bioassay selection, experimental design and quality control/assurance for use in effluent assessment...
TRANSCRIPT
Bioassay Selection, Experimental Design and Quality Control/Assurance
for use in Effluent Assessment and Control
IAN JOHNSON,1,* MATT HUTCHINGS,2 RACHEL BENSTEAD,3 JOHN THAIN4
AND PAUL WHITEHOUSE1
1WRc-NSF, Henley Road, Medmenham, Marlow, Buckinghamshire, SL7 2HD, UK2AstraZeneca, Brixham Environmental Laboratory, Freshwater Quarry, Brixham, Devon, TQ5 8BA, UK3Environment Agency, 4 The Meadows, Waterberry Drive, Waterlooville, Hampshire, PO7 7XX, UK
4Centre for Environment, Fisheries and Aquaculture Science, Fisheries Laboratory,Remembrance Avenue, Burnham-on-Crouch, Essex, CM0 8HA, UK
Abstract. In the UK Direct Toxicity Assessment Programme, carried out in 1998-2000, a series ofinternationally recognised short-term toxicity test methods for algae, invertebrates and fishes, and rapidmethods (ECLOX and Microtox) were used extensively. Abbreviated versions of conventional tests (algalgrowth inhibition tests, Daphnia magna immobilisation test and the oyster embryo-larval developmenttest) were valuable for toxicity screening of effluent discharges and the identification of causes andsources of toxicity. Rapid methods based on chemiluminescence and bioluminescence were not generallyuseful in this programme, but may have a role where the rapid test has been shown to be an acceptablesurrogate for a standardised test method. A range of quality assurance and control measures wereidentified. Requirements for quality control/assurance are most stringent when deriving data for char-acterising the toxic hazards of effluents and monitoring compliance against a toxicity reduction target.Lower quality control/assurance requirements can be applied to discharge screening and the identificationof causes and sources of toxicity.
Keywords: direct toxicity assessment, whole effluent test, rapid bioassays
Introduction
Effluent discharges which are to be assessed bydirect toxicity assessment (DTA) contain a rangeof substances with varying physico-chemicalproperties which exert toxic effects through a rangeof mechanisms. Mechanisms of action could in-clude non-polar narcosis, polar narcosis,uncoupling of oxidative phosphorylation, choli-nesterase inhibition, membrane irritation, centralnervous system convulsion, respiratory blocking,
cell division inhibition and photosynthesis inhibi-tion.
It is widely recognised that no single bioassay canbe used to assess the toxic effects of every differentmode of action because not all relevant target sitesare found in a single organism. Therefore, whenusing toxicity tests for assessing and controllingeffluent discharges, a battery of test methods isprobably required. Assessment of potential impactsin the receiving water to which an effluent is dis-charged requires inclusion of species representingdifferent trophic levels, typically algae (primaryproducers), invertebrates (primary consumers) andfishes (secondary consumers). This assessment ofthe potential effects of effluents on different trophic
*To whom correspondence should be addressed:
Tel: +01491-636585; Fax: +01491-636501;
E-mail: [email protected]
Ecotoxicology, 13, 437–447, 2004
� 2004 Kluwer Academic Publishers. Manufactured in The Netherlands.
levels was agreed at the outset of the DTA De-monstration Programme (Environment Agency,1999) and is consistent with the conventional ap-proach to hazard assessment used in the manage-ment and regulation of new and existing chemicals.The use of different taxonomic groups withinaquatic ecosystems has been recognised by reg-ulators in Europe (Umweltbundesamt, 1997) aswellas Canada (Environment Canada, 1999) and theUnited States (USEPA, 1991).
However, it was also recognised that the waythat bioassays are used (in terms of the test designand the quality assurance/quality control applied)needs to be tailored for different applications, thusensuring that the data generated are ‘fit for pur-pose.’ These different applications (with differentobjectives) are shown in detail in Fig. 1 ofWhitehouse et al. (2003) and relate to:� screening discharges for effluent toxicity;� characterising the toxic hazards of effluents as
part of a risk assessment process;� assessing the toxic impact of point source dis-
charges on the receiving water environment;� identifying the cause(s) and source(s) of final
effluent toxicity (using toxicity identificationevaluations and toxicity source evaluations); and
� monitoring compliance against a toxicityreduction target.
In the River Esk (UKWIR, 2001a) and LowerTees Estuary (UKWIR, 2001b, c) case studiesdifferent bioassays were evaluated for differentapplications and this paper describes our findingswith respect to:
(a) the suitability of different test species in dif-ferent aspects of the DTA framework;
(b) prospects for developing abbreviated, highthroughput tests for screening purposes;
(c) differences in the design and interpretation ofscreening and full characterisation tests; and
(d) the different QA andf QC demands of testsused for different operational roles within aDTA framework.
Evaluation of bioassays and rapid methods in the
River Esk and Lower Tees Estuary case studies
Table 1 summarises the test methods used in thecase studies at the freshwater River Esk locationand the estuarine/marine Lower Tees Estuarylocation. These methods are widely used in reg-ulatory chemical hazard assessment schemes andall have internationally recognised and standar-dised protocols which have been promulgated bybodies such as the Organisation for EconomicCo-operation and Development (OECD), Inter-national Standards Organisation (ISO) and theInternational Council for the Exploration of theSeas (ICES). The OECD procedures relate tothe evaluation of pure substances (and formula-tions) while the ISO and ICES procedures can beapplied to environmental samples as well as puresubstances.
In both the River Esk and Lower Tees Estuarycase studies effluent screening also involved the useof rapid methods (the bacterial (Vibrio fischeri)bioluminescence test and the enhanced chemilu-minescence (ECLOX) test). These rapid tests wereused to establish whether they represented a cost-effective alternative to standard methods for cer-tain operational roles, such as screening.
The methods used in the DTA DemonstrationProgramme were selected by the DTA MethodsWorking Group which also established a series of
Table 1. Summary of the short-term toxicity tests used in the River Esk (freshwater) and Lower Tees Estuary (estuarine/marine) case
studies
Test methods (Standard procedures)
Freshwater Estuarine/marine
72 h Algal (Pseudokirchneriella subcapitata)
growth inhibition (OECD, 201)
72 h Algal (Skeletonema costatum) growth inhibition (ISO 10253)
48 h Daphnia magna immobilisation (OECD, 202) 24 h Oyster (Crassostrea gigas) embryo-larval development (ICES, Vol.11)
48 h Juvenile marine copepod lethality (ISO 14669)
96 h Juvenile rainbow trout (Onchorhynchus mykiss)
lethality (OECD, 203)
96 h Juvenile turbot (Scophthalmus maximus)
lethality (ISO/WD 15990)
438 Johnson et al.
test guidelines used in the programme (Environ-ment Agency, 1998).
Three basic test designs were adopted for thedifferent applications investigated in the case stu-dies (Table 2):1. A limited concentration, high-throughput
design for screening effluent discharges andtoxicity identification and source evaluationexercises. These high throughput methods areintended primarily to reduce costs relative to thefull concentration range tests principally by� reducing the timescales within which data
can be generated;� reducing resources devoted to testing (num-
bers of test organisms, materials and espe-cially effort) and
� increasing the number of samples which canbe tested for a unit cost relative to the stan-dard method.
2. A full (or modified full) concentration range testusing the standardised method for effluentcharacterisation and monitoring against atoxicity reduction target.
3. A single concentration bioassay using the stan-dardised method for assessing the impact ofdischarges in the receiving water.
Table 3 summarises the nature of the testing car-ried out in the River Esk and Lower Tees Estuarycase studies for different applications of the DTAapproach. The strategy for the River Esk casestudy has been described in detail in UKWIR(2000a) and comprised screening (on-site and at anecotoxicology laboratory) and characterisation ofthe Langholm sewage treatment works (STW)final effluent. This testing was accompanied byreceiving water monitoring at eight locationsupstream and downstream of the STW discharge.The strategy for the evaluation of effluent dis-
charges in the Lower Tees Estuary case study isdescribed in detail in UKWIR (2001b). In sum-mary this comprised:� Screening of 12 effluent discharges (A to L)
on-site and at an ecotoxicology laboratory ontwo occasions.
� Subsequent screening of 8 priority discharges (Ato C, G, H, I, K and L) on-site and at an eco-toxicology laboratory on a further one occa-sion.
� Subsequent screening of 4 discharges (A, C, Gand L), with receiving water monitoring aroundthe discharges.In both case studies work was subsequently
carried out to identify the causes and/or sources ofeffluent toxicity and details of this are provided inthe accompanying paper by Hutchings et al.(2003).
Results of the River Esk and Lower Tees Estuary
case studies
River Esk case study
In the River Esk case study, initial screening onthree occasions with the D. magna immobilisa-tion test and the rapid methods revealed fluctu-ating toxicity in the Langholm STW final effluentwhen the traders discharging to the works wereoperating (Table 4). In contrast, the lowesttoxicities (highest EC50 values) in all bioassayswere found when the traders were not dischar-ging to the STW during a holiday period (August1998). The D. magna immobilisation bioassayshowed the greatest toxicity responses in theSeptember 1998 samples, but no toxicity on twoother occasions (July and August 1998). The
Table 2. Applications of short-term methods within the DTA approach to control acutely toxic and complex effluents
Sample type Application Test design
Effluent Screening Limited concentration range, reduced cost high
throughput method (where available)
Characterisation Full concentration range, standardised method
Toxicity identification and source evaluation exercises Limited concentration range, reduced cost high
throughput method (where available)
Monitoring against a toxicity reduction target Modified full concentration range, standardised method
Receiving water Assessment of the impact of point discharges Single concentration bioassay method
Bioassay Selection for Effluents 439
bioluminescence bioassay measured no, or limitedtoxicity in the effluent on all of the three sam-pling occasions, although this may reflect limitedsensitivity to key pollutants such as diazinon andpropetamphos. The ECLOX test measured‘toxicity’ on all three sampling occasions but itwas evident that the responses observed couldalso be partially or completely due to the bio-chemical oxygen demand (BOD) of the samplesrather than toxicity.
Characterisation of the Langholm STW finaleffluent revealed toxicity in the 48-h D. magnaimmobilisation test but no, or limited, toxicity inalgal (Pseudokirchneriella subcapitata) growthinhibition, or rainbow trout (Oncorhynchus
mykiss) lethality tests (Table 5). On three testoccasions the NOEC values in the D. magna testswere 9–18% effluent, whereas the NOEC values forthe algal growth and fish lethality tests were12.4—>98.4% and >50% effluent, respectively.
There was only limited evidence of short-termlethal ecotoxicity in the River Esk. Only onereceiving water sample from downstream ofLangholm STW (in May 1999) showed toxicity inthe D. magna immobilisation test during a series offive monitoring exercises (July, August andSeptember 1998 and May and June 1999). How-ever, river flow during sampling events was higherthan usual due to unusually high rainfall (up to193% of the long term average on a monthly
Table 4. Summary of the screening data for the Langholm STW final effluent
Sampling date
Discharges to
Langholm STW
Range of EC50 values
(% effluent) in the
ECLOX test
Range of EC50 values
(% effluent) in the
Microtox test
(% effect at 91%)
Range of EC50 values
(% effluent) in
the Daphnia magna
immobilisation test
22–23 July 98 Traders working 1.29–2.67% v/v 88–91% v/v (15–63%) >100% v/v
10–11 August 98 Trader holiday 8.47–>20% v/v >91% v/v (0–25%) >100% v/v
22–23 September 98 Traders working 1.18–3.4% v/v >91% v/v (6–42%) <10–60.3% v/v
Table 3. Summary of the nature of the testing carried out in the River Esk and Lower Tees Estuary case studies for different
applications of the DTA approach
Application River Esk case study Lower Tees Estuary case study
Effluent screening One discharge on three occasions using 24 h
Daphnia magna immobilisation tests
(4 concentrations) and rapid methodsa
Twelve discharges (A to L) on two occasions and 8
priority discharges (A to C, G to L) (on one further
occasion) using the 24 h oyster embryo-larval
development test (4–5 concentrations) and
rapid methodsa
Effluent One discharge on three occasions using: One priority discharge on five occasions using:
characterisation � 72 h algal (Pseudokirchneriella subcapitata)
growth inhibition test (7 concentrations)
� 72 h algal (Skeletonema costatum) growth inhibition
test (5 concentrations)
� 48 h Daphnia magna immobilisation test
(10 concentrations)
� 24 h oyster embryo larval development test
(5 concentrations)
� 96 h rainbow trout (Oncorhynchus mykiss) � 48 h Tisbe battagliai lethality test (5 concentrations)
lethality test (6 concentrations) � 24h juvenile turbot (Scophthalmus maximus)
lethality test (5 concentrations)
Monitoring of
receiving waters
Eight locations upstream/downstream of
discharge on five occasions using 24 h Daphnia
magna immobilisation tests and rapid methodsa
Seven locations around two impacting (C and L) and
two non-impacting (A and G) discharges (and
discharges themselves) on one occasion using the oyster
embryo larval development test and rapid methodsa
Toxicity reduction
evaluation exercises
Sewer network on two occasions using the
Daphnia magna immobilisation test
Sewer network on three occasions using the algal
(Skeletonema costatum) growth inhibition test and
the oyster embryo larval development test
a The rapid methods were the enhanced chemiluminescence (ECLOX) test and the bacterial (Vibrio fischeri) bioluminescence test.
440 Johnson et al.
basis), giving rise to a high level of dilution of anycontaminants (Table 6).
An investigation of the cause(s) and source(s) ofLangholm STW final effluent toxicity was alsocarried out using the most sensitive species(D. magna). This study identified certain organo-phosphate pesticides (diazinon and to a lesserdegree propetamphos) as key toxic constituents,although other toxic constituents could not beidentified (see accompanying paper by Hutchingset al., 2003). Toxicity tracking in the sewer net-work with the Daphnia acute immobilisation testalso proved useful (alongside chemical analysis) inidentifying the traders implicated in the toxicity ofthe Langholm STW final effluent.
Lower Tees Estuary case study
Of the 12 discharges (A to L) screened in the LowerTees Estuary case study on two occasions with theoyster embryo-larval development test and therapid methods, 11 demonstrated short-term toxi-city. The tests exhibited to varying degrees thecapacity to discriminate between the 11 toxic dis-charges. The OEL and chemiluminescence testswere more sensitive than the bioluminescence test.
However, as in the River Esk case study, the che-miluminescence test may have responded to BODrather than toxicity. Subsequently, eight ‘priority’discharges (A to C, G, H, I, K and L), which wereperceived to have the greatest potential impact onthe receiving water based on toxicity and flow datawere tested using the OEL test. The results con-firmed those of the initial screening exercise interms of the levels of effluent toxicity measured(Table 7).
Two ‘priority’ discharges (C and L) with sig-nificant predicted receiving water impact (i.e. evenafter dilution, acute toxicity was predicted in thereceiving water) and two discharges of limited per-ceived impact (A and G) were then investigated indetail. Short-term ecotoxicity could be measured inreceiving water samples collected from within thearea of the estuary where the discharges werelocated, using the OEL test. Discharge L was con-sidered to be the highest priority discharge, havingthe largest potential impact on the receiving water(see accompanying paper by Girling et al., 2004).
Characterisation of discharge L showed that the72-h algal (Skeletonema costatum) growth inhibi-tion test and the 24-h OEL test were more sensitiveto the effluent than the 48-h Tisbe battagliai leth-
Table 5. Summary of the data from the characterisation of the Langholm STW final effluent
72 h P. subcapitata growth
inhibition test data (% effluent)
48 h D. magna immobilisation test
data (% effluent)
96 h O. mykiss lethality test
data (% effluent)
Sampling date NOEC EC50 NOEC
EC50 (95% Confidence
intervals) NOEC EC50
9 March 99 12.4 >49.8 18 24 (18–32) >56 >56
12 March 99 >98.2 >98.2 18 22 (20–24.1) 100 >100
8 July 99 24.9 >98.2 9 12.3 (11.7–13) >50 >50
Table 6. Responses in the Daphnia magna immobilisation test and the rapid methods observed with receiving water samples from the
River Esk case study
Sampling date
River flow
conditions
Changes in light output
(EC50 values) in the
ECLOX test at receiving
water locations
Changes in light output in
the Microtox test, relative to
controls at receiving water
locations
Changes in response in
the Daphnia magna
immobilisation test at
receiving water locations
22–23 July 98 Moderate/high �12.5% �19.8% 0–15%
10–11 August 98 Moderate �25% �16.7% Not tested
22–23 September 98 Low �55.6% �29.9% 0% at all locations
5 May 99 Low/moderate Not tested Not tested 9.2–60%
13 July 99 Low Not tested Not tested 0% at all locations
Bioassay Selection for Effluents 441
ality test and the 24-h juvenile turbot (Scophthal-mus maximus) lethality test (Table 8). Both meth-ods were subsequently used for toxicity trackingand toxicity identification evaluation (TIE) ex-ercises (see accompanying paper by Hutchings etal. 2004).
In the Lower Tees Estuary case study it wasfound that the scale of toxicity tracking in acomplex sewer network (that is the number ofsamples generated) rather than the capabilities ofthe test methods adopted may limit data acquisi-tion. The use of rapid, small-scale, bioassays washighly useful in allowing cost-effective toxicitytracking exercises to be undertaken using standardtest organisms, although it did not prove possibleto identify the source of the toxicity in discharge L.
Once a sensitive test (or tests) has been identi-fied in a toxicity reduction evaluation it may be
appropriate to modify the conventional method torender it more amenable for use in a toxicitytracking exercise in which concurrent testing oflarge numbers of samples is possible. However,experimenters will need to guard against makingchanges to methodology that compromise thesensitivity of the test.
On-site testing in the case study also provedsuccessful in maximising sample throughput. Theshorter lines of communication afforded byon-site testing also helped when it was foundnecessary to adapt testing programmes. On-sitetesting facilities should be used where possible,but it is important to ensure that such facilitiesare fit for purpose particularly with regard to themaintenance of healthy test organisms, and theability to maintain suitable environmental condi-tions during testing.
Table 7. Summary of the screening data from the Lower Tees Estuary case study
Discharge Type of discharge
Range of EC50 values
(% effluent) in
the ECLOX test
Range of EC50 values
(% effluent) in
the Microtox test
Range of EC50 values
(% effluent) in the oyster
embryo-larval development test
A Chemical production plant 2.3–5.3 9.7–>100 22–>100
B Chemical production plant 0.7–2.8 1.2–13 0.31–2.4
C Chemical production plant 0.24–0.67 6.6–30 0.24–1.0
D Chemical storage site 2.2–4.7 21–38 0.53–65
E Chemical production plant 2.7–7.1 74–>100 0.47–7.3
F Chemical and pharmaceutical
production plant
0.36–1.3 >100 5.0–44
G Chemical production plant 2.0–3.6 4.1–7.6 1.1–50
H Petroleum refinery 0.55–3.0 1.5–53 2.0–9.7
I Inorganic chemical processing
plant
0.83–3.1 37–>100 3.4–>100
J Cooling water and treated sewage 34–>100 >90 >100
K Coking plant cooling and
wash water
0.22–1.5 22–>100 0.55–5.2
L Chemical production plant 0.98–4.8 25–>100 3.4–27
All discharges were screened on three occasions except discharges D, E, F and J which were screened on two occasions.
Table 8. Summary of the data from the characterisation of Discharge L in the Lower Tees Estuary case study
Sampling date
72 h S.costatum growth
inhibition test data
(% effluent)
24 h Oyster embryo-larval
development test data
(% effluent)
48 h Tisbe battagliai
lethality test data
(% effluent)
96 h S. maximus lethality
test data (% effluent)
17/6/99 4.1 3.5 Not tested 18
2/7/99 9.0 9.2 43 >32
13/7/99 4.2 10 29 >32
14/7/99 4.5 6.9 57 13
17/7/99 8.3 4.8 47 >32
442 Johnson et al.
Measuring toxic effects of effluents in a DTA
approach
Types of methods (including test designs and testendpoints) for different applications
The type of method(s) used to test environmentalsamples (such as effluents and receiving waters) andthe experimental design adopted (e.g., number ofexposure concentrations, interval between testconcentrations and test duration)will dependon theobjectives of the study. Table 9 summarises theobjectives of different applications within a DTAframework and, based on the data from the casestudies, the resulting test designs in terms of the levelof replication, test concentrations and test end-points.
Effluent screening and identification of the cause(s)and source(s) of toxicity
Screening of effluents or investigating the cause(s)and source(s) of effluent toxicity (using toxicityidentification and source evaluation exercises) canrequire testing of large numbers of samples inwhich the objective is to provide information onthe rank order of toxicity. This then allowsdecisions to be made as to where subsequent test-ing effort is directed. Experience from the RiverEsk and Lower Tees Estuary case studies hashighlighted the value of abbreviated versions ofconventional tests (algal growth inhibition tests, D.magna immobilisation test and the oyster embryo-larval development test). Procedures have nowbeen developed for conducting high throughput
Table 9. Summary of the objectives and test designs required for different applications within a DTA framework
Test design
Application Objective Replication Test concentrations Test endpoint
Effluent screening To provide data on the
toxicity of a discharge and
enable multiple discharges to
a water body to be ranked,
thereby establishing
priorities for further action
Unreplicated, if it has
been established that
there is an adequate
degree of within-test
precision
At least 5 concentrations Toxicity thresholda as
a minimum but also
EC10, EC20 or NOEC
if deemed appropriate
Effluent
characterisation
To provide detailed and
robust data on the toxicity of
a discharge to allow
an assessment of the acute
risk posed to a receiving water
Replicated which permits
an assessment of test
precision through the
calculation of confidence
intervals
A concentration series
should be selected to
allow a value to be
obtained for the selected
endpoint and should be
based on screening test
data
EC10, EC20, EC50,
NOEC or MATC
Effluent toxicity
identification and
source evaluations
To obtain an understanding
of the causes and sources of
toxicity of an effluent
discharge
Unreplicated, if it has
been established that
there is an adequate
degree of within-test
precision
A concentration series
should be selected to
allow a value to be
obtained for the selected
endpoint
Toxicity thresholda as
a minimum but also
EC10, EC20 or NOEC
if deemed appropriate
Monitoring of
effluent toxicity
(against a toxicity
reduction target)
To determine whether
the discharge is achieving
the toxicity reduction target
Replicated A concentration series
should be selected to
allow a value to be
obtained for the selected
endpoint and should be
based on effluent
characterisation data
Toxicity reduction
targetb
Receiving water
monitoring
To assess the magnitude of
effects measured at receiving
water sites (including the
point of protection), relative
to those at control locations
Replicated Undiluted receiving water
samples from locations
chosen on a site-specific
basis
Level of response in
sample, relative to an
appropriate control
a Effluent concentration that results in no adverse acute effect.b Effluent concentration at which there is no acute toxicity as predicted at the point of protection.
Bioassay Selection for Effluents 443
and minituarised algal growth tests, D. magnaimmobilisation tests, OEL tests and T. battagliailethality tests (Environment Agency, 2001). In theminiaturised tests, potential space constraints havebeen circumvented so that exposure of the organ-isms to effluents takes place in microtitre wellplates placed within temperature-controlled in-cubators. This avoids the need for large, non-mo-bile temperature controlled facilities and so enablestesting to be carried out on-site and even at remotelocations. Throughput with these systems may beup to 40 samples per day with unit costs in theregion of �50 per sample. For discharges tofreshwater, it is recommended that abbreviatedalgal growth inhibition tests and D. magna im-mobilisation tests are used to screen discharges,with the most sensitive of these being used to in-vestigate the cause(s) and source(s) of toxicity. Fordischarges to the estuarine/marine environment,algal growth inhibition tests and OEL or T. bat-tagliai lethality tests should be used. Evidence fromthe case studies indicated that fish acute toxicitydata are usually of lower sensitivity than thosefrom algal and invertebrate toxicity tests and,therefore, the use of fish at the effluent screeningstage is unlikely to serve any useful purpose.However, if there is a reason to suspect the pre-sence of toxicants to which fish are particularlysusceptible (e.g. ammonia, cyanide) then a fishacute toxicity test using a reduced number of ani-mals per test concentration may be warranted.
For these applications, where only anapproximate rank order of effluents may be nee-ded, a high throughput test using a limited con-centration series is recommended, and a simplevisual determination of a toxicity threshold (thehighest effluent concentration at which no adverseeffects are evident) is usually adequate. It is im-portant to emphasise that it will not be possible toestimate a reliable LOEC or NOEC from suchtests. The visual assessment of the highest con-centration causing no measurable adverse acuteeffects is referred to as the ‘toxicity threshold’ and isexpressed in terms of % (v/v) effluent.
Effluent characterisation
The acute toxicity of effluents is best characterisedusing a battery of test methods representing dif-
ferent trophic levels. Unlike the tests used foreffluent screening or the identification of the cau-se(s) and source(s) of toxicity, replicated testdesigns should be used that permit an assessmentof precision on point estimates of toxicity (e.g.EC50) through the calculation of confidence limits(where data and endpoints permit).
There are advantages and limitations of pointestimates or hypothesis testing approaches andthese have been discussed in detail in UKWIR(2000d). In the effluent characterisation stage, fullconcentration series tests are conducted and any ofEC10/EC20/EC50, NOECacute or MATCacute valuescan be used as the test endpoint as long as thefollowing requirements are met:
1. For hypothesis testing (estimation of LOECacute
and NOECacute)
� An estimate of the power of the tests (that isthe least significant difference that can beresolved between control and treatmentgroups) is reported.
� The most powerful statistical significance testis used, consistent with the data.
� The interval between test concentrations isnot excessive (it does not exceed 2.2-fold andis reported).
� Adequate control over random error isdemonstrated by QA/QC procedures (this isdiscussed further below).
� Reporting of the Maximum AcceptableThreshold Concentration (MATCacute)—thegeometric mean of the LOECacute andNOECacute—is preferable to the NOECacute
alone because it is less sensitive to the dis-position of test concentrations.
2. For regression-based methods (to estimate anECx value).
� Measures of error at the defined ECx valueshould be reported.
� The value of x should be consistent within astudy; EC20 estimates are preferable to esti-mates of EC10.
� Goodness of fit with the regression modelused should be reported.
� It may not be possible to calculate EC50
estimates for discharges which exhibit onlylow acute toxicity, in which case other options
444 Johnson et al.
(EC10/EC20, NOECacute or MATCacute) arepreferred.
Monitoring of effluent toxicity against a toxicityreduction target
Effluent toxicity monitoring should be performedusing the species specified in the toxicity reductiontarget (that is the ultimate level of effluent toxicitywhich is agreed will be acceptable to satisfy theobjectives of the toxicity reduction study), whichwill invariably be the most sensitive of those usedduring toxicity characterisation.
When monitoring effluent toxicity, the limit testdesign (based on assessing effects at a singleexposure concentration) should be ‘embedded’ inthe concentration-response design. In other words,one of the test concentrations should correspondto that predicted to occur at the ’point of protec-tion’ (the point in the receiving water beyond amixing zone, at which zero effects are expected andrequired). This permits a simple ‘pass/fail’ assess-ment against the toxicity reduction target whilstretaining the ability to determine the extent of any‘non-compliance’. The DTA Technical Guidance(UKWIR, 2001d) recommends that within-testvariability should be constrained to the extent thatit is possible to discriminate a difference of at least25% in response between control and test groupsat this concentration. If this is not routinely pos-sible, it will be necessary to increase the level ofreplication at this particular test concentration toensure that the test is capable of discriminatingbetween a ‘pass’ and ‘fail’ response with this levelof discriminatory power.
To ensure that there have been no significantchanges in effluent toxicity, occasional testing withthe full trophic level battery is recommended. Thismay need to be performed annually unless plannedchanges in processes or treatment take place, whentesting with the full suite of trophic level testswould be warranted.
Quality assurance/quality control requirements
A consistent feature of the testing requirementsfor using a DTA approach is that emphasis isplaced on the generation of reliable and robustdata, especially when important regulatory or
commercial decisions may result (Whitehouseet al., 1996). In order to support this approach theregulatory ecotoxicology testing quality scheme(RETQS), trialled in the Demonstration Pro-gramme, should be implemented. This requireslaboratories carrying out regulatory effluent toxi-city testing to adopt approved test methods, topractice certain quality assurance procedures andto take part in external proficiency testing basedon reference toxicants as a means of ensuring adegree of quality control in the performance oftoxicity tests (Whitehouse et al., 1996). However,as is explained below, the rigour with which theserequirements are applied depends on the intendeduse of the data generated and, in particular, on thesignificance of decisions made on the basis of thetest data generated.
Effluent screening and identification of cause(s)and source(s) of toxicity
Failure to generate sound screening data couldlead to false conclusions about a discharge’s pos-sible impact. This could, in turn, lead to the pre-mature exclusion of discharges which actuallypose a risk to receiving water quality (false nega-tive). When attempting to identify cause(s) andsource(s) of toxicity it is important to bear in mindthat poor quality data could be misleading andcould result in erroneous conclusions and wastedinvestment.
Effluent screening and campaigns aimed atidentifying cause(s) and source(s) of toxicity maybe performed on-site (and at remote locations). Inthis case formal compliance with QA systems (suchas GLP or UKAS accreditation) is rarely possible.However, in such cases the following requirementsare suggested:� Personnel conducting screening tests should be
acquainted with the issues involved in qualityassurance and quality control and this is bestachieved by screening being carried out only byRETQS-approved laboratories.
� Reference toxicant tests should be run atintervals during a programme of effluentscreening and monitored against internal QCcharts to check that bias or variability is notexcessive compared to that seen normally.Submission of reference toxicant data as part
Bioassay Selection for Effluents 445
of the external RETQS requirements isunnecessary.
Effluent characterisation
Important regulatory decisions, possibly with sig-nificant commercial implications, may be made onthe basis of the data generated in the effluentcharacterisation stage. It is therefore importantthat the data are reliable and free from excessivebias and variability. With this in mind, all effluentcharacterisation testing should be subject to therequirements of the RETQS scheme.
Essentially this involves addressing issues ofdata integrity and auditability by requiringlaboratories to:� Comply with either the Principles of Good
Laboratory Practice (GLP) or to be accreditedunder the United Kingdom AccreditationScheme (UKAS) for the specified methods;
� Participate in an external quality controlscheme which involves providing referencetoxicant data to an external body so that pre-cision and accuracy can be assessed against pre-defined targets.
� Perform tests according to approved testguidelines.
Monitoring of effluent toxicity
If compliance assessment indicates excessiveeffluent toxicity, this could lead to importantregulatory and commercial consequences. It isimportant, therefore, that compliance monitoringdata are as free from excessive bias and variabilityas possible because errors of this type may resultin inaccurate conclusions. Thus, the same con-cerns apply at this stage as were applied whentoxicity data were generated at the characterisa-tion stage. It follows that a high level of qualityassurance and quality control is appropriate fortoxicity monitoring. Full compliance with therequirements of the RETQS scheme is, therefore,recommended.
Receiving water monitoring
The same quality assurance/quality control con-ditions as described for effluent screening shouldbe applied to receiving water monitoring.
Conclusions
Experiences gained from the River Esk and LowerTees Estuary case studies were that, whilst therewas a good deal of consistency in the test speciesused in different operational roles, the details of theconduct and interpretation of the tests may differconsiderably. Similarly, the level of QA andQC canbe varied according to the significance of decisionsmade on the basis of the data generated.For screening discharges for toxicity and seekingto identify the causes and sources of effluenttoxicity, abbreviated versions of conventional tests(algal growth inhibition tests, D. magna immo-bilisation test and the OEL test) were valuable.The data from the case studies indicated that rapidchemiluminescence and bioluminescence methodswere not generally applicable, but may have a roleif the results from the rapid method are correlatedwith the standard method and can be readilyinterpreted in terms of the potential environmentalimplications.
In order to standardise the approaches to betaken for testing effluents and receiving waterswithin a DTA framework, a series of guidelineshave been produced which specify test methodol-ogies in some detail (Environment Agency2002a,b). These documents should be used along-side the Technical Guidance for the Implementa-tion of DTA for effluent control (UKWIR 2001d).
Attention also needs to be given to measuresdesigned to promote data quality and consistencyin testing, through formal QA/QC procedures. Thequality control/assurance requirements need to bemost stringent when deriving data for characteris-ing the toxic hazards of effluents and monitoringcompliance against a toxicity reduction target. Lessstringent quality control/assurance requirementscan be applied to discharge screening and theidentification of causes and sources of toxicity.
Acknowledgements
The authors thank the following organisations,without whose help these investigations could nothave been undertaken: Northumbrian Water,North East Region of the Environment Agency,SEPA and West of Scotland Water. In addition tothis we would also like to thank the various traders
446 Johnson et al.
and industrialists who co-operated with ourinvestigations.
References
Environment Agency (1998). Draft Short-term Ecotox-
icological Methods Guidelines for Effluent and Receiving
Water Assessment. Environment Agency, Bristol, UK.
Environment Agency (1999). A Proposed Scheme to Ensure the
Quality of Data Generated by Laboratories Undertaking
Regulatory Ecotoxicological Testing. R&D Technical
Report P166, Environment Agency, Bristol, UK.
Environment Agency (2002a). Ecotoxicity Test Methods for
Effluent and Receiving Water Assessment—Comprehensive
Guidance. Environment Agency, Bristol, UK.
Environment Agency (2002b). Ecotoxicity Test Methods for
Effluent and Receiving Water Assessment—Supplementary
Advice to International Test Guidelines. Environment
Agency, Bristol, UK.
Environment Canada (1999). Guidance Document on Appli-
cation and Interpretation of Single-Species Tests in En-
vironmental Toxicology. Report EPS1/RM/34. Method
Development and Application Section. Environmental
Toxicology Section, Environment Canada.
Girling, A.E., Riddle, A.M., Mitchell, G.M., Chown. P.,
Tinsley, D., Buckler, C., Johnson, I. and Benstead, R.
(2004). Estimating Spatial Patterns of Effluent Exposure
Concentrations in Direct Toxicity Assessment Studies,
Ecotoxicology 13, 449–61.
Hutchings, M., Johnson, I., Hayes, E., Girling, A., Thain, J.,
Thomas, K., Benstead, R., Whale, G., Wordon, J.,
Maddox, R. and Chown. P. (2004). Toxicity Reduction
Evaluation, Toxicity Identification Evaluation and Toxicity
Tracking in Direct Toxicity Assessment. Ecotoxicology 13,
475–84.
UKWIR (2001a). UK Direct Toxicity Assessment (DTA)
Demonstration Programme—River Esk Project. Technical
Report 00/TX/02/02. United Kingdom Water Industry
Research, London, UK.
UKWIR (2001b). UK Direct Toxicity Assessment (DTA)
Demonstration Programme—Lower Tees Estuary Project
Part I. Technical Report 00/TX/02/03. United Kingdom
Water Industry Research, London, UK.
UKWIR (2001c). UK Direct Toxicity Assessment (DTA)
Demonstration Programme—Lower Tees Estuary Project
Part II. Technical Report 00/TX/02/04. United Kingdom
Water Industry Research, London, UK.
UKWIR (2001d). TechnicalGuidance for the Implementation of
Direct Toxicity Assessment (DTA) for Effluent Control:
Addressing Water Quality Problems in Catchments where
Acute Toxicity is an Issue. Technical Report 00/TX/02/07.
United KingdomWater Industry Research, London, UK.
Umweltbundesamt (1997). Report on The OSPAR Workshop
on Ecotoxicological Evaluation of Waste Water. Berlin,
September 23–24, 1997.
US EPA (1991). Technical Support Document for Water
Quality-based Toxics Control. Report EPA/505/2-90-001.
Office of Water, United States Environmental Protection
Agency.
Whitehouse, P., van Dijk, P.A.H., Delaney, P.J., Roddie, B.D.,
Redshaw, C.J. andTurner, C. (1996) The precision of aquatic
toxicity tests: its implications for the control of effluents by
Direct Toxicity Assessment. In J.F. Tapp, S.M. Hunt and
J.R. Wharfe (eds). Toxic Impacts of Wastes on the Aquatic
Environment, pp. 44–53, Royal Society of Chemistry.
Bioassay Selection for Effluents 447