was in-
cubated for four months, then the autochthonous OM was con-
centrated using the portable reverse-osmosis unit.
The second set of NOM was collected from the Experi-
mental Lakes Area (ELA) in July 2000. The ELA is located
in northwestern Ontario, 60 km east of Kenora between
498379N to 498479N and 938339W to 938509W on the south-
western margin of the Canadian shield. The ELA lakes are
generally small and are surrounded by boreal forest. We col-
lected NOM from Lake 114, Lake 227, Lake 239, and Lake
305 as well as from the east inflow stream to Lake 239. The
third set of NOM used in this study was collected from sites
in southern Ontario and northwestern Québec during the sum-
mer and fall of 2000. In Ontario, we sampled Luther Marsh
near Guelph (Canada) (438579N, 808269W), nutrient-rich Sanc-
tuary Pond located in Point Pelee National Park (Canada)
(418599N, 828339W), and Lake Ontario (438459N, 788009W).
Sites in the Abitibi region of Québec were Lac Jouve (498339N,
788179W), Lac Mistaouac (498259150N, 788419W), and a
Environ. Toxicol. Chem. 23, 2004 2891
dammed wetland herein named Ducks Unlimited Pond
(498269N, 788209W).
Data presentation
Sigma Plot was used to graph data, and correlation coef-
ficients (695% confidence intervals) were calculated using the
Pearson product moment in SigmaStat Statistical Software,
Version 2.0 (SPSS, Chicago, IL, USA). All data in tables are
expressed as means 61 standard deviation (n 5 number of
samples measured).
RESULTS
96-h Cu exposures
Three sets of 1-mM Cu exposures were run. In the first set
of exposures, trout survival increased in the presence of an
average of 8.6 mg C/L NOM collected in British Columbia
compared to the Cu-only exposure. Water chemistry plus times
to 50% mortality (LT50) for each exposure are given in Table
1. Only 20% mortality was seen with the optically dark Mis-
sion Creek NOM at the end of 96 h, with an estimated LT50
of 180 h (Table 1). Optically dark, allochthonous-like Duck
Lake NOM decreased Cu toxicity more than Wood Lake and
Kalamalka Lake NOM; the autochthonous-like mesocosm
NOM decreased Cu toxicity better than expected based on its
color but had more Ca (Table 1). In the second set of exposures,
all treatments using NOM concentrated at the ELA (mean of
6.1 mg C/L; Table 1) decreased Cu toxicity compared to the
Cu-only exposure (LT50 5 12 h). Of the NOM treatments,
trout survival was greatest with the addition of very optically
dark Aldrich humic acid where only 10% mortality was seen
over the course of 96 h (estimated LT50 ;200 h). Optically
dark Suwannee River and Lake 239 east inflow NOM also
decreased Cu toxicity well, with LT50s of 96 and 43 h, re-
spectively (Table 1). Natural organic matter from Lakes 305,
227, 239, and 114 were similar in decreasing Cu toxicity with
LT50s of 24, 27, 29, and 31 h, respectively.
A third set of Cu toxicity tests was run with NOM collected
from southern Ontario and northwestern Québec. As with the
preceding toxicity tests, the Cu-only exposure was very toxic
to trout (LT50 5 11 h), whereas with NOM added this toxicity
was decreased (mean of 6.2 mg C/L; Table 1). Optically dark
Luther Marsh NOM decreased Cu toxicity completely over 96
h (estimated LT50 ;200 h), and Suwannee River NOM de-
creased Cu toxicity the second most, resulting in 40% mortality
over 96 h, with an estimated LT50 of 110 h. Both Lac Jouve
and Sanctuary Pond NOM decreased Cu toxicity, with an LT50
of 96 h each. The Ducks Unlimited NOM provided interme-
diate protection (LT50 5 43 h), and both Lac Mistaouac and
Lake Ontario NOM decreased Cu toxicity the least well, with
an LT50 of 24 h each.
96-h Pb exposures
Based on initial toxicity tests with Pb and NOM done in
ion-poor water in our lab, the 96-h toxicity tests with NOM
were run with extra Ca added to each exposure. Calcium ad-
ditions were to prevent increased Pb toxicity in the presence
of NOM, compared to exposures to Pb alone (data not shown),
presumably because the small amount of Ca in solution was
also complexed by NOM, leading to ion losses by the fish (see
Discussion section). In the experiment using NOM from the
ELA, most of the 2 mM Pb plus NOM exposures were less
toxic than the Pb-only exposure (LT50 5 66 h); however, in
the presence of Lake 239 NOM and especially the optically
 2892

Table 1. Chemistry of exposure water from 96-h Cu toxicity tests using juvenile rainbow trout in the presence or absence of 5 mg C/L, nominally, natural organic matter (NOM) concentrated in the
field using a portable reverse-osmosis unit. Water temperature throughout the experiments was held between 9 and 118C. If time to 50% fish mortality (LT50) was not reached in 96 h, an estimate of
LT50 was made (in parentheses). SAC is the specific absorbance coefficient. Cu-gillCu is the calculated Cu-gill burden without considering F, the NOM metal binding quality factor. Cu-gillCu(F) is
the calculated Cu-gill burden considering this factor. TOC is total organic carbon. Water chemistry measurements are expressed as a mean 61 standard deviation (n). See text for details; Aldrich HA
(Sigma, St. Louis, MO, USA); HA 5 humic acid; all study locations except the Suwannee River (GA, USA) are found in Canada
Environ. Toxicol. Chem. 23, 2004

Cu-
NOM exposure Ca (mM) Na (mM) Cl (mM) TOC (mg C/L) pH Cu (mM) SAC LT50 (h) Cu-gillCu F gillCu(F )

Control 38 6 1 (9) 317 6 8 (9) 46 6 8 (9) 1.3 6 0.4 (9) 6.8 6 0.3 (9) 0.04 6 0.03 (9) — — — — —
Cu only 44 6 7 (3) 276 6 6 (3) 86 6 39 (3) 1.0 6 0.5 (3) 6.3 6 0.1 (3) 0.89 6 0.09 (3) — 15 — — —
Mesocosm 483 6 36 (9) 2,654 6 45 (9) 1,690 6 76 (9) 8.9 6 0.8 (9) 7.0 6 0.4 (9) 1.02 6 0.03 (9) 4.8 6 3.7 (9) 72 0.68 0.49 1.43
Kalamalka Lake 118 6 2 (7) 2,543 6 49 (7) 609 6 10 (7) 8.2 6 0.6 (7) 6.7 6 0.2 (7) 1.19 6 0.08 (7) 5.5 6 3.9 (7) 30 0.63 0.53 1.71
Wood Lake 80 6 1 (6) 2,202 6 22 (6) 728 6 44 (6) 8.8 6 0.3 (5) 7.4 6 0.3 (6) 1.13 6 0.04 (6) 6.7 6 5.7 (5) 24 0.40 0.59 1.27
Duck Lake 105 6 2 (9) 1,225 6 21 (9) 421 6 31 (9) 8.5 6 0.5 (9) 7.0 6 0.2 (9) 1.01 6 0.03 (9) 13.1 6 5.5 (9) 56 0.40 0.80 0.63
Mission Creek 204 6 2 (9) 1,728 6 21 (9) 600 6 27 (9) 8.3 6 0.4 (9) 7.1 6 0.3 (9) 0.97 6 0.08 (9) 20.7 6 1.7 (8) .96 (180) 0.53 0.94 0.59
Control 29 6 2 (9) 306 6 21 (9) 120 6 17 (9) 1.3 6 0.5 (9) 7.0 6 0.1 (9) 0.02 6 0.01 (9) — — — — —
Cu only 30 6 1 (3) 311 6 22 (3) 165 6 15 (3) 1.4 6 0.7 (3) 6.6 6 0.5 (3) 0.90 6 0.06 (3) — 12 — — —
Lake 227 48 6 2 (6) 394 6 12 (6) 236 6 24 (5) 6.0 6 0.4 (6) 6.9 6 0.2 (6) 0.97 6 0.02 (6) 17.0 6 1.8 (6) 27 0.55 0.89 0.86
Lake 305 46 6 2 (4) 387 6 8 (4) 185 6 46 (4) 6.1 6 0.5 (4) 7.0 6 0.3 (4) 1.00 6 0.02 (4) 12.0 6 1.4 (4) 24 0.56 0.77 1.30
Lake 239 50 6 3 (5) 1,680 6 34 (5) 1,154 6 61 (5) 6.4 6 0.7 (5) 6.9 6 0.1 (5) 0.94 6 0.03 (5) 15.9 6 2.2 (5) 28 0.42 0.86 0.67
Lake 239 EIF 45 6 5 (9) 384 6 8 (9) 238 6 25 (9) 6.5 6 0.4 (9) 6.9 6 0.2 (9) 1.02 6 0.14 (9) 27.9 6 2.0 (9) 42 0.49 1.03 0.44
Lake 114 49 6 1 (7) 393 6 10 (7) 232 6 62 (7) 6.2 6 0.3 (7) 6.9 6 0.2 (7) 0.99 6 0.02 (7) 12.3 6 1.8 (7) 30 0.53 0.78 1.23
Suwannee River 49 6 2 (9) 380 6 11 (9) 278 6 17 (8) 6.1 6 0.5 (9) 6.9 6 0.2 (9) 0.98 6 0.02 (9) 26.4 6 2.0 (9) 96 0.54 1.02 0.52
Aldrich HA 44 6 4 (9) 388 6 18 (9) 313 6 19 (9) 5.7 6 0.5 (9) 6.9 6 0.2 (9) 0.96 6 0.02 (9) 54.1 6 5.5 (9) .96 (200) 0.60 1.24 0.32
Control 5 6 2 (9) 298 6 28 (9) 105 6 16 (9) 2.0 6 0.7 (9) 6.9 6 0.2 (9) 0.03 6 0.01 (9) — — — — —
Cu only 7 6 2 (3) 364 6 159 (3) 102 6 17 (3) 1.9 6 1.2 (3) 6.8 6 0.0 (3) 0.87 6 0.06 (3) — 11 — — —
Lac Mistaouac 10 6 2 (7) 2,058 6 124 (7) 1,443 6 24 (7) 6.5 6 0.8 (7) 6.9 6 0.1 (7) 0.89 6 0.04 (7) 11.8 6 1.7 (7) 24 0.15 0.76 0.45
Luther Marsh 13 6 1 (8) 340 6 15 (8) 241 6 40 (8) 6.4 6 0.6 (9) 6.9 6 0.2 (9) 0.90 6 0.11 (9) 29.1 6 3.7 (9) .96 (200) 0.18 1.04 0.16
Sanctuary Pond 14 6 2 (9) 333 6 47 (9) 127 6 23 (9) 6.3 6 0.5 (9) 6.9 6 0.1 (9) 0.87 6 0.01 (7) 9.2 6 6.2 (9) 96 0.18 0.69 0.92
Ducks Unlimited Pond 13 6 2 (9) 334 6 26 (9) 129 6 40 (9) 6.5 6 0.8 (8) 6.8 6 0.2 (9) 0.97 6 0.03 (9) 17.6 6 2.0 (8) 42 0.22 0.89 0.34
Lac Jouve 12 6 1 (8) 327 6 32 (8) 199 6 30 (9) 6.2 6 0.5 (8) 6.9 6 0.1 (9) 1.25 6 0.02 (9) 21.9 6 9.2 (8) 96 1.04 0.96 1.27
Lake Ontario 18 6 3 (9) 2,981 6 874 (9) 1,135 6 27 (9) 5.8 6 0.5 (9) 6.9 6 0.2 (9) 0.89 6 0.03 (9) 5.2 6 7.2 (9) 24 0.28 0.51 1.72
Suwannee River 12 6 1 (9) 376 6 19 (9) 267 6 26 (9) 6.4 6 0.7 (9) 6.8 6 0.2 (9) 0.93 6 0.02 (9) 18.6 6 2.1 (9) .96 (110) 0.19 0.91 0.27
M.L. Schwartz et al.
Effect of NOM source on acute metal toxicity to trout Environ. Toxicol. Chem. 23, 2004 2893

Table 2. Chemistry of exposure water from 96-h Pb toxicity tests using juvenile rainbow trout in the presence or absence of 2.5 mg C/L, nominally, natural organic matter (NOM). Water temperature
throughout the experiments was 98C. If time to reach 50% fish mortality (LT50) was not reached in 96 h, an estimate of LT50 was made (in parentheses). SAC is the specific absorbance coefficient.
Pb-gillPb is the calculated Pb-gill burden without considering F, the NOM metal binding quality factor. Pb-gillPb(F ) is the calculated Pb-gill burden considering this factor. TOC is total organic
carbon. Water chemistry measurements are expressed as a mean 61 standard deviation (n). See text for details, Aldrich HA (Sigma, St. Louis, MO, USA); HA 5 humic acid; all study locations

gillPb(F)
light, autochthonous-like Lake 114 NOM, the Pb toxicity was

0.70
1.01
0.58
0.74
1.42
0.62
0.61

0.58
0.48
0.94
0.63
0.67
0.86
0.76
still the same or greater than for Pb alone (LT50 5 63 and 48

Pb-

—
—

—
—
h, respectively; Table 2). Water chemistry from this experiment
is given in Table 2. Exposures to Pb were not run using NOM
from BC sources because we had exhausted those concentrates.

0.86
0.76
0.83
1.02
0.62
0.99
1.24

0.92
1.11
0.75
0.96
0.93
0.58
0.94
—
—

—
—
F
Even with additional Ca added to the exposures, greater Pb
toxicity was usually seen when NOM was included in the

Pb-gillPb
exposure waters using NOM collected from Ontario and Qué-

0.46
0.55
0.38
0.78
0.73
0.61
1.02

0.51
0.64
0.48
0.56
0.56
0.49
0.66
—
—

—
—
bec (Table 2). All NOM treatments resulted in lower LT50s
(58–75 h) than the Pb-only exposure (LT50 5 84 h), except
when Luther Marsh NOM was used (LT50 5 96 h). However,

.96 (105)
the general pattern of LT50s from both sets of exposures agrees

LT50 (h)
well with the SAC of each NOM. Specifically, the Pb plus

66
69
72
63
66
48
66

84
60
96
75
69
69
58
67
—

—
optically lightest NOM (Lake Ontario and Lake 114) had the
lowest LT50s (58 and 48 h, respectively), and optically dark
Suwannee River, Luther Marsh, and Aldrich humic acid de-

(6)
(7)
(5)
(8)

(8)
(8)

(9)
(9)
(9)
(9)
(9)

(8)
(6)

(7)
creased Pb toxicity the most.

6 4.2
6 4.1
6 4.8
6 3.3

6 1.2
6 7.2

6 9.9
6 8.9
6 8.7
6 5.1
6 4.1

6 2.8
6 1.9

6 6.1
SAC

—
—

—
—
96-h Cd exposures

16.3
11.5
14.7
26.4

24.3
54.1

19.5
36.1
11.4
21.9
20.1

20.6
7.5

6.4
The 96-h Cd toxicity tests with NOM collected from the
ELA showed that some NOMs decreased Cd toxicity while
others did not. Optically light Lake 305 and Lake 114 NOM

(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(7)
(8)
except the Suwannee River (GA, USA) are found in Canada
did not decrease the toxicity of 1 mM Cd compared to the Cd-

Pb (mM)

0.01
0.42
0.19
0.12
0.17
0.17
0.21
0.21
0.11
0.03
0.28
0.23
0.14
0.11
0.13
0.04
0.16
0.14
only exposure (LT50 5 48 h for each; Table 3). After 50%

6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
mortality was reached, fish in the Lake 305 and Lake 114 OM

0.01
1.85
2.14
2.19
2.26
2.36
2.22
2.06
2.24
0.00
1.74
2.00
1.89
1.79
1.80
1.84
2.26
1.89
exposures with Cd died more quickly than in the Cd-only
exposure, indicating that the presence of these NOMs actually
worsened Cd toxicity (data not shown). Optically darker Lake

(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(7)
(8)
239 and Suwannee River NOM decreased Cd toxicity slightly,

0.2
0.2
0.2
0.2
0.3
0.1
0.1
0.2
0.2
0.1
0.1
0.2
0.1
0.2
0.1
0.1
0.2
0.2
with both resulting in an LT50 of 57 h. Lake 227 NOM and pH

6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
Aldrich humic acid both decreased Cd toxicity to a greater 6.9
6.9
7.0
6.9
7.0
7.0
6.9
6.9
6.9
6.8
6.8
6.8
6.8
6.8
6.8
6.8
6.8
6.8
degree (LT50s 5 66 h), and Lake 239 east inflow NOM pro-
vided the greatest decrease in Cd toxicity (LT50 5 72 h). Water
TOC (mg C/L)

chemistry of these exposures is given in Table 3.

(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(7)
(8)
The LT50s from the Cd exposures with NOM from southern
1.0
1.0
1.1
0.8
0.9
0.7
0.4
0.6
0.6
0.5
0.4
0.7
0.3
0.5
0.3
0.4
0.2
0.3
Ontario and Québec revealed that in some cases NOM de-
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
creased Cd toxicity, as was seen in the previous experiment,
1.6
1.9
4.4
4.2
4.2
4.0
3.8
3.8
3.4
1.1
1.1
3.0
3.4
3.6
3.4
3.5
3.3
3.4
but the presence of some NOMs increased Cd toxicity (Table
3). Natural organic matter concentrated from the Ducks Un-
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)

(9)
(9)
(9)
(9)
(7)
(8)
limited Pond, Lac Mistaouac, and Lake Ontario did not de-
crease Cd toxicity or increased its toxicity (LT50 5 42, 50,
Cl (mM)

19
30
23
26
30
18
32
22
20
20
71
99

14
25
46
37
13
28
6
6
6
6
6
6
6
6
6
6
6
6

6
6
6
6
6
6
54, and 59 h, respectively). The presence of Sanctuary Pond
NOM in the exposure did not alter Cd toxicity (LT50 5 59
1,154
379
411
364
345
930
356
391
400
394
390
386

334
268
268
297
895
324
h). Optically dark Suwannee River and Lac Jouve NOM both
decreased Cd toxicity equally (LT50s 5 66 h), while dark
120 (9)

Luther Marsh NOM decreased Cd toxicity most (LT50 5 78

54 (9)

36 (7)
33 (9)
18 (9)
33 (9)
20 (9)
66 (9)

20 (9)
16 (9)
37 (9)
29 (9)
16 (9)

16 (9)
43 (9)
10 (9)
26 (9)

14 (8)
Na (mM)

h).
6

6
6
6
6
6
6

6
6
6
6
6

6
6
6
6

6
6

Correlations of metal toxicity with NOM color as SAC

1,266

1,700
295
384
274
356
952

385
327
306
361
324

307
259
280
287

287
346

We had some indication that more colored, aromatic NOM

decreases acute metal toxicity to fish most [4,28]. Although
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(9)
(7)
(8)
Ca (mM)

more sophisticated Cu and H1 titrations of NOM were done,

4
3
4
3
2
2
4
3
2
6
5
8
3
6
4
4
8
3

we wished to see if a simple optical measurement, specific

6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6

absorbance coefficient (SAC), yields the same information as

87
92
88
86
98
86
89
87
82
88
88
96
83
83
85
84
90
84

more complex measurements of metal or H1 binding capacity

[32]. We plotted metal toxicity as LT50 against SAC for each
Ducks Unlimited Pond

suite of metal exposures. For Cu (Fig. 2A) and Pb (Fig. 2B),

the linear relationships were highly significant (Cu LT50 vs
Suwannee River

Suwannee River
Sanctuary Pond
NOM exposure

SAC, r 5 0.702, p , 0.001; Pb LT50 vs SAC, r 5 0.845, p

Lac Mistaouac
Lake 239 EIF

Luther Marsh

Lake Ontario

, 0.001). No significant relationship was observed between

Aldrich HA

Lac Jouve
Lake 227
Lake 305
Lake 239

Lake 114

Cd LT50 and SAC (r 5 0.481, p 5 0.082; Fig. 2C). Corre-

Pb only

Pb only
Control

Control

lations between 96-h LT50 and TOC concentrations were far

from significant (Cu: r 5 20.067, p 5 0.787; Pb: r 5 20.163,
Table 3. Chemistry of exposure water from 96-h Cd toxicity tests using juvenile rainbow trout in the presence or absence of 5 mg C/L, nominally, natural organic matter (NOM). Water temperature
2894

throughout the experiments was held between 9 and 108C. SAC is the specific absorbance coefficient. Cd-gillCd is the calculated Cd-gill burden without considering F, the NOM metal binding quality
factor. Cd-gillCd(F ) is the calculated Cd-gill burden considering this factor. TOC is total organic carbon. Water chemistry measurements are expressed as a mean 61 standard deviation (n). See text
for details; HA 5 humic acid; Aldrich HA (Sigma, St. Louis, MO, USA); all study locations except the Suwannee River (GA, USA) are found in Canada

Cd-
Exposure Ca (mM) Na (mM) Cl (mM) TOC (mg C/L) pH Cd (mM) SAC 96-h LT50 Cd-gillCd F gillCd(F )

Control 38 6 1 (9) 210 6 12 (9) 136 6 84 (9) 1.2 6 0.5 (8) 6.9 6 0.2 (9) 0.01 6 0.00 (9) — — — — —
Cd only 41 6 2 (8) 212 6 11 (8) 139 6 58 (8) 1.4 6 0.6 (8) 6.9 6 0.2 (8) 1.11 6 0.03 (8) — 48 — — —
Lake 227 51 6 4 (9) 270 6 11 (9) 173 6 45 (8) 6.0 6 0.4 (9) 7.0 6 0.2 (9) 1.12 6 0.05 (9) 13.8 6 1.1 (9) 66 1.92 0.81 1.93
Lake 305 46 6 1 (7) 286 6 16 (7) 99 6 9 (7) 6.2 6 0.5 (7) 6.8 6 0.2 (7) 1.10 6 0.03 (7) 9.7 6 0.5 (7) 48 1.91 0.70 1.94
Lake 239 52 6 5 (7) 1,515 6 36 (7) 1,018 6 37 (7) 6.0 6 0.3 (7) 6.8 6 0.4 (7) 1.07 6 0.07 (7) 14.9 6 1.4 (7) 57 1.90 0.84 1.92
Lake 239 EIF 47 6 2 (9) 244 6 10 (9) 189 6 26 (9) 6.1 6 0.3 (9) 6.7 6 0.5 (9) 1.14 6 0.06 (9) 25.8 6 1.4 (9) 72 1.91 1.01 1.91
Lake 114 50 6 2 (7) 252 6 9 (7) 182 6 64 (7) 6.2 6 0.5 (7) 6.8 6 0.2 (7) 1.10 6 0.02 (7) 9.5 6 1.2 (7) 48 1.91 0.7 1.93
Suwannee River 52 6 2 (7) 282 6 7 (7) 322 6 146 (7) 6.2 6 0.6 (7) 6.7 6 0.2 (7) 1.14 6 0.05 (7) 23.5 6 2.2 (7) 57 1.91 0.98 1.91
Aldrich HA 49 2 (9) 271 22 (9) 261 21 (9) 6.8 0.2 (9) 1.06 0.06 (9) 51.4 (9) 66 1.91 1.22 1.89
Environ. Toxicol. Chem. 23, 2004

6 6 6 5.6 6 0.4 (9) 6 6 6 5.4

Control 5 6 1 (8) 329 6 30 (8) 116 6 10 (8) 2.2 6 0.7 (9) 6.9 6 0.2 (9) 0.04 6 0.00 (9) — — — — —
Cd only 8 6 3 (8) 404 6 20 (8) 152 6 28 (8) 2.1 6 0.6 (9) 6.9 6 0.1 (9) 1.03 6 0.05 (9) — 59 — — —
Lac Mistaouac 10 6 2 (7) 2,288 6 91 (7) 1,559 6 35 (7) 6.6 6 0.7 (8) 6.8 6 0.2 (8) 1.00 6 0.07 (7) 12.8 6 2.6 (8) 50 1.90 0.79 1.94
Luther Marsh 12 6 1 (9) 396 6 25 (9) 290 6 131 (9) 6.9 6 0.6 (9) 6.8 6 0.3 (9) 0.96 6 0.03 (9) 24.4 6 4.8 (9) 78 1.89 0.99 1.90
Sanctuary Pond 14 6 3 (8) 396 6 29 (7) 154 6 20 (8) 6.5 6 0.7 (9) 6.8 6 0.1 (9) 0.97 6 0.04 (8) 13.8 6 4.0 (8) 59 1.90 0.81 1.93
Ducks Unlimited Pond 13 6 3 (7) 479 6 79 (7) 202 6 61 (7) 6.6 6 0.6 (8) 6.9 6 0.2 (8) 0.93 6 0.10 (8) 19.9 6 4.0 (8) 42 1.90 0.93 1.91
Lac Jouve 14 6 3 (9) 384 6 30 (9) 191 6 26 (9) 6.4 6 0.4 (8) 6.9 6 0.1 (9) 0.96 6 0.13 (9) 21.7 6 8.0 (8) 66 1.91 0.95 1.91
Lake Ontario 19 6 3 (8) 3,300 6 140 (8) 1,244 6 46 (8) 6.2 6 0.7 (8) 6.9 6 0.2 (9) 0.95 6 0.05 (9) 1.1 6 4.3 (8) 54 1.90 0.03 1.98
Suwannee River 13 6 2 (8) 426 6 14 (8) 294 6 43 (8) 6.5 6 0.8 (9) 6.9 6 0.2 (9) 0.98 6 0.04 (9) 20.3 6 2.4 (9) 66 1.91 0.93 1.92

for details.

Metal-gill modeling
p 5 0.010, n 5 13).
similar amounts of NOM for each exposure (Tables 1 to 3).

significant (Cu: r 5 0.563, p 5 0.015, n 5 18; Pb: r 5 0.682,

(;4 h per SAC unit), next for Pb (;1 h per SAC unit), and
M.L. Schwartz et al.

p 5 0.578; Cd: r 5 20.008, p 5 0.978) because we added

least and not significantly for Cd (;20 min per SAC unit).
absorbance coefficients (SAC). A good correlation of LT50 and SAC
Fig. 2. Correlations of Cu (A), Pb (B), and Cd (C) toxicity as time

the second Cd exposure were lower than in the first (Table 3).
higher (8.3–8.9 mg C/L; Table 1) than in the next two sets of
that as SAC increased, fish survival increased, most for Cu
Because Aldrich humic acid had a very high SAC and

seen with these two metals. Without the Aldrich data points,

Cu exposures (5.7–6.5 mg C/L), and the Ca concentrations in

NOMs used in this study, correlations with Cu and Pb 96-h
degree than lightly colored NOM of a more aliphatic nature. See text

However, water chemistries were not identical within each of

The general relationship seen in Figure 2A, B, and C was
the correlations of Cu and Pb LT50 against SAC were still
This was done to see whether the results using Aldrich humic
generally decreased metal toxicity much better than any of the
was seen with Cu and Pb but not with Cd, indicating that more ar-
to 50% mortality (LT50) with natural organic matter (NOM) specific

the Cu, Pb, or Cd exposures (see Tables 1 to 3). For example,

the TOC concentrations for the first set of Cu exposures were
LT50 and SAC were run with the Aldrich data points removed.
omatic, optically dark NOM decreases Cu and Pb toxicity to a greater

acid were responsible for the highly significant correlations

Effect of NOM source on acute metal toxicity to trout
Water chemistry differences could potentially confound the
LT50-versus-SAC relationships, so metal-gill modeling was
used to take into account differences in water chemistry lead-
ing to potential differences in cation competition at metal-gill
binding sites and metal complexation by NOM.
We modeled these interactions using MINEQL1 (Ver 4.5;
Environmental Research Software, Hallowell, ME, USA) us-
ing the log K values given in Figure 1 and Appendix 1. We
used the Cu-gill and Cd-gill models of Playle et al. [13] and
Playle [17] plus the Pb-gill model of Macdonald et al. [4]. The
Cu-gill and Cd-gill models were modified according to Mac-
donald et al. [4] by adding Ca-NOM and H-NOM interactions
and by increasing the number of Cu or Cd binding sites on
NOM to 0.2 mmol per mg C (see Appendix). For the Pb-gill
model, we used the suggested H-gillPb log K value of 5.0 and
used 0.5 mmol Pb binding sites per mg C [4]. For calculation
of correlation coefficients, we used the exposure duration if
the LT50 was greater than the exposure time [4].
For the first calculations, NOM concentration alone was
considered (e.g., ignoring differences in NOM source). Using
measured water chemistry from the exposures (mean values)
and first ignoring NOM source differences, we calculated metal
gill loads (e.g., Cu-gillCu, Pb-gillPb, Cd-gillCd; Tables 1 to
3). For all three metals, no significant correlations were ob-
served between LT50 and calculated gill metal concentrations
(Cu: r 5 0.116, p 5 0.637; Pb: r 5 0.419, p 5 0.130; Cd: r
5 0.033, p 5 0.912). These MINEQL1 simulations indicate
that differences in concentrations of ions and NOM cannot
alone explain the variability in LT50 values. That is, NOM
metal binding quality must be taken into account.
We used SAC as a simple method to weight the different
NOM types. To do this, we calculated the NOM weighting
factor, F, using the relationship F 5 0.31·ln(SAC340). Essen-
tially, this relationship converts an SAC value of 25 (e.g.,
optically dark, allochthonous Luther Marsh or Suwannee River
NOM) to 1.0 (high-quality NOM) and NOM with lower SAC
(e.g., Lake Ontario, Kalamalka Lake) to lower values (lower-
quality NOM), as indicated in Tables 1 to 3. The NOM con-
centration (e.g., total NOM metal binding sites, in mM) is
multiplied by the F value to yield a new NOM concentration,
which is put into the metal-gill model. For example, for the
mesocosm NOM with a mean SAC value of 4.8 (Table 1), an
F value of 0.49 is calculated, which decreases the NOM con-
centration from 1.78 mM (8.9 mg C/L TOC multiplied by 0.2
mmol/mg C) to 0.87 mM. With about half as many NOM sites
to bind Cu, the calculated load of Cu on the gills increases
from 0.68 nM (Cu-gillCu) to 1.43 nM (the Cu-gillCu[F] value;
Table 1).
Once NOM metal binding quality is considered using F,
the correlation between LT50 and calculated gill Cu burden
becomes nearly significant, with greater calculated gill Cu bur-
den correlating reasonably well with shorter fish survival (r
5 0.427, p 5 0.068). The correlation between LT50 and cal-
culated gill Pb burden also becomes nearly significant (r 5
0.529, p 5 0.052), as does the relationship for gill Cd (r 5
0.459, p 5 0.063). These modeling results, going from far
from significant correlations between LT50 and calculated met-
al-gill burdens when not considering NOM metal binding qual-
ity (e.g., p 5 0.130–0.912) to nearly significant correlations
when the NOM metal binding quality factor, F, is considered
(p 5 0.052–0.068), indicate that some index of NOM metal
binding quality is necessary when considering the biological
influence of NOM-metal interactions.
Environ. Toxicol. Chem. 23, 2004 2895
Other optical parameters of NOM
As other possible indices of NOM metal binding quality,
we considered the fluorescence index and percent aromaticity
calculated from this value. Fluorescence measurements were
made on the NOM concentrates from Québec and Ontario
using the method of McKnight et al. [29] to estimate the per-
cent aromaticity of each NOM to compare with SAC mea-
surements on the same samples. The ratio of emission intensity
at 450 and 500 nm yields the slope of the curve between 450
and 500 nm, which represents the decline in emission with
increasing wavelengths. A plot of fluorescence index against
percent aromaticity determined by 13C-NMR (nuclear magnetic
resonance) revealed a negative relationship (fig. 3 in McKnight
et al. [29]), meaning that percent aromaticity increases as the
fluorescence index decreases. Using the equation of the line
from their data ( y 5 3.94x20.316, where y is fluorescence index
and x is % aromaticity), the percent aromaticity for each of
our NOM samples was calculated (Table 4). Some of the fluo-
rescence indices we obtained were beyond the range of those
in McKnight et al. [29], so the percent aromaticities of those
samples (Luther Marsh and Lac Jouve) may not reflect the
true aromatic nature of these samples. Nonetheless, a good
correlation was observed between percent aromaticity and
SAC (r 5 0.864, p 5 0.012, n 5 7), which indicates that SAC
is a good proxy measurement for aromatic content of NOM.
We compared LT50 values for these seven sites with SAC
and with calculated percent aromaticity. The Cu LT50 data for
these seven sites correlated with SAC (r 5 0.792, p 5 0.034,
n 5 7) and were also strongly correlated with percent aro-
maticity (r 5 0.901, p 5 0.006, n 5 7). Similarly, the Pb LT50
data correlated with SAC (r 5 0.770, p 5 0.006, n 5 7) and
were also strongly correlated with percent aromaticity (r 5
0.899, p 5 0.006, n 5 7). Surprisingly, the Cd LT50 values
also correlated with percent aromaticity (r 5 0.779, p 5 0.039,
n 5 7), even though no correlation existed between the Cd
toxicity data and SAC (r 5 0.471, p 5 0.286, n 5 7). Fluo-
rescence measurements were not made on the other NOM
samples, but SAC–fluorescence relationships appear to be a
fruitful avenue for future research.
DISCUSSION
We exposed rainbow trout to Cu, Pb, and Cd in ion-poor
water with added natural organic matter from various sources
to provide data regarding biologically meaningful differences
in NOM metal binding quality. The general pattern seen was
that optically darker NOM decreases acute metal toxicity to
trout to a greater degree than does optically lighter NOM (Fig.
2A to C). A simple optical measurement such as the SAC,
which reflects NOM aromaticity (Table 4), may be a reasonable
method to incorporate NOM metal binding quality as well as
NOM quantity into biotic ligand models that predict metal
effects on aquatic organisms.
Previous data of ours indicated that optically darker Luther
Marsh NOM decreased the acute toxicity to trout of a mixed
metal solution to a greater degree than did optically lighter
Sanctuary Pond NOM, and we introduced a NOM metal bind-
ing quality factor, F [28]. A recent Pb-gill binding model yield-
ed a better correlation of Pb toxicity (as LT50 in h) with
calculated gill Pb burdens when this NOM quality factor was
used [4]. Our current Cu, Pb, and Cd toxicity data also cor-
related better with gill metal burdens calculated using the
NOM metal binding factor ( p 5 0.052 – 0.068 vs p 5 0.130–
0.912 without F ). These metal data from our laboratory agree
2896 Environ. Toxicol. Chem. 23, 2004 M.L. Schwartz et al.

Table 4. Optical characteristics of natural organic matter (NOM) samples collected in Ontario and Québec (Canada). Percent aromaticity was calculated using the fluorescence index of each sample
and the equation y 5 3.94x20.316 from McKnight et al. [29]. Specific absorbance coefficients (SAC) are the mean of measured values given in Tables 1 to 3. TOC 5 total organic carbon; LT50 5 time
with published NOM and polycyclic aromatic hydrocarbon
data where aromatic organic matter decreased PAH accumu-

Cd LT50
lation by Caenorhabditis elegans and Daphnia magna to a

(h)

54
50
59
42
66
66
78
greater degree than did less aromatic organic matter [33,34].
The NOM metal binding quality factor F is essentially a
multiplication factor for the number (or strength) of metal
binding sites per mg C. In general, molar absorbance at lower
wavelengths (200–400 nM) increases with molecular weight
and aromaticity, so SAC can be a reliable yet simple mea-
surement for aromatic content of NOM [29]. We used F 5
Pb LT50
(h)

0.31·ln(SAC) instead of F 5 0.21·ln(SAC) used before, which

58
60
75
69
67
69
96
was based on very limited data [28]. Using F 5 0.31·ln (SAC)
yields F 5 1.0 for an SAC of 25 for high-quality Luther Marsh
or Suwannee River NOM (Tables 1 to 3); using this formula
sets these optically dark, allochthonous NOMs to unity as
opposed to setting their F values to 0.68 using the former
formula. Low-quality, autochthonous-like NOM from lakes
.96 (110)

.96 (200)

have SAC values around 5 (Tables 1 to 3) which yield F 5

Cu LT50
(h)

0.5 (or 0.34 using the old formula). Using either formula, F
24
24
96
42

96

varies about twofold from high to low metal binding quality

endpoints for NOM (e.g., marsh vs lake NOM).
A twofold difference in NOM metal binding quality for
acute metal exposures is too large to ignore. Our previous data
showed 1.5- and 1.7-fold ranges in F using the old calculation
[4,28]. Bryan et al. [35] showed that a fulvic acid weighting
Aromaticity

factor for dissolved organic carbon improved Cu-NOM model

predictions for freshwater; their weighting factor had a range
21.7
24.7
27.0
31.1
32.6
40.9
61.8
(%)

of about 2.3-fold for the Windermere Humic Aqueous Model

to reach 50% mortality

VI (WHAM; Natural Environment Research Council, Swidon,

UK) considering Fe precipitation. More in-depth analysis of
our British Columbia NOM concentrates has shown good
agreement between measured gill Cu burden, Cu21 determined
Fluorescence index
(450/500 ratio)

by Cu ion selective electrode, and bioavailable Cu measured

by diffusive gradients in thin films [36], with a 2.1-fold range
1.49
1.43
1.39
1.33
1.31
1.22
1.07

of F [37].
The metal toxicity–versus–SAC relationships (Fig. 2A to
C) yielded about a 4-h increase in LT50 (trout survival time)
per SAC unit for Cu, about a 1-h increase per SAC unit for
Pb, and only about a 20-min increase per SAC unit for Cd,
which was not significant. The weak influence of NOM metal
(cm2/mg)

binding quality and also quantity for Cd (Fig. 2C and Table

SAC

16.9
11.5
19.8
19.8
21.2
29.9
4.2

3) can be related to the weaker binding of Cd to NOM (log

KCd-NOM 5 7.4) compared to Cd binding to the gills (log KCd-
gillCd 5 8.6). Van Ginneken et al. [38] proposed an even lower
binding constant for the Cd-humic acid complex (log KCd-HA
5 6.6), which would make the difference even greater. In
contrast, both Cu and Pb are modeled to bind about 50 and
250 times more strongly to NOM than to the gills (Fig. 1 and
(mg C/L)
TOC

Appendix 1), so both NOM metal binding quality and quantity

5.1
5.6
5.9
5.7
5.4
5.7
6.0

have greater influence (Tables 1 and 2 and Fig. 2A and B). In

agreement with our toxicity data, the Baltic blue mussel My-
tilus trossulus exposed for 21 d in Baltic seawater accumulated
less Cu in the presence of more optically dense NOM of greater
C content (e.g., more aromatic NOM), while the same Nor-
wegian NOM samples did not affect Cd accumulation by the
mussels [39].
Ducks Unlimited Pond

In Macdonald et al. [4], we discussed why the inclusion of

an NOM metal binding quality factor in developing biotic
Suwannee River
Sanctuary Pond
Lac Mistaouac

ligand models is important. Reasons include possible effects

Luther Marsh
NOM sample

Lake Ontario

of global warming plus atmospheric ozone depletion on re-

Lac Jouve

duced inflow of high-quality NOM to lakes as precipitation

decreases and greater degradation of high-quality NOM as
more ultraviolet light arrives at lake surfaces [40]. Remaining
Effect of NOM source on acute metal toxicity to trout
NOM would be of lower metal binding quality and therefore
would confer less protection against metal toxicity. For these
and other reasons [4], inclusion of a NOM metal binding qual-
ity factor in BLMs is important to make them more flexible
and more comprehensive.
Although SAC is easy to measure with standard laboratory
spectrophotometers and is a good proxy measurement for
NOM aromaticity (e.g., Table 4), SAC measurements are not
very reliable at low TOC concentrations, especially for opti-
cally light, authochthonous-like NOM. In practice, NOM con-
centrations of 5 to 10 mg C/L are needed to yield reliable
absorbance readings using a standard 1-cm path length. There-
fore, NOM will usually need to be concentrated to yield re-
liable SAC values; reverse osmosis is a good method to con-
centrate NOM [41]. In addition, although ultrapure water is
used as the zero solution, a reference NOM for a high SAC
reference value is needed. For a high reference standard we
suggest using an aromatic NOM, such as Suwannee River
NOM available from the International Humic Substances So-
ciety, or a similar swamp NOM (e.g., Luther Marsh) that yields
SAC340 values of 25 to 30 and therefore F 5 1.0 (see Tables
1 to 3). Alternatively, very optically dense commercial Aldrich
humic acid can be used as a reference standard, with an SAC340
of about 50 and therefore F 5 1.2.
Titration of organic matter typically reveals functional
groups with pKa ranges of 3.5 to 6.0 and 7.0 to 8.5, which are
likely carboxylic acids and b-dicarbonyls, enols, and alcohols,
respectively, along with a higher pKa range of 9.0 to 10.5 due
to phenolic acids [42]. Phenol groups, associated with aromatic
carbon structures that absorb light, are thought to bind metals
strongly, whereas carboxylic acids and alcohols tend to exhibit
weaker metal binding and are generally associated with ali-
phatic carbon [24]. Using single conditional log K values for
metal-NOM binding (Fig. 1 and Appendix) is a simplification
of a complex system [24,27,43]. However, for toxicologically
relevant concentrations of NOM and Cu, Pb, and Cd, single
log K values for metal-NOM interactions may be adequate
[7,44] as long as NOM metal binding quality is considered as
well as NOM quantity. Because NOM differs in color de-
pending on source, where lighter, autochthonous-like NOM is
more aliphatic compared to optically dark, allochthonous
NOM, which is more aromatic, a measure of color such as
SAC appears to be a good proxy measurement for estimating
metal binding behavior of NOM.
Certainly, SAC is not the only optical method to quantify
NOM aromaticity. The fluorescence index of McKnight et al.
[29] works well (Table 4). Imai et al. [45] found that ultraviolet
absorbance (260 nM) to dissolved organic carbon ratios were
useful for evaluating NOM characteristics. For example, al-
lochthonous river water had a high ultraviolet/dissolved or-
ganic carbon ratio, lake water had an intermediate ratio, and
domestic sewage had a low ultraviolet/dissolved organic car-
bon ratio; these ratios are essentially SAC values at a lower
wavelength. Another promising optical analysis of NOM is
excitation-emission matrix spectroscopy [46–48 and referenc-
es therein]. These NOM fingerprints are useful in determining
the terrigenous nature of NOM due to the humic-like peaks at
wavelengths of 260 nM (excitation), 380 to 460 nM (emission),
350 nM (excitation), and 420 to 480 nM (emission). These
scans are more sophisticated than SAC but are hard to interpret
quantitatively unless converted to a fluorescence index [29] or
unless peak volumes in specific regions are integrated [48].
In our laboratory we use ion-poor water for maximal metal
Environ. Toxicol. Chem. 23, 2004 2897
interactions at fish gills and therefore high metal toxicity and
add constituents such as NOM to the water to quantify their
influence on this worst-case metal toxicity. For the Pb expo-
sures we supplemented the ion-poor water with Ca to eliminate
detrimental effects of NOM plus Pb seen occasionally by us
when developing the Pb-gill binding model [4]. That is, it
appears that sometimes NOM binds what little Ca is in the
ion-poor water so that residual Pb not bound by NOM can
interact at the gills; adding Ca decreases this effect but does
not always eliminate it (see Table 2). In addition to supplying
more Ca to bind at the gills, more Ca in water reduces the
diffusion gradient of Ca to the water from inside the fish (;2
mM inside [49]), which increases survival time, especially for
small fish [50].
In retrospect, we should also have added more Ca to the
ion-poor Cd exposure water to decrease the detrimental effect
of NOM sequestering scarce Ca and sometimes increasing Cd
toxicity (Table 3). However, Ca sequestration by NOM may
not explain all the effects of NOM on Cd toxicity; studies with
approximately 1 mg C/L Aldrich humic acid showed increased
Cd toxicity to daphnids through some unknown mechanism
[51–53]. Note that this Ca sequestering effect of NOM does
not appear with Cu, where NOM was always clearly beneficial
(Table 1; see also Richards and Playle [3]), likely because Cd
and Pb act as Ca analogues [4], whereas Cu does not (Cu
inhibits Na uptake at fish gills; see Grosell et al. [50]). Recent
aluminum-gill modeling also indicated that Ca binding to fish
gills can be decreased by NOM [54]. In our experience det-
rimental effects of NOM through sequestered Ca appear only
in very ion-poor water. For example, Pb was never more toxic
to trout in diluted natural water samples compared to Pb ex-
posures in ion-poor water, which represent worst-case condi-
tions (e.g., table 1 in Macdonald et al. [4]).
In summary, our metal toxicity data indicate that biologi-
cally meaningful differences indeed exist in NOM isolated
from different sources. The general pattern for Cu, Pb, and to
some degree Cd is that optically darker, more allochthonous
NOM binds metals better than optically lighter, more autoch-
thonous-like NOM, as demonstrated by metal toxicity to rain-
bow trout. A relatively simple NOM metal binding quality
factor based on SAC, which reflects aromatic content of NOM,
suggests about a twofold difference in biologically meaningful
metal binding quality between optically dark and optically
light NOM end members. Whether this pattern holds for other
metals such as Al, Ag, and inorganic Hg is currently under
study in our laboratories.
Note added in proof—De Schamphelaere et al. (2004; [55])
also concluded in their study with Daphnia magna that UV
absorbance of NOM may be good measure of biological and
toxicological differences in Cu binding by NOM, results sim-
ilar to ours for fish.
Acknowledgement—This research was funded by a Strategic Grant
from the Natural Science and Engineering Research Council of Can-
ada, which includes support from BHP Billiton Base Metals and Nor-
anda-Falconbridge, and by Wilfrid Laurier University. Chad Luider
and Kristen Mueller are thanked for their help in the collection of
organic matter in British Columbia and at the Experimental Lakes
Area.
REFERENCES
1. Playle RC, Gensemer RW, Dixon DG. 1992. Copper accumulation
on gills of fathead minnows: Influence of water hardness, com-
2898 Environ. Toxicol. Chem. 23, 2004
plexation and pH of the gill micro-environment. Environ Toxicol
Chem 11:381–391.
2. Bury NR, McGeer JC, Wood CM. 1999. Effects of altering fresh-
water chemistry on physiological responses of rainbow trout to
silver exposure. Environ Toxicol Chem 18:49–55.
3. Richards JG, Playle RC. 1999. Protective effects of calcium
against the physiological effects of exposure to a combination of
cadmium and copper in rainbow trout (Oncorhynchus mykiss).
Can J Zool 77:1035–1047.
4. MacDonald A, Silk L, Schwartz ML, Playle RC. 2002. A lead-
gill binding model to predict acute lead toxicity to rainbow trout
(Oncorhynchus mykiss). Comp Biochem Physiol C Comp Phar-
macol 133:227–242.
5. Pagenkopf GK. 1983. Gill surface interaction model for trace-
metal toxicity to fishes: Role of complexation, pH, and water
hardness. Environ Sci Technol 17:342–347.
6. Morel FMM. 1983. Principles of Aquatic Chemistry. John Wiley,
New York, NY, USA.
7. Playle RC. 1998. Modeling metal interactions at fish gills. Sci
Total Environ 219:147–163.
8. Paquin PR, Santore RC, Wu KB, Kavvadas CD, DiToro DM.
2000. The biotic ligand model: A model of the acute toxicity of
metals to aquatic life. Environmental Science and Policy 3:S175–
S182.
9. Paquin PR, Gorsuch JW, Apte S, Batley GE, Bowles KC, Campbell
PGC, Delos CG, Di Toro DM, Dwyer RL, Galvez F, Gensemer
RW, Goss GG, Hogstrand C, Janssen CR, McGeer JC, Naddy RB,
Playle RC, Santore RC, Schneider U, Stubblefield WA, Wood CM,
Wu BK. 2002. The biotic ligand model: A historical overview.
Comp Biochem Physiol C Comp Pharmacol 133:3–35.
10. DiToro DM, Allen HE, Bergman HL, Meyer JS, Paquin PR, San-
tore RC. 2001. Biotic ligand model of the acute toxicity of metals.
1. Technical basis. Environ Toxicol Chem 20:2383–2396.
11. Santore RC, DiToro DM, Paquin PR, Allen HE, Meyer JS. 2001.
Biotic ligand model of the acute toxicity of metals. 2. Application
to acute copper toxicity in freshwater fish and daphnia. Environ
Toxicol Chem 20:2397–2402.
12. Playle RC, Dixon DG, Burnison K. 1993. Copper and cadmium
binding to fish gills: Modification by dissolved organic carbon
and synthetic ligands. Can J Fish Aquat Sci 50:2667–2677.
13. Playle RC, Dixon DG, Burnison K. 1993. Copper and cadmium
binding to fish gills: Estimates of metal-gill stability constants
and modeling of metal accumulation. Can J Fish Aquat Sci 50:
2678–2687.
14. Janes N, Playle RC. 1995. Modeling silver binding to gills of
rainbow trout (Oncorhynchus mykiss). Environ Toxicol Chem 14:
1847–1858.
15. Richards JG, Playle RC. 1998. Cobalt binding to gills of rainbow
trout (Oncorhynchus mykiss): An equilibrium model. Comp
Biochem Physiol C Comp Pharmacol 119:185–197.
16. Laurén DJ, McDonald DG. 1987. Acclimation to copper by rain-
bow trout, Salmo gairdneri: Physiology. Can J Fish Aquat Sci
44:99–104.
17. Giles MA. 1984. Electrolyte and water balance in plasma and
urine of rainbow trout (Salmo gairdneri) during chronic exposure
to cadmium. Can J Fish Aquat Sci 41:1678–1685.
18. Hollis L, McGeer JC, McDonald DG, Wood CM. 1999. Cadmium
accumulation, gill Cd binding, acclimation, and physiological ef-
fects during long term sublethal Cd exposure in rainbow trout.
Aquat Toxicol 46:101–119.
19. Erickson RJ, Benoit DA, Mattson VR, Nelson HP Jr, Leonard
EN. 1996. The effects of water chemistry on the acute toxicity
of copper to fathead minnows. Environ Toxicol Chem 15:181–
193.
20. Richards JG, Burnison K, Playle RC. 1999. Natural and com-
mercial dissolved organic matter protects against the physiolog-
ical effects of a combined cadmium and copper exposure on rain-
bow trout (Oncorhynchus mykiss). Can J Fish Aquat Sci 56:407–
418.
21. McGeer JC, Szebedinszky C, McDonald DG, Wood CM. 2002.
The role of dissolved organic carbon in moderating the bioavail-
ability and toxicity of Cu to rainbow trout during chronic wa-
terborne exposure. Comp Biochem Physiol C Comp Pharmacol
133:147–160.
22. Morel FMM, Hering JG. 1993. Complexation. In Principles and
Applications of Aquatic Chemistry. John Wiley, New York, NY,
USA, pp 319–420.
M.L. Schwartz et al.
23. Peuravuori J, Pihlaja K. 1999. Some approaches for modeling
dissolved aquatic organic matter. In Keskitalo J, Eloranta P, eds,
Limnology of Humic Waters. Backhuys, Leiden, The Netherlands,
pp 11–39.
24. Perdue EM. 1998. Chemical composition, structure, and metal
binding properties. In Hessen DO, Tranvik LJ, eds, Aquatic Hu-
mic Substances. Springer-Verlag New York, NY, USA, pp 41–
61.
25. Mason AZ, Jenkins KD. 1995. Metal detoxification in aquatic
organisms. In Tessier A, Turner DR, eds, Metal Speciation and
Bioavailability in Aquatic Systems. John Wiley, New York, NY,
USA, pp 479–608.
26. Smith DS, Bell RA, Kramer JR. 2002. Metal speciation in natural
waters with emphasis on reduced sulfur groups as strong metal
binding sites. Comp Biochem Physiol C Comp Pharmacol 133:
65–74.
27. Tipping E. 2002. Cation Binding by Humic Substances. Cam-
bridge University Press, Cambridge, UK.
28. Richards JG, Curtis PJ, Burnison BK, Playle RC. 2001. Effects
of natural organic matter source on reducing metal toxicity to
rainbow trout (Oncorhynchus mykiss) and on metal binding to
their gills. Environ Toxicol Chem 20:1159–1166.
29. McKnight DM, Boyer EW, Westerhoff PK, Doran PT, Kulbe T,
Anderson DT. 2001. Spectrofluorometric characterization of dis-
solved organic matter for indication of precursor organic material
and aromaticity. Limnol Oceanogr 46:38–48.
30. Findlay SEG, Sinsabaugh RL. 2003. Aquatic Ecosystems: Inter-
activity of Dissolved Organic Matter. Academic, New York, NY,
USA.
31. Wetzel RG. 1983. Limnology, 2nd ed. Saunders College Pub-
lishing, Toronto, ON, Canada.
32. Haitzer M, Abbt-Braun G, Traunspurger W, Steinberg CEW. 1999.
Effects of humic substances on the bioconcentration of polycyclic
aromatic hydrocarbons: Correlations with spectroscopic and
chemical properties of humic substances. Environ Toxicol Chem
18:2782–2788.
33. Haitzer M, Burnison K, Hoss S, Traunspurger W, Steinberg CEW.
1999. Effects of quantity, quality, and contact time of dissolved
organic matter on bioconcentration of benzo[a]pyrene in the nem-
atode Caenorhabditis elegans. Environ Toxicol Chem 18:459–
465.
34. Gourlay C, Tusseau-Vuillemin M-H, Garric J, Mouchel J-M.
2003. Effect of dissolved organic matter of various origins and
biodegradabilities on the bioaccumulation of polycyclic aromatic
hydrocarbons in Daphnia magna. Environ Toxicol Chem 22:
1288–1294.
35. Bryan SE, Tipping E, Hamilton-Taylor J. 2002. Comparison of
measured and modelled copper binding by natural organic matter
in freshwaters. Comp Biochem Physiol C Comp Pharmacol 133:
37–49.
36. Davison W, Zhang H. 1994. In situ speciation measurements of
trace components in natural waters using thin-film gels. Nature
367:546–548.
37. Luider CD, Crusius J, Playle RC, Curtis PJ. 2004. Influence of
natural organic matter source on copper speciation as demon-
strated by Cu binding to fish gills, by ion selective electrode, and
by DGT gel sampler. Environ Sci Technol 38:2865–2872.
38. Van Ginneken L, Bervoets L, Blust R. 2001. Bioavailability of
Cd to the common Carp, Cyprinus carpio, in the presence of
humic acid. Aquat Toxicol 52:13–27.
39. Pempkowiak J, Kożuch J, Grzegowska H, Gjessing ET. 1999.
Biological versus chemical properties of natural organic matter
isolated from selected Norwegian lakes. Environ Int 25:357–366.
40. Schindler DW. 2001. The cumulative effects of climate warming
and other human stresses on Canadian freshwaters in the new
millennium. Can J Fish Aquat Sci 58:18–29.
41. Sun L, Perdue EM, McCarthy JF. 1995. Using reverse osmosis
to obtain organic matter from surface and ground waters. Water
Res 29:1471–1477.
42. Clair TA, Kramer JR, Sydor M, Eaton D. 1991. Concentration
of aquatic dissolved organic matter by reverse osmosis. Water
Res 25:1033–1037.
43. Town RM, Filella M. 2000. Dispelling the myths: Is the existence
of L1 and L2 ligands necessary to explain metal ion speciation
in natural waters? Limnol Oceanogr 45:1341–1357.
44. McGeer JC, Playle RC, Wood CM, Galvez F. 2000. A physio-
logically based biotic ligand model for predicting the acute tox-
Effect of NOM source on acute metal toxicity to trout Environ. Toxicol. Chem. 23, 2004 2899

icity of waterborne silver to rainbow trout in freshwaters. Environ
Sci Technol 34:4199–4207.
45. Imai A, Fukushima T, Matsushige K, Kim YH. 2001. Fraction-
ation and characterization of dissolved organic matter in a shallow
eutrophic lake, its inflowing rivers, and other organic matter
sources. Water Res 35:4019–4028.
46. Coble PG. 1996. Characterization of marine and terrestrial DOM
in seawater using excitation-emission matrix spectroscopy. Mar
Chem 51:325–346.
47. Blough NV, Del Vecchio R. 2002. Chromophoric DOM in the
coastal environment. In Hansell DA, Carlson CA, eds, Biogeo-
chemistry of Marine Dissolved Organic Matter. Academic, New
York, NY, USA, pp 509–578.
48. Chen W, Westerhoff P, Leenheer JA, Booksh K. 2003. Fluores-
cence excitation-emission matrix regional integration to quantify
spectra for dissolved organic matter. Environ Sci Technol 37:
5701–5710.
49. Wood CM, Hogstrand C, Galvez F, Munger RS. 1996. The phys-
iology of waterborne silver toxicity in freshwater rainbow trout
(Oncorhynchus mykiss) 1. The effects of ionic Ag1. Aquat Toxicol
35:93–109.
50. Grosell M, Nielsen C, Bianchini A. 2002. Sodium turnover rate
determines sensitivity to acute copper and silver exposure in
freshwater animals. Comp Biochem Physiol C Comp Pharmacol
133:287–303.
51. Winner RW. 1984. The toxicity and bioaccumulation of cadmium
and copper as affected by humic acid. Aquat Toxicol 5:267–274.
52. Winner RW. 1985. Bioaccumulation and toxicity of copper as
affected by interactions between humic acid and water hardness.
Water Res 19:449–455.
53. Winner RW, Gauss JD. 1986. Relationship between chronic tox-
icity and bioaccumulation of copper, cadmium, and zinc as af-
fected by water hardness and humic acid. Aquat Toxicol 8:149–
161.
54. Wood CM, Matsuo AY, Wilson RW, Gonzalez RJ, Patrick MJ,
Playle RC, Val AL. 2003. Protection by natural blackwater against
disturbances in ion fluxes caused by low pH exposure in fresh-
water stingrays endemic to the Rio Negro. Physiol Biochem Zool
76:12–27.
55. De Schamphelaere KAC, Vasconcelos FM, Tack FMG, Allen HE,
Janssen CR. 2004. Effect of dissolved organic matter source on
acute copper toxicity to Daphnia magna. Environ Toxicol Chem
23:1248–1255.

APPENDIX
Log-conditional equilibrium constants (K ) and number of binding sites for the metal-gill binding models (left columns), plus some stability
constants from the MINEQL1 chemical equilibrium program (Environmental Research, Hallowell, ME, USA) (right columns). See text for
details; NOM 5 natural organic matter

Complex Log K Complex Log K

Cu-gillCu 7.4 CuCO3(aq) 6.7

Pb-gillPb 6.0 Cu(CO3)222 10.1
Cd-gillCd 8.6 CuHCO13 12.1
Ca-gillCu 3.4 CuOH1 27.7
Ca-gillPb 4.0 PbCO3(aq) 6.4
Ca-gillCd 5.0 Pb(CO3)222 9.9
H-gillCu 5.4 PbHCO13 13.1
H-gillPb 5.0 PbOH1 27.6
H-gillCd 6.7 CdCO3(aq) 4.3
Cu-NOM 9.1 Cd(CO3)222 7.2
Pb-NOM 8.4 CdHCO13 10.6
Cd-NOM 7.4 CdOH1 212.7
Ca-NOM 5.0 CaCO3(aq) 3.0
H-NOM 4.0 CaHCO13 11.5
Total -gillCu sites: 2.0 nmol/L Total Cu-NOM sites: 0.2 mmol per mg C
Total -gillPb sites: 6.5 nmol/L Total Pb-NOM sites: 0.5 mmol per mg C
Total -gillCd sites: 2.0 nmol/L Total Cd-NOM sites: 0.2 mmol per mg C