Integrated Environmental Assessment and Management Volume 4, Number 1pp.

1523
2008 SETAC

15

Consideration of Exposure and Species Sensitivity of

Triclosan in the Freshwater Environment
Marie Capdevielle,* Roger Van Egmond,` Mick Whelan, Donald Versteeg, //
Matthias Hofmann-Kamensky,# Josef Inauen,# Virginia Cunningham, and Daniel Woltering``

ABSTRACT
Triclosan (TCS) is a broad-spectrum antimicrobial used in consumer products including toothpaste and hand soap. After
being used, TCS is washed or rinsed off and residuals that are not biodegraded or otherwise removed during wastewater
treatment can enter the aquatic environment in wastewater effluents and sludges. The environmental exposure and toxicity
of TCS has been the subject of various scientific and regulatory discussions in recent years. There have been a number of
publications in the past 5 y reporting toxicity, fate and transport, and in-stream monitoring data as well as predictions from
aquatic risk assessments. State-of-the-science probabilistic exposure models, including Geography-referenced Regional
Exposure Assessment Tool for European Rivers (GREAT-ER) for European surface waters and Pharmaceutical Assessment and
Transport Evalutation (PhATEe) for US surface waters, have been used to predict in-stream concentrations (PECs). These
models take into account spatial and temporal variability in river flows and wastewater emissions based on empirically
derived estimates of chemical removal in wastewater treatment and in receiving waters. These model simulations (based on
realistic use levels of TCS) have been validated with river monitoring data in areas known to be receiving high wastewater
loads. The results suggest that 90th percentile (low flow) TCS concentrations are less than 200 ng/L for the AireCalder
catchment in the United Kingdom and between 250 ng/L (with in-stream removal) and 850 ng/L (without in-stream removal)
for a range of US surface waters. To better identify the aquatic risk of TCS, a species sensitivity distribution (SSD) was
constructed based on chronic toxicity values, either no observed effect concentrations (NOECs) or various percentile adverse
effect concentrations (EC1025 values) for 14 aquatic species including fish, invertebrates, macrophytes, and algae. The SSD
approach is believed to represent a more realistic threshold of effect than a predicted no effect concentration (PNEC) based
on the data from the single most sensitive species tested. The log-logistic SSD was used to estimate a PNEC, based on an
HC5,50 (the concentration estimated to affect the survival, reproduction and/or growth of 5% of species with a 50%
confidence interval). The PNEC for TCS was 1,550 ng/L. Comparing the SSD-based PNEC with the PECs derived from GREATER and PhATE modeling to simulate in-river conditions in Europe and the United States, the PEC to PNEC ratios are less than
unity suggesting risks to pelagic species are low even under the highest likely exposures which would occur immediately
downstream of wastewater treatment plant (WWTP) discharge points. In-stream sorption, biodegradation, and photodegradation will further reduce pelagic exposures of TCS. Monitoring data in Europe and the United States corroborate the
modeled PEC estimates and reductions in TCS concentrations with distance downstream of WWTP discharges. Environmental metabolites, bioaccumulation, biochemical responses including endocrine-related effects, and community level
effects are far less well studied for this chemical but are addressed in the discussion. The aquatic risk assessment for TCS
should be refined as additional information becomes available.
Keywords: Triclosan

Aquatic risk assessment

Species sensitivity distribution

INTRODUCTION
The environmental exposure and toxicity of triclosan
(TCS) has been the subject of various scientific and regulatory
discussions in recent years. These discussions have been
motivated by the proximity of the predicted environmental
concentration (PEC) to the predicted no effect concentration
(PNEC) based on most sensitive species conventional aquatic
risk assessment methods (for example, as described in the EU
* To whom correspondence may be addressed: marie_capdevielle@colpal.com
Published on the Web 9/13/2007.

Probabilistic exposure assessment

Technical Guidance Documents; EC 2003). Although the

conventional approach provides a reasonable (screening-level)
1st approximation of environmental risk, this risk is derived
using conservative assumptions and assessment factors and
may not reflect actual environmental situations.
Triclosan is a broad-spectrum antimicrobial used in a
variety of consumer products including toothpaste, shampoos,
deodorants, skin lotions, and hand soaps. Its concentration in
these products is typically in the range of 0.1% to 0.3%. After
being used, TCS is washed or rinsed off and may enter the
environment via local wastewater treatment plants (WWTPs)

Original Research

(Received 25 March 2007; Accepted 11 September 2007)

Review

Colgate-Palmolive, 909 River Road, Piscataway, New Jersey 08855, USA

`Unilever, Safety and Environmental Assurance Centre, Sharnbrook, Bedford MK44 1LQ, United Kingdom
Cranfield University, Department of Natural Resources, School of Applied Sciences, Building 88, Cranfield, Bedfordshire, Cranfield,
Bedfordshire, MK43 0AL, United Kingdom
// Procter and Gamble, Central Product Safety, 11810 East Miami River Road, Cincinnati, Ohio 45252, USA
#Ciba Specialty Chemicals, Environmental Safety & Compliance, Regulatory Services, Klybeck 424.214, CH-4002, Basel.214, CH-4002,
Basel, Switzerland
GlaxoSmithKline, Corporate Environment, Health and Safety, 200 N. 16th Street, Philadelphia, Pennsylvania 19102, USA
``Alexandria, Virginia, USA

16

where typically 90% to 98% is removed as a result of

biodegradation and sorption (Singer et al. 2002; Bester 2003,
2005; Sabaliunas et al. 2003; Thompson et al. 2005; Heidler
and Halden 2006; Waltman et al. 2006). Between 50% and
70% of the removal is attributed to biodegradation (Singer et
al. 2002; Heidler and Halden 2006). Triclosan has been shown
in numerous studies to be biodegradable and photo-instable
causing it to continue to breakdown following its release into
the aquatic environment. Federle and Schwab (2003)
reported a mineralization half-life of 2.5 to 3.5 d in a
laboratory river water die-away study with radiolabeled TCS.
Measured in-stream removal half-lives for TCS have been
reported to range between 2.1 h (Sabaliunas et al. 2003) and
11.6 h (Morrall et al. 2004) but are likely to be considerably
higher (of the order of days) in lakes (e.g., Singer et al. 2002;
Tixier et al. 2002). A TCS half life of 11 d for the Ruhr River
in Germany was suggested by Bester (2005). However, this
estimate was based on concentrations in single grab samples
collected on the same day from stations separated by an
estimated time-of-travel of about 6 d and thus is difficult to
consider as part of this evaluation.
There is some experimental evidence on TCS environmental degradates. A series of studies (e.g., Latch et al. 2005)
show that photodegradation of TCS produced 2,4-dichlorophenol and 2,8-dichlorodibenzo-p-dioxin (DCDD). The
2,4-dichlorophenol itself is known to be biodegradable as
well as photodegradable (EC 2000). For DCDD, a conversion
rate of 1% has been reported and estimated half-lives suggest
that it is photolabile as well (Aranami and Readman 2007).
The formationdecay kinetics of DCDD are also reported by
Sanchez-Prado et al. (2006). Several papers report the
formation of the potential biotransformation product methyl-triclosan. Observed concentrations in surface waters are
generally at or below the limit of quantitation (Singer et al.
2002). These concentrations suggest that methyl-triclosan is
formed, but also degraded with a slower kinetic than TCS.
Further discussion of TCS metabolites is beyond the scope of
this paper, although additional information on formation and
degradation half-lives would be helpful.
Heretofore, the most recent aquatic risk assessment for
TCS was published by Reiss et al. (2002). This was based on a
probabilistic exposure estimation that combined measured
and modeled fate and transport data along with the screening
level PNEC (that is, using the lowest reported chronic
concentration). According to Reiss et al. (2002), measured
TCS concentrations in US wastewater effluents range from
0.2 to 2.7 lg/L, and the lowest no-observed-effect concentration (NOEC) for algae, the most sensitive species, is less
than 1 lg/L. Thus Reiss et al. (2002) concluded that effluents
with TCS concentrations at the high end of the range are
discharged into rivers with low dilution, TCS concentrations
may exceed the algal NOEC.
The conventional PNEC for TCS is 70 ng/L, which was
derived from the lowest reported NOEC for most sensitive
aquatic species, Scenedesmus subspicatus (i.e., 700 ng/L
reported by Orvos et al. [2002]). This effect is believed to
be algistatic rather than algicidal since algae exposed to TCS
at concentrations up to 13 lg/L do recover after dilution of
the test medium to noninhibitory concentrations (Orvos et al.
2002). Other species tend to be less sensitive to TCS. Orvos
et al. (2002) report acute and chronic toxicity data for
activated-sludge microorganisms, green and blue green algae,
marine and freshwater diatoms, duckweed, Daphnia magna

Integr Environ Assess Manag 4, 2008M Capdevielle et al.

and Ceriodaphnia, fathead minnows, bluegill sunfish, and

rainbow trout.
Based on probabilistic exposure estimates, Reiss et al.
(2002) concluded that, the risk to fish, invertebrates, and
vascular aquatic plants appear to be of no concern. However,
for some rivers with small dilution volumes during low flow
conditions, it is possible that some algae species will be
affected immediately downstream of WWTP discharges, but
because of dissipation in the water column environmental
concentrations are reduced downstream, leading to reduced
risk to aquatic organisms. Reiss et al. (2002) did not include
a probabilistic effects assessment endpoint. The species
sensitivity distribution (SSD) approach is believed to provide
a more robust and realistic indication of the threshold of
effect in the natural environment, compared with the
conventional (single species) approach (Versteeg et al.
1999). Guidance provided in the EU Technical Guidance
Documents (Part II, Section 3.3) suggests that a reliable
PNEC can be derived by the SSD method if the database
contains at least 10 NOECs for different species covering at
least 8 taxonomic groups. The EU Technical Guidance
Documents further recommends that a PNEC derived using
this method should be the concentration below which 5% of
species are affected, taking into account a 50% confidence
interval, termed the HC5,50. The US Environmental
Protection Agencys (USEPAs) Guidelines for Ecological Risk
Assessment (USEPA 1998, section 5.1) describe a comparable
method to derive effect distributions from single-species
toxicity data. This is an accepted and often used approach for
risk estimation (e.g., van Straalen 2002; USEPA 2003; Hose
and Van den Brink 2004; EUFRAM 2005; OECD 2005;
Straub and Stewart 2007). In fact, Versteeg et al. (1999) have
shown that the HC5,50 with at least 5 species in the
distribution is protective of communities exposed in mesocosms (i.e., the HC5,50 , NOECmesocosm).
In this paper, an aquatic risk assessment for TCS is
presented which combines a probabilistic exposure assessment (validated using monitoring data) with a PNEC derived
from an SSD. The SSD analysis was based on published
chronic toxicity data for 14 aquatic species. The exposure
assessment uses realistic estimates of wastewater loads,
removal during treatment, dilution, and fate of TCS (not
the breakdown products) in receiving surface waters for
various catchments in Europe and North America. While the
focus of this paper is on those studies that are used in aquatic,
pelagic risk assessment as routinely practiced in the United
States and Europe, other studies which evaluate biochemical
responses, including endocrine-related effects, are considered
in the discussion.

METHODS
Toxicity assessment
A single SSD was constructed from published chronic
NOEC and EC1025 values for 14 aquatic species using
methods described in Versteeg et al. (1999). The data were
not adjusted for ionization although, with a pKa of 8.1, some
of the TCS in the aquatic environment is expected to be
ionized. Biological membranes are generally permeable to unionized molecules and relatively impermeable to the ionized
species of molecules (Cohn 1979), thus failure to take the
ionized fraction into account in this toxicity assessment is
believed to be conservative. Consistent with this, Orvos et al.

Triclosan Aquatic Risk AssessmentIntegr Environ Assess Manag 4, 2008

(2002) demonstrated that the ionized form of TCS is less

toxic to algae than the neutral form. An acute and chronic
PNEC for TCS was estimated using the HC5,50 value in an
SSD that was generated using published toxicity data (not
adjusted for ionization). The method for compiling distributions of single-species toxicity test data is detailed in Versteeg
et al. (1999). In those situations where more than 1 study was
available for a given species, the geometric mean value was
used. Where required, NOEC and EC1025 values were
combined into an estimate of the low effect concentration
for a species. A log-logistic curve was fitted to the data using
nonlinear regression.
Exposure assessment
A modeling approach was also used to predict TCS
exposure in surface waters in Europe and in the United
States. Predicted environmental concentrations were derived
and compared with measured TCS data.
Europe
Geography-referenced Regional Exposure Assessment Tool
for European Rivers (GREAT-ER) was used to predict
statistical distributions of river water TCS concentrations in
the AireCalder catchment in the United Kingdom. The
United Kingdom is considered to be a high-consumption
country for TCS (typically 1 g TCS per capita per y). The
GREAT-ER is a GIS-based model which was developed to
calculate the PECs (also referred to in GREAT-ER as Csims, or
simulated concentrations) for down-the-drain chemicals in
European surface waters (Feijtel et al. 1997; Koormann et al.
2006). The model uses spatially referenced data on the
locations and sizes of WWTPs, chemical removal during
wastewater treatment, a digital river network and associated
flow statistics and in-stream removal processes to calculate instream exposures. Key input parameters such as removal in
WWTPs and river flow are described using probability density
functions, which are combined using Monte Carlo simulation
(Warn and Brew 1980). Results are expressed as probability
distributions for chemical concentrations in individual river
reaches, which can be visualized as color-coded maps or as
concentration profiles. Model output has been compared with
monitoring data for a number of catchments across Europe
and has been shown to generate reasonably accurate
predictions (e.g., Gandolfi et al. 2000; Schroder et al. 2002).
The AireCalder catchment (approximately 1,960 km2 at
its closing section) is situated in West Yorkshire, in the north
of England. It is characterized by a high population density
(approximately 925 inhabitants/km2; Keller et al. 2007) and
includes the cities of Leeds and Bradford as well as smaller
towns and villages. A large number of WWTPs with various
sizes, discharge effluent into receiving waters with dilution
factors as low as 2. The catchment was one of the main casestudy areas for the GREAT-ER project (Holt et al. 2003). The
Aire below its confluence with the Calder carries treated
sewage effluent from a population of 1,900,000 and
approximately 80% of Yorkshires industrial effluent (Holt
et al. 2003). In dry weather, approximately 65% of the flow in
the Aire above the confluence with the Calder is derived from
treated effluent (Holt et al. 2003). A limited number of grab
samples (6 samples from 4 different stations) have been
collected from the AireCalder system and analyzed for TCS
by the Environment Agency of England and Wales. All
sampling stations were located downstream of WWTPs.

17

Samples were analyzed at the Environment Agency Nottingham laboratory by GC-MS and following the method
described by McAvoy et al. (2002) but substituting acylation
in place of silylation for TCS derivatization. The sampling
dates were 10, 11, and 18 February; 3 and 17 March; and 5
April 2005. Standard laboratory procedures were followed
(e.g., glass bottles, 4 8C storage in the dark, less than 10-d
holding times, blanks and blank spiked analytical quality
control samples). The recoveries were greater than 90% and
the reported limit of quantification was 5 ng/L.
For the application of GREAT-ER to the AireCalder, the
following assumptions were made: a use rate of 1 g TCS per
capita per year, a uniform distribution of WWTP removal
rates with a range of 90% to 98%, and 1st order kinetics for instream removal of TCS with a rate constant of 0.21/h
corresponding to a half-life of 3.3 h. The in-stream removal
rate was derived from a monitoring study employing dyetracing conducted by Sabaliunas et al. (2003) in the Mag
Brook, a tributary of the river Calder. Sorption, biodegradation, and photodegradation all contributed to the losses of
TCS in the field. This value (i.e., half-life of 3.3 h) is
considered to be characteristic of relatively small and shallow
streams of Northern Europe and its use here is reasonable
since it was estimated in the same catchment as the one
discussed. However, the actual rate constant for TCS will
probably vary significantly in space (particularly with river
depth; Boeije et al. 2000) and seasonally (e.g., with temperature). It should be noted that the rate constant assumed here
refers to primary degradation (i.e., disappearance of the
parent molecule) rather than to complete mineralization.
North America
For modeling river water exposure to TCS in North
America, the Pharmaceutical Assessment and Transport
Evalutation (PhATEe) model was applied. PhATE uses a
mass balance approach to estimate PECs in 11 catchments
selected to be representative of the range of hydrologic
regions in the United States (Anderson et al. 2004). These
catchments range widely in size and include between 19 and
more than 1,100 WWTPs as point sources of TCS. The recent
paper by Anderson et al. (2004) includes the results of TCS
simulations using PhATE. The key input parameters for TCS
were 600,000 kg/y annual use; no loss prior to entering
wastewater treatment; WWTP removal of 32% for primary
treatment only, 90% for primary plus secondary treatment,
and 95% for tertiary treatment. In-stream losses of TCS in the
evaluations were assumed to take place with a rate constant of
0.23/d (corresponding to a half-life of approximately 3 d), or
zero. The 0.23/d value was taken from a laboratory
mineralization study reported by Federle and Schwab
(2003) which is likely to be conservative with respect to
TCS degradation in the field. Biodegradation in vitro is often
limited by poor contact with fixed-film microbial biomass.
Furthermore, complete mineralization in the laboratory will
necessarily take longer than primary conversion of parent
TCS to resulting metabolites (Struijs and van den Berg 1995).
Better agreement between modeled and measured data was
obtained with an in-stream loss of zero. This may be due to
the fact that the measured results were from samples taken
close to sewage outfalls, while the PhATE model calculates
concentrations at the end of river segments.
In addition to their modeling efforts, Anderson and
colleagues (2004) reported some model corroboration using

18

comparisons of field data for TCS reported by Kolpin et al.

(2002) with PhATE model PECs for the same locations on a
point-by-point basis. The field data were derived from a
nationwide US Geological Survey stream survey project
which involved 139 streams across the country (Kolpin et
al. 2002) chosen from areas known to be receiving relatively
high wastewater loads.
Risk assessment
Ecological risk assessments make comparisons between a
PEC and a PNEC. In conventional assessments a PEC to
PNEC ratio greater than unity indicates a risk (i.e., exposures
exceed the adverse effect threshold for sensitive species). In
this study the PECs derived from the PhATE and GREAT-ER
models were compared to PNEC from SSD.

RESULTS AND DISCUSSION

Predicted no effect concentration
Chronic toxicity values for the 14 species range from 0.53
lg/L for Scenedesmus to 290 lg/L for Oryzias (Table 1). The
algae are consistently more sensitive than either invertebrates
or fish. The lowest chronic NOEC is 0.69 lg/L, derived from
a laboratory toxicity test in which TCS was tested with a pure
culture of Scenedesmus subspicatus exposed for 4 d. For
comparative purposes, acute toxicity values for 13 species are
included in Table 2. The acute values range from 1.4 lg/L for
Scenedesmus to 3,000 lg/L for Chironomus. Again, algae are
the most sensitive species to TCS exposure. There is overlap
between the acute and chronic toxicity values but generally
the acute value is twice (or more) of the chronic value for the
same species.
The available chronic toxicity data was used to develop a
log-logistic distribution (Figure 1). The goodness of fit of the
SSD to the test data met statistical criteria using the
Kolmogorov Smirnov test. The resulting chronic PNEC
(HC5,50 value) is 1.55 lg/L TCS (Figure 1). The species
used to construct the SSD are diverse with broad taxonomic
distribution; the chronic toxicity values came from 14 species
representing fish, invertebrates, and aquatic plant taxa,
exceeding the minimum number of species specified in the
EU Technical Guidance Documents; the most sensitive taxon,
aquatic plants, is over-represented (50% of the data); the most
sensitive biomass endpoint was used to derive chronic values
for algae (growth rate was less sensitive); and, the effect on
algae is believed to be algistatic rather than algicidal. Based on
this information, we expect that chronic, continuous exposure to the PNEC value would be expected to have some
effect on approximately 5% of the species exposed but would
not be expected to have extensive effects on the exposed
community. This PNEC is expected to be protective of the
community in general as studies have demonstrated the
HC5,50 is conservative relative to mesocosm and fieldderived NOECs (Emans et al. 1993; Versteeg et al. 1999).
Thus, the HC5,50 is used as the PNEC. While not included in
the PNEC derivation, the result of a published Microtox assay
(Tatarazako et al. 2004) indicates the sensitivity of these
bacteria (inhibition concentration [IC], IC25 70 lg/L) is in
the same range as that for the other aquatic organisms.
Predicted environmental concentration
EuropeThe modeled PEC distributions in the Aire
Calder catchment, a densely populated area in the United

Integr Environ Assess Manag 4, 2008M Capdevielle et al.

Kingdom, provide an example of a representative worst case

TCS exposure scenario. The model generates a PEC
probability distributions for each river reach, which reflect
variability in key variables such as removal in WWTPs and
river flow. Figure 2 shows the spatial distribution of the 90th
percentile TCS concentrations in the rivers Aire and Calder
predicted by GREAT-ER. The highest 90th percentile PECs
result from a combination of high emission and low river
flow and are shown in red. A profile of changes in measured
and modeled TCS concentrations along the main channel of
the river Aire is also shown in Figure 2. Comparison of the
model prediction with the measured TCS concentration
data, collated by the Environment Agency of England and
Wales (mean 6 1 standard deviation), suggests that GREATER simulations are reasonable for this system, although the
measured data set is rather limited (4 locations, 6 samplings
per location). This confirms an earlier unpublished comparison of GREAT-ER predictions with a more extensive water
quality data set (Holt et al. 2003) that showed that GREATER could match measured concentrations of down-the-drain
chemicals for this catchment. The highest PEC was
generated for the river reach downstream of Bradford. At
this point, the 90th percentile PEC is approximately 180 ng/
L (see insert in Figure 2) and the mean PEC is approximately
100 ng/L.
Estimated TCS use in the United Kingdom is high
compared to other EU countries and therefore aquatic risks
based on UK exposure estimates are reasonably conservative
for the European situation. The stream characteristics
simulated in the AireCalder catchment study may be
directly applied to other relatively small and shallow streams
of Northern Europe. Predicted environmental concentrations
could be different for larger and deeper streams where flow
rate, depth, and turbidity would affect in-stream dilution,
sorption, biodegradation, and photodegradation.
North AmericaFigure 3 provides a comparison of PhATEmodeled and US Geological Surveymeasured TCS levels for
US surface waters. The cumulative probability distribution of
measured TCS concentrations, as reported in Kolpin et al.
(2002), is shown as solid triangles. Nondetects are represented by open triangles. Modeled PECs generated by PhATE
are also shown as a cumulative distribution of concentrations
generated in all river reaches in all catchments considered in
the model. Two model curves are shown in Figure 3: 1 for low
flow conditions (90th percentile PEC 850 ng/L) and 1 for
mean flow conditions (90th percentile PEC 60 ng/L). The
PhATE simulation depicted in Figure 3 was generated
assuming zero in-stream removal of TCS. The same simulation run with an in-stream removal rate of 0.23/d predicts
considerably lower low-flow and mean-flow TCS 90th
percentile PECs (250 ng/L and 25 ng/L respectively). These
PECs are between one-third and one-half the concentrations
predicted using zero in-stream removal.
However, there is better correlation between the measured
data and PhATE predictions when zero in-stream removal is
assumed, especially for the low flow condition. According to
the US Geological Survey authors (Kolpin et al. 2002), the
selection of sampling sites in their nationwide survey was
intentionally focused on streams considered susceptible to
contamination from human, industrial, and agricultural
wastewater loading. The higher TCS concentrations generally
represent locations close to WWTPs where travel times are
too short to allow significant in-stream removal.

Triclosan Aquatic Risk AssessmentIntegr Environ Assess Manag 4, 2008

19

Table 1. Chronic aquatic toxicity data for triclosan

Taxa

Species

Durationa

Endpoint

Referenceb

Comment

156

NOEC

Hatchability and fry survival

9 d ph

290

IC25

Hatchability and fry survival

21 d

Value
(lg/L)

Fish

Oryzias latipes

Fish

O. latipes

Fish

Danio rerio

10 d ph

200

NOEC

Not reported

Fish

D. rerio

14 d ph

160

IC25

Hatchability and fry survival

Fish

Oncorhynchus mykiss

61 d ph

34.1

NOEC

Hatchability and fry survival

Invertebrate

Brachionus calyciflorus

2d

50

NOEC

Not reported

Invertebrate

Ceriodaphnia dubia

7d

182

NOEC

pH 8.5, survival and

reproduction

Invertebrate

C. dubia

7d

NOEC

pH 7.0, survival and

reproduction

Invertebrate

C. dubia

7d

NOEC

Not reported

Invertebrate

C. dubia

7d

170

IC25

Survival and reproduction

Invertebrate

Daphnia magna

21 d

40

NOEC

Survival and reproduction

Invertebrate

Chironomus tentans

10 d

80

NOEC

Survival and growth

Invertebrate

Hyalella azteca

10 d

50

EC10

Survival and growth

Algae

Selenastrum
capricornutum

4d

2.44

EC25

Biomass

Algae

S. capricornutum

4d

2.6

NOEC

Not reported

Algae

S. capricornutum

4d

3.4

IC25

growth

Algae

Scenedesmus
subspicatus

4d

0.69

NOEC

Biomass

Algae

S. subspicatus

3d

0.53

NOEC

Nominal, growth

Algae

Skeletonema

4d

EC20

Biomass

Algae

Anabaena

4d

0.67

EC25

Biomass

Algae

Anabaena

4d

0.97

EC10

Biomass

Algae

Navicula

4d

10.7

EC20

Biomass

Macrophyte

Lemna

10 d

62.5

EC20

Biomass

66

ph posthatch.
b
1 Ishibashi et al. 2004; 2 Tatarazako et al. 2004; 3 Ferrari et al. 2002; 4 Orvos et al. 2002; 5 Dussault et al. 2004.

Risk assessment
The chronic PNEC derived from the SSD and based on
toxicity data for 14 aquatic species is 1,550 ng/L TCS. In
comparison, predicted 90th percentile concentrations based
on a state-of-the science exposure estimation model (GREATER) for a worst-case catchment in the United Kingdom,
suggest that concentrations rarely exceed 200 ng/L TCS and
this might occur in a limited number of locations (e.g.,
immediately downstream of significant WWTP effluents). For
North America, PhATE model predictions suggest that TCS
concentrations as high as 850 ng/L may occur (without instream removal and 250 ng/L with in-stream removal). Both
of these PECs represent 90th percentile (low-flow) simulations. However, in all cases, the PEC to PNEC ratios are less
than unity suggesting risks to sensitive aquatic species are low
even under the highest likely exposures which would occur

immediately downstream of WWTP discharge points. Hence,

at current TCS usage, TCS is not expected to adversely impact
aquatic communities immediately below WWTP effluents. Instream sorption, biodegradation, and photodegradation of
TCS will tend to reduce pelagic exposures further downstream (e.g., Morrall et al. 2004). Monitoring data in Europe
and the United States corroborate the modeled PEC estimates
as well as expected reductions in TCS concentrations with
increasing distance downstream of WWTP discharges.
Additional considerations for TCS
This paper presents a risk assessment for pelagic species
continuously exposed to TCS released through WWTP
effluents. The authors recognize that based on the physical
and chemical characteristics of TCS, (e.g., log Kow) it will sorb
to solids and it may be found in sediments and soils but a
discussion of risk in these compartments is beyond the scope

20

Integr Environ Assess Manag 4, 2008M Capdevielle et al.

Table 2. Acute aquatic toxicity data for triclosan

Taxa

Species

Duration (d)

Value (lg/L)

Endpoint

Referencea

Fish

Oryzias latipes

399

LC50

Fish

Pimephales promelas

260

LC50

Fish

Lepomis macrochirus

440

LC50

Invertebrate

Daphnia magna

343.8

EC50

Invertebrate

Ceriodaphnia dubia

184.7

EC50

Invertebrate

Chironomus tentans

10

3,000

LC50

Invertebrate

Hyalella azteca

10

1,000

LC50

Algae

Selenastrum capricornutum

4.46

EC50

Algae

Scenedesmus subspicatus

1.4

EC50

Algae

Skeletonema

EC50

Algae

Anabaena

1.6

EC50

Algae

Navicula

19.1

EC50

Macrophyte

Lemna

10

62.5

EC50

66

1 Ishibashi et al. 2004; 2 Orvos et al. 2002; 3 Dussault et al. 2004.

of this paper. This pelagic assessment has focused on the use

of predicted and measured water concentrations and data
from laboratory toxicity tests. However, there are a variety of
other fate and effect-related studies that may suggest future
research and may be relevant to better understand TCS fate
and effects in the environment. For example, TCS has been
detected in human breast milk (Adolfsson-Erici et al. 2002),
although in its 2002 report on TCS and its use in consumer
products, the European Commissions Scientific Steering
Committee concluded that any triclosan absorbed by the
human body is rapidly excreted and no long-term or large-

scale accumulation occurs (EC 2002). In addition, TCS

bioconcentrates in fish (Orvos et al. 2002) and other
organisms. Coogan et al. (2007) analyzed filamentous algae
(Cladophora spp.) sampled in the vicinity of a WWTP found
increasing TCS concentrations downstream of the outfall
despite decreasing TCS levels in water. Triclosan has been
reported in the bile of wild fish exposed to a WWTP effluent
(Adolfsson-Erici et al. 2002). While no data are currently
available that correlate tissue concentrations with effect levels
(e.g. Environmental Effects Residue Database; USACE 2006),
the contribution of bioconcentration is factored into this

Figure 1. Species sensitivity distribution. The log-logistic nonlinear regression of chronic aquatic toxicity data. Upper and lower confidence limits are shown with
dashed lines. The resulting HC5,50 value, below which 5% (a 0.05 cumulative probability) of the species responses (no observed effect concentrations [NOECs] or
various percentile adverse effect concentrations [ECx]) occur taking into account a 50% confidence interval, is 1.55 lg/L as shown in red on the x-axis.

Triclosan Aquatic Risk AssessmentIntegr Environ Assess Manag 4, 2008

21

Figure 2. Simulated 90th percentile triclosan (TCS) concentrations in the UK AireCalder catchment using Geography-referenced Regional Exposure
Assessment Tool for European Rivers (GREAT-ER) simulation model along with direct comparison of modeled and measured (in blue) TCS levels in the same river
reach (upper right insert).

assessment as chronic exposure toxicity tests are sufficient in

duration to include effects of bioconcentration on growth,
survival, and reproduction of test species.
In this assessment, ecologically relevant endpoints of
growth, survival, and reproduction are used to evaluate risk
to multiple species and the ecosystem. Studies that evaluate
biochemical responses, including endocrine-related effects,
are difficult to build into a risk assessment unless effects on

survival and reproduction also occur. However, biochemical

endpoints can be used to understand mode of action, suggest
biomarkers of exposure, and rapidly identify species that may
be uniquely sensitive. For example, Canesi et al. (2007) have
shown that mussel immune cells, hemocytes, when exposed
to TCS at 290 to 29,000 lg/L in vitro and to nominal
concentrations of 0.29, 2.9, and 29 ng/g dry weight by
injection showed effects on various enzymes. Endocrine-

Figure 3. The cumulative probability distribution of all triclosan concentrations in US surface waters reported by the US Geological Survey and predicted
environmental concentration (PECs) generated by Pharmaceutical Assessment and Transport Evalutation (PhATETM) for all model segments. Detected triclosan
shown as solid triangles m and nondetects as open triangles D. Model results are shown based on zero in-stream removal. Reproduced with permission,
Copyright 2004, American Chemical Society (Anderson et al. 2004).

22

related effects have been investigated by Foran et al. (2000),

Ishibashi et al. (2004), and Veldhoen et al. (2006). Veldhoen
exposed premetamorphic North American bullfrogs, Rana
catesbeiana, to TCS at 0.15, 1.5, and 22 lg/L and followed
tadpole developmental stages for 18 d in 3,5,3-triiodothyronine (T3) injected and uninjected individuals. 3,5,3-Triiodothyronine injection was used to induce metamorphosis in
the bullfrog, although this species was not ready to undergo
this process naturally at this life stage. Coadministration of T3
and TCS interfered with thyroid-related gene expression and
caused tadpoles to develop faster than controls. Uninjected
tadpoles exposed to TCS showed no difference from controls
in development but did demonstrate a small, transient change
in thyroid receptor alpha in the brain. Ishibashi et al. (2004)
exposed medaka (Oryzias latipes) to TCS at concentrations up
to 137 lg/L for 21 d. Gonadosomatic index was increased in
adults exposed to TCS and hepatic vitellogenin levels were
increased in males exposed to 13 and 61 lg/L TCS. Despite
these changes in biomarkers, there were no adverse effects on
reproduction, hatchability, cumulative mortality, growth, or
sex ratio of offspring at doses up to 137 lg/L. The authors
concluded that a metabolite of TCS was probably weakly
estrogenic. Foran et al. (2000) measured sex ratio and anal fin
length in medaka and concluded that TCS was not estrogenic
but could be weakly androgenic due to the presence of longer
dorsal and anal fins on fish exposed to 100 lg/L.
These biomarker-related effects are difficult to build into a
risk assessment as they involve unrealistic exposure modes
(Veldhoen et al. 2006; Canesi et al. 2007), demonstrate
effects only at elevated concentrations (Foran et al. 2000), or
show a biochemical effect with no effect on growth, survival,
or reproduction of the species (Ishibashi et al. 2004). A recent
paper by Dayan (2007) showed no indication of endocrine
disruption in mammals treated with TCS in relevant high
dose tests of fertility and fetal and neonatal development or in
chronic toxicity and carcinogenicity experiments. There is
good conservation in endocrine systems (e.g., receptor
structure and function) across wide taxonomic distances and
thus an evaluation of endocrine effects on mammals are likely
relevant to fish and amphibians. Endocrine-related effects in
mammals appear to be unlikely and it is reasonable to
question if endocrine-related effects in fish and amphibians
will be supported by additional studies.
The focus of this paper is the application of a recognized
SSD approach to derive a PNEC for pelagic aquatic species
based on standardized single species toxicity tests. The single
species toxicity data (see Figure 1) demonstrate that algae
(including diatoms, green and bluegreen) are the most
sensitive standard test species by a considerable margin over
invertebrates and fish. Investigations of toxicity to algal
assemblages or communities (naturally occurring or artificially derived) can provide a point of comparison to the
predictions from the laboratory tests. Several such algal
assemblage studies have been reported (Wilson et al. 2003;
White et al. 2005; R. Hunt, unpublished data, 2006) in which
exposed natural algal or periphyton communities have been
exposed to TCS. Effects were observed at concentrations
ranging from 0.015 to 1,100 lg/L. In these studies, the lack of
analytical confirmation of exposures, the lack of dose
response, lack of concurrence across studies and issues with
the level of taxonomy makes interpretation difficult. In
addition, the use of natural algal and bacterial communities
in static exposure systems may lead to TCS degradation and
release of the algal nutrient CO2 which may itself have altered

Integr Environ Assess Manag 4, 2008M Capdevielle et al.

algal communities in these studies. At this point in time, valid,

ecologically interpretable algal assemblage/community
NOEC values do not exist for TCS.

CONCLUSIONS
This pelagic aquatic risk assessment for TCS combined a
more sophisticated toxicity (PNEC) methodology with a
probabilistic exposure (PEC) assessment. Based on the PNEC
derived using a species sensitivity distribution and the PECs
derived from GREAT-ER and PhATE modeling to simulate
in-river conditions in Europe and the United States, the PEC
to PNEC ratios are less than unity suggesting risks to sensitive
aquatic species are low even under the highest likely
exposures that would occur immediately downstream of
WWTP discharge points. In reality, in-stream sorption,
biodegradation, and photodegradation of TCS will tend to
reduce pelagic exposures further. Monitoring data in Europe
and the United States corroborate the modeled PEC
estimates as well as expected reductions in TCS concentrations with increasing distance downstream of WWTP
discharges. Environmental metabolites, bioaccumulation,
biochemical responses including endocrine-related effects,
and community level effects are far less well studied and the
aquatic risk assessment for TCS should be continuously
refined as additional information becomes available.

REFERENCES
Adolfsson-Erici M, Pettersson M, Parkkonen J, Sturve J. 2002. Triclosan, a
commonly used bactericide found in human milk and in the aquatic
environment in Sweden. Chemosphere 46:14851489.
Anderson PD, DAco VJ, Shanahan P, Chapra SC, Buzby ME, Cunningham VL,
Duplessie BM, Hayes EP, Mastrocco FJ, Parke NJ, Rader JC, Samuelian JH,
Schwab BW. 2004. Screening analysis of human pharmaceutical compounds
in U.S. surface waters. Environ Sci Technol 38:838849.
Aranami K, Readman J. 2007. Photolytic degradation of triclosan in freshwater
and seawater. Chemosphere 66:10521056.
Bester K. 2003. Triclosan in a sewage treatment processbalances and
monitoring data. Water Res 37:38913896.
Bester K. 2005. Fate of triclosan and triclosan-methyl in sewage treatment plants
and surface waters. Arch Environ Contam Toxicol 49:917.
Boeije GM, Wagner J-O, Koormann F, Vanrolleghem PA, Schowanek DR, Feijtel TCJ.
2000. New PEC definitions for river basins applicable to GIS-based environmental exposure assessment. Chemosphere 40:255265.
Canesi L, Ciacci C, Lorusso LC, Betti M, Gallo G, Pojana G, Marcomini A. 2007.
Effects of triclosan on Mytilus galloprovincialis hemocyte function and
digestive gland enzyme activities: Possible modes of action on non target
organisms. Comp Biochem Physiol Chem Toxicol Pharmacol 145:464472.
Cohn VH. 1979. Transmembrane movement of drug molecules. In: La Du BN,
Mandel HG, Way EL, editors. Fundamentals of drug metabolism and drug
disposition. New York (NY): Robert E. Krieger. p 316.
Coogan MA, Edziyie RE, LaPoint TW, Venables BJ. 2007. Algal bioaccumulation of
triclocarban, triclosan and methyl-triclosan in a North Texas wastewater
treatment plant receiving stream. Chemosphere 67:19111918.
Dayan A. 2007. Risk assessment of triclosan (Irgasant) in human breast milk. Food
Chem Toxicol 45:125129.
Dussault E, Solomon K, Sibley P. 2004. Aquatic toxicity of carbamazepine,
atorvastatin and triclosan to benthic invertebrates [poster]. In: Society of
Environmental Toxicology and Chemistry (SETAC) 25th Annual Meeting; 2004
Nov 1418; Portland, OR. Pensacola (FL): SETAC. Poster IP006.
[EC] European Commission. 2000. IUCLID Dataset on 2,4-dichlorophenol: CAS No.
120-83-2, February 18, European Chemicals Bureau. European Commission.
58 p.
[EC] European Commission. 2003. Technical guidance document on risk assessment in support of Directive 93/67/EEC (new notified substances) and
Regulation 1488/94 (existing substances). European Chemicals Bureau
Institute for Health and Consumer Protection. Part II. 328 p.

Triclosan Aquatic Risk AssessmentIntegr Environ Assess Manag 4, 2008

[EC] European Commission. 2002. Health and Consumer Protection Directorate
General, Opinion on Triclosan Resistance. June 2002. Brussels (BE): EC. Section
V1.1.
Emans HJB, van de Plassche EJ, Canton JH, Okkerman PC, Sparenburg PM. 1993.
Validation of some extrapolation methods used for effect assessment. Environ
Toxicol Chem 12:21392154.
[EUFRAM] European Framework for Probabilistic Risk Assessment of the Environmental Impacts of Pesticides. 2005. Introducing probabilistic methods into
the ecological risk assessment of pesticides. Draft EUFRAM Report, Vol 1, Ver
6. York (UK): Central Science Laboratory.
Federle TW, Schwab EL. 2003. Triclosan biodegradation in wastewater treatment
plant effluent diluted in various river waters. 103rd American Society for
Microbiology General Meeting; 2003; Washington, DC.
Feijtel TCJ, Boeije G, Matthies M, Young A, Morris G, Gandolfi C, Hansen B, Fox K,
Holt M, Koch V, Schroeder R, Cassani G, Schowanek D, Rosenblom J, Niessen
H. 1997. Development of a geographically-referenced regional exposure
assessment tool for European riverGREAT-ER. Chemosphere 34:23512374.
Ferrari B, Paxeus N, Pinto G, Pollio A, Andreozzi R, Marotta R, Vogna D, Lyberatos
G, Garric J. 2002. Fate and effect of triclosan in the aquatic ecosystems: data
for a risk assessment [poster]. In: Society of Environmental Toxicology and
Chemistry Annual Meeting in Europe; 2002 May 1216; Vienna, Austria.
Brussels (BE): SETAC Europe. Poster 73-04.
Foran CM, Bennett ER, Benson WH. 2000. Developmental evaluation of a
potential non-steroidal estrogen: Triclosan. Marine Environ Res 50:153156.
Gandolfi C, Facchi A, Whelan MJ, Cassani G, Tartari G, Marcomini A. 2000.
Validation of the GREAT-ER model in the Lambro catchment. CLER Review
6:8291.
Heidler J, Halden RU. 2006. Mass balance assessment of triclosan removal during
conventional sewage treatment. Chemosphere 66:362369.
Holt MS, Fox KK, Daniel M, Buckland H. 2003. Linear alkylbenzene sulfonate and
boron monitoring in four catchments in the UK. Contribution to GREAT-ER
#11. Sci Total Environ 314316:271288.
Hose GC, Van den Brink PJ. 2004. Confirming the species sensitivity concept for
endosulfan using laboratory, mesocosm, and field data. Arch Environ Contam
Toxicol 47:511520.
Ishibashi H, Matsumura N, Hirano M, Matsuoka M, Shiratsuchi H, Ishibashi Y,
Takao Y, Arizono K. 2004. Effects of triclosan on the early life stages and
reproduction of medaka Oryzias latipes and induction of hepatic vitellogenin.
Aquat Toxicol 67:167179.
Keller V, Rees HG, Fox KK, Whelan M.J. 2007. A practical approach for estimating
the concentrations of down-the-drain chemicals locally and regionally: An
application in England and Wales. Environ Pollut 148:334342.
Kolpin DW, Furlong ET, Meyer MT, Thurman EM, Zaugg SD, Barber LB, Buxton HT.
2002. Pharmaceuticals, hormones and other organic wastewater contaminants in U.S. streams, 19992000: A national reconnaissance. Environ Sci
Technol 36:12021211.
Koormann F, Rominger J, Schowanek D, Wagner J-O, Schroder R, Wind T, Silvani
M, Whelan MJ. 2006. Modelling the fate of down-the-drain chemicals in
rivers: An improved software for GREAT-ER. Environmental Modelling &
Software 21:925936.
Latch DE, Packer JL, Stender BL, VanOverbek J, Arnold WA, McNeill K. 2005.
Aqueous photochemistry of triclosan: Formation of 2,4-dichlorophenol, 2,8dichlorodibenzo-p-dioxin, and oligomerization products. Environ Toxicol
Chem 24:517525.
McAvoy DC, Schatowitz B, Jacob M, Hauk A, Eckoff WS. 2002. Measurement of
triclosan in wastewater treatment systems. Environ Toxicol Chem 21:1323
1329.
Morrall D, McAvoy D, Schatowitz B, Inauen J, Jacob M, Hauk A, Eckhoff W. 2004.
A field study of triclosan loss rates in river water (Cibolo Creek, TX).
Chemosphere 54:653660.
[OECD] Organization for Economic Cooperation and Development. 2005. OECD
SIDS Initial Assessment Report of LAS. In: 20th SIAM meeting; 1921 April

23

2005; Paris, France. Washington, DC: Council for LAB/LAS Environmental

Research. The CLER Review, 10, December 2005. 104 p.
Orvos DR, Versteeg DJ, Inauen J, Capdevielle M, Rothenstein A, Cunningham V.
2002. Aquatic toxicity of triclosan. Environ Toxicol Chem 21:13381349.
Reiss R, Mackay N, Habig C, Griffin J. 2002. An ecological risk assessment for
triclosan in lotic systems following discharge from wastewater treatment
plants in the United States. Environ Toxicol Chem 21:24832492.
Sabaliunas D, Webb SF, Hauk A, Jacob M, Eckhoff WS. 2003. Environmental fate
of triclosan in the River Aire basin, UK. Water Res 37:31453154.
Sanchez-Prado L, Llompart M, Lores M, Garcia-Jares C, Bayona JM, Cela R. 2006.
Monitoring the photochemical degradation of triclosan in wastewater by UV
light and sunlight using solid-phase microextraction. Chemosphere 65:1338
1347.
Schro
der R, Schulze C, Matthies M. 2002. Concentration of LAS and boron in the
Itter. Comparison of measured data with results obtained by simulation with
the GREAT-ER software. Environ Sci Pollut Res 9:130135.
Singer H, Muller S, Tixier C, Pillonel L. 2002. Triclosan: Occurrence and fate of a
widely used biocide in the aquatic environment: Field measurements in
wastewater treatment plants, surface waters, and lake sediments. Environ Sci
Technol 36:49985004.
Straub J, Stewart K. 2007. Deterministic and probabilistic acute-based environmental risk assessment for naproxen for Western Europe. Environ Toxicol
Chem 26:795806.
Struijs J, van den Berg R. 1995. Standardized biodegradability tests: Extrapolation
to aerobic environments. Water Res 29:255262.
Tatarazako N, Ishibashi H, Teshima K, Kishi K, Arizono K. 2004. Effects of triclosan
on various aquatic organisms. Environ Sci 11:130133.
Thompson A, Griffen P, Stuetz R, Cartmell E. 2005. The fate and removal of
triclosan during wastewater treatment. Water Environ Res 77:6367.
Tixier C, Singer HP, Canonica S, Muller SR. 2002. Phototransformation of triclosan
in surface waters: A relevant elimination process for this widely used biocide
Laboratory studies, field measurements, and modeling. Environ Sci Technol
36:34823489.
[USACE] US Army Corps of Engineers. 2006. Environmental Effects Residue
Database. http://el.erdc.usace.army.mil/ered/index.html#misc. Accessed 24
August 2007.
[USEPA] US Environmental Protection Agency. 1998. Guidelines for ecological risk
assessment. Risk assessment forum. Washington DC: Office of Research and
Development. EPA/630/R-95/002F.
[USEPA] US Environmental Protection Agency. 2003. Ambient aquatic life water
quality criteria for atrazine. Washington DC: Office of Water. EPA-822-R-03023.
van Straalen NM. 2002. Theory of ecological risk assessment based on species
sensitivity distributions. In: Posthuma L, Suter GW II, Traas TP, editors. Species
sensitivity distributions in ecotoxicology. Boca Raton (FL): Lewis. p 3748.
Veldhoen N, Skirrow RC, Osachoff H, Wigmore H, Clapson DJ, Gunderson MP, Van
Aggelen G, Helbing CC. 2006. The bacterial agent triclosan modulates thyroid
hormone-associated gene expression and disrupts postembryonic anuran
development. Aquat Toxicol 80:217227.
Versteeg DJ, Belanger S, Carr CJ. 1999. Understanding single species and model
ecosystem sensitivity: A data based comparison. Environ Toxicol Chem
18:13291346.
Waltman E, Venables B, Waller W. 2006. Triclosan in a North Texas wastewater
treatment plant and the influent and effluent of an experimental constructed
wetland. Environ Toxicol Chem 25:367372.
Warn AE, Brew JS. 1980. Mass balance. Water Res 14:14271434.
White KL, Haggard BE, Matlock MD, Kim JW. 2005. Periphytic chlorophyll-a
response to triclosan exposure application of a passive diffusion periphytometer. Applied Engineering in Agriculture 21:307311.
Wilson BA, Smith VH, Denoyelles F, Larive CK. 2003. Effects of three
pharmaceutical and personal care products on natural freshwater algal
assemblages. Environ Sci Technol 37:17131719.