The OECD Lemna growth inhibition test
f
Development and ring-testing of draft OECD test Guideline
Research and Development
Technical Report
EMA003
ENVIRONMENT AGENCY
-------
All puips used in production of thts paper is sourced from sustainable managed forests and are elemental
chlorine free and wood free
-------
WRc-NSF Ltd
• Our Ret:
5 October 1999
Dr Jerry Smrchek
US EPA
Existing Chemicals Division
Risk Assessment Division (7403)
Office of Pollution Prevention and Toxics
U S EPA Headquarters,
401 M St, SW
Washington,
DC 20460
Dear Jerry
Lemna ring-test report
I hope you are well and that you received the copy of the Leipzig SETAC poster I sent
you back in May. Please find enclosed your copy of the Lemna ring-test report which I
trust you will find of interest. As you can see, the ring-test went well with a good number
of participants. It was unfortunate that there was not more data on L. gibba submitted,
but this uneven spread is a risk you take when you allow participants to choose what
they do. As participation in the ring-test was voluntary it was not possible to instruct
each participant which option to use.
As a result of this ring-test a few minor modifications were made to the draft Guideline,
but these are not significant and it remains virtually the same as ring-tested. We have
enjoyed collaborating on this project with yourself and the EA and feel that the outcome
has been worth the hard work on behalf of the Steering Group and all those who
participated. The next stage will be submission of the draft for adoption by the OECD,
but this stage of the process is in Marie’s hands.
Best wishes for now, and hope to see you at SETAC Brighton next year,
LL .
Ian Sims
WRc.NSF LId, Henley Road, Medmenham. Marlow, Bucks SL7 2HD, UK WRc-NSF Ltd
Tel 4 44 (0)1491 636500 Fax +44 (0)1491 636501 a lont venture of
WRc NSF Ltd Registered in England No 3754780 WRc plc and
Registered Office Frankland Road Blagrove Swindon Wiltshire SN5 8YF England NSF International
-------
The OECD Lemna growth inhibition test
R&D Technical Report EMA 003
I Sims, P Whitehouse, R Lacey
Research Contractors:
WRc plc
Environment Agency
USEPA Office of Prevention, Pesticides and Toxic Substances
Further copies of this report are available from:
Environment Agency R&D Dissemination Centre, do
WRc, Frankland Road, Swindon, Wilts SN5 8YF
tel: 01793-865000 fax: 01793-514562 e-mail: publications@wrcplc.co.uk
-------
Publishing Organisation:
Environment Agency
Rio House
Waterside Drive
Aztec West
Almondsbury
Bristol BS32 4UD
Tel: 01454 624400 Fax. 01454 624409
ISBN 1 85705 1 378
© Environment Agency 1999
All rights reserved. No part of this document may be reproduced, stored in a retrieval system. or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or
otherwise without the prior permission of the Environment Agency
The views expressed in this document are not necessarily those of the Environment Agency. Its
officers, servant or agents accept no liability whatsoever for any loss or damage arising from the
interpretation or use of the information, or reliance upon views contained herein.
Dissemination status
Internal: Released to Regions
External: Released to Public Domain
Statement of use
National OECD Co-ordinators decided that high prionty to the development of a Lemna growth
inhibition test. The recommendation originates from the OECD Danish detailed Review paper on
“Aquatic Testing methods for Pesticides and Industrial Chemicals”. This reflects concerns about
relying on tests using unicellular algae for predicting the hazard of substances to higher plants which.
of course, play an important role in aquatic ecosystems. For example, unicellular algae exhibit low
sensitivities to certain phytotoxic compounds because they lack certain sites of action which are
present in higher plants. Furthermore, there are practical difficulties in performing algal toxicity tests
with coloured and turbid samples and samples with low dissolved oxygen. The methods using the
floating macrophyte, Lemna, have found lavour in studies of phytotoxicity because they are easy to
handle and have been shown to be sensitive to a wide range of contaminants, including complex
effluents.
Research contractor
This document was produced under R&D Project EMA 003(Package 7.6)1 by.
WRc plc Office of Prevention, Pesticides and Toxic Substances
Henley Road USEPA Headquarters
Medmenham 401 M St. SW
Marlow Washington D C.
B uckinghamshire USA
SL7 2HD
Tel: 01491 571531
Fax: 0149! 579094
WRc Report No.: EA 4784
R&D Technical Report EMA 003
-------
Environment Agency’s Project Manager
The Environment Agency’s Project Manager for R&D Project EMA 003/ was:
Dr. David Forrow.
USEPA’s Project Manager
The USEPA;s Project Manager for R&D Project EMA 003 was.
Jerry Smrchek.
R&D Technical Report EMA 003
-------
CONTENTS Page
LIST OF TABLES ii
LIST OF FIGURES
EXECUTIVE SUMMARY 1
KEY WORDS 3
1. INTRODUCTION 5
1.1 Background to development of an OECD Guideline for a Lemna growth
inhibition test 5
1.2 Development of draft test Guideline 5
1.3 Scope of the OECD test guideline 7
2. PHASE 1: RESEARCH TO DEVELOP THE DRAFT GUIDELINE 9
2.1 Introduction 9
2.2 Plant material 9
2.3 Growth media 10
2.4 Test vessels and volumes 13
2.5 Test endpoints 13
2.6 Expression of toxicity and data analysis 15
3. PHASE 2: RING-TESTING OF DRAFT OECD TEST GUIDELINE,
DATED JUNE 1998 17
3.1 Objectives of ring-test 17
3.2 Design of ring-test 17
3.3 Participation in the ring-test 18
3.4 Statistical analysis of data 19
3.5 Results and Discussion 20
4. CONCLUSIONS 34
REFERENCES 37
APPENDICES
APPENDIX A NAMES AND AFFILIATIONS OF PARTICIPANTS IN THE
RING TEST OF THE OECD LEMNA GROWTH TEST 41
R&D Technical Report EMA 003 1
-------
APPENDIX C
GUIDANCE TO RING-TEST PARTICIPANTS AND PRO-
FORMA DATASHEETS
EC 50 VALUES ESTIMATED FROM RING-TEST
GUiDANCE ON THE TESTING OF EFFLUENTS,
LEACHATES AND RECEIVING WATERS IN THE LEMNA
GROWTh INHIBITION TEST
LIST OF TABLES
Table 1.1 The Lemna Steenng Group, 1997-1999
Table 1.2 Ring-test participation
Table 2.1 Interlaboratory companson of Lemna growth in different media
Table 3.1 Data sets submitted by participants, broken down by test organism
and by toxicant, and showing the number of repeat tests performed
and included in the statistical analysis of the nng-test results.
Table 3.2 Number of participants reporting EC 50 s for different endpoints with
Lemna minor and potassium dichromate.
Table 3.3 Number of participants reporting EC 50 s for different endpoints with
Lemna minor and DCP.
Table 3.4 Number of participants reporting EC 50 s for different endpoints with
Lemna gibba and potassium dichromate.
Table 3.5 Number of participants reporting EC 50 s for different endpoints with
Lemna gibba and DCP.
Table 3.6 EC 50 values for L. minor calculated using average specific growth-
rate based on frond number (E C 5 os) and final biomass based on dry
weight (E bC5O).
Results of REML analysis
Control limits for accuracy in Lemna growth inhibition tests
LIST OF FIGURES
Box-whisker plots of EC 50 values (as mg 11 and % deviation from
median EC 50 )
Effects of doubling time on sensitivity of Lemna minor to potassium
dichromate
Effects of doubling time on sensitivity of Lemna minor to DCP
APPENDIX D
APPENDIX E
71
83
87
6
7
12
18
23
23
24
24
25
27
31
28
29
Table 3.7
Table 3.8
Figure 3.1
Figure 3.2
Figure 3.3
R&D Technical Report EMA 003
ii
-------
EXECUTIVE SUMMARY
In a review of test guideline requirements in 1995, an aquatic higher plant toxicIty test method
was highlighted as a priority for development by the OECD. This led to an initiative by KEMI
(Sweden) to develop a guideline based on aquatic macrophytes of the genus, Lemna, which
was subsequently taken forward by WRc (UK).
A Steering Group was established and research needs (‘Phase 1’ studies) identified which
would enable a test Guideline to be drafted. These were concerned with the choice of test
species, test media, endpoints and data analysis and were addressed between 1997 and 1998
through an experimental programme by members of the Steering Group. The results and
conclusions drawn from these studies are discussed. The draft method was then subjected to
international ring-testing (‘Phase 2’) in 1998-99.
Significant features of the test guideline are that it permits the use of two Lemna species. L
minor and L. gibba. Essentially, it involves determination of the effects of substances on the
increase in the size of a population of test plants compared to that in control vessels. The main
purpose of the OECD test Guideline is for hazard assessment of chemicals. However, the
Lemna growth inhibition test also has value for assessing the phytotoxicity of effluents.
leachates and receiving waters because it is less prone to interference by properties of these
samples, such as attenuation of illumination, than tests using unicellular algae. Specific
guidance on the testing of such samples is appended.
All OECD test Guidelines provide some latitude in methodology and the same is true of the
draft OECD Lemna test Guideline. Whilst this is necessary for practical reasons, it contributes
to vanability in estimates of toxicity between laboratories and also within a laboratory. A ring-
test was designed to assess key performance characteristics of the draft test method, especially
compliance with the critical quality criteria, and to assess the variability of the draft Guideline,
specifically the repeatability of the method within laboratories and reproducibility between
laboratories. Thirty-seven laboratories took part in the ring-test and more than half conducted
the requested five or more repeat tests. All participants used one or both of the specified
reference toxicants (potassium dichromate and 3,5-dichlorophenol).
More laboratories submitted data for L. minor than for L. gibba and potassium dichromate was
used more extensively in the ring-test than was 3,5-dichiorophenol. Adherence to the draft
Guideline in the ring-test was good. All participants supplied EC 50 data calculated from
average specific growth rates based on frond number, as required by the Guideline, and most
(83 %, n = 191) met the test acceptability criterion for control doubling time. Those tests that
failed this criterion were usually conducted under conditions of low illumination and/or low
temperature. Few exceedences of water quality criteria occurred, although the commonest
problem was an increase in pH of the control medium greater than that advised in the draft
Guideline. However, this did not seem to be associated with reduced growth rates.
The ranges of measured EC 50 valves for the two species of Lemna and two test substances are
summarised below (Table Si). When normahsed for differences in sensitivity, the data show
greater variability in tests when potassium djchromate was used than when 3,5-dichlorophenol
was used as the reference toxicant. Much of the variability seen in tests with potassium
dichromate could be attributed to the shallow dose-response assoc iated with this substance
R&D Technical Report EMA 003 1
-------
and consequent difficulties in determining an EC 50 with accuracy; in a significant number of
cases, it had been necessary for laboratories to extrapolate beyond the range of test
concentrations.
Table Si Interquartile ranges of EC 50 values generated in the interlaboratory ring-
test
Test substance
Lem
na minor (mg 1.1)
Lemna gibba
(mg 1.1)
Potassium dichromate
2
-4
(n
= 61)
8 - 30 (n = 28)
3 ,5-dichiorophenol
2.7 -
3.4
(n = 52)
6.00 - 7.00* (n =
4)
* = range of reported EC 50 values.
The data were analysed to identify the contnbutions to variability in EC 50 values resulting
from variability within laboratones and between laboratories. Those EC 50 values from tests
where the doubling time for control frond number was exceeded or where the EC 50 was
extrapolated beyond the concentration range tested were excluded from this analysis. As
expected, both sources of vanability were greater in expenments using potassium dichromate
but the between-laboratory variability generally accounted for a greater proportion of the
observed vanability than the variability found within-laboratones. in tests where L. minor and
potassium dichromate had been employed, two laboratones were responsible for a high
proportion of the between-laboratory vanability, suggesting their EC 50 estimates were rather
extreme.
Using the estimates of variance within-laboratones and between-laboratones, Quality Control
critena have been derived for accuracy (expressed in terms of the deviation from the
‘consensus’ mean EC 50 ) and for precision (the range of EC 50 values generated from repeat
tests within a laboratory). When used in conjunction with reference toxicant data, these
critena may be used by laboratones to assess the accuracy and precision with which they
perform Lemna growth inhibition tests. They may also be used by regulators when assessing
the quality of data to be used for chemical classification and nsk assessment purposes so that
excessive bias and variability can be identified.
Quality Control! criteria have been derived for L. minor when either potassium dichromate or
3,5-dich lorophenol are used as reference toxicants. However, useful criteria cannot yet be
established for L. gibba due to a lack of suitable data. Of the two toxicants used, 3,5-
dich lorophenol is to be preferred because of the difficulties in interpreting the very shallow
dose-response associated with potassium dichromate. Further work should concentrate on tests
using L. gibba and 3,5-dich lorophenol, performed according to the OECD draft Guideline so
that Quality Control cntena can be derived for this species also.
R&D Technical Report EMA 003 2
-------
ACKNOWLEDGEMENTS
We wish to thank the UK Environment Agency and US Environmental Protection Agency for
funding this project. The substantial contnbutions made by members of the OECD Lemna
Steenng Group and participants in the ring-test are also gratefully acknowledged.
KEY WORDS
Lemna minor, Lemna gibba, Organisation for Economic Co-operation and Development, test
Guideline, ring-test, potassium dichromate, 3 ,5-dich lorophenol, repeatability, reproducibility,
quality control.
R&D Technical Report EMA 003 3
-------
R&D Technical Report EMA 003 4
-------
1. INTRODUCTION
1.1 Background to development of an OECD Guideline for a Lemna
growth inhibition test
In 1998 the Organisation for Economic Co-operation and Development (OECD) produced a
review of aquatic toxicity test methods used for assessing the toxic effects of pesticides and
industrial chemicals (Kristensen et al, 1998). The aim of this review was to identify toxicity
test methods for which protocols and/or guidelines did not exist, and taxonomic groups that
were poorly represented in existing OECD Guidelines. The methods identified were assigned
a priority rating, as Groups 1, 2 or 3. Methods classified with the highest rating (Group I)
involved species for which no OECD Guideline existed but which were deemed necessary for
existing risk-assessment schemes. Furthermore, it was considered that these methods would
require only a small or moderate amount of work prior to nng-testing.
One of the methods assigned to Group 1 was a freshwater pelagic macrophyte toxicity test
using growth as the end-point of toxicity. The favoured genus, based on a body of existing
research, was Lemna; L. gibba for warm water environments and L. minor for cold water
environments. The genus Lemna comprises a group of small floating aquatic plants which are
to be found in many temperate regions of the world. This group of plants is useful for toxicity
testing because plants are small, reproduce rapidly and are relatively easily maintained in the
laboratory (Wang and Freemark, 1995).
The 1998 review prioritised the Lemna test because the existence of several draft standardised
test methods (ASTM 1991, EPA 1996, NAEP 1990) indicated that test protocols were already
at an advanced stage of development.
1.2 Development of draft test Guideline
During initial work on development of the draft test Guideline performed by KEMI it became
clear that certain aspects of the Guideline could not be finalised because of a lack of
knowledge in some areas. Consequently a Steering Group was set up with representatives
from several OECD countries (USA, Netherlands, UK, France, Italy and Sweden) and chaired
by WRc. Table 1.1 lists those individuals who served on this Steering Group during 1997-
1999. The Group met at the Joint Research Centre, Ispra in March 1997 at which the need for
further work to develop a draft test Guideline was agreed. This programme of work was
referred to as ‘Phase 1’ and is described in detail in Section 2 of this report.
R&D Technical Report EMA 003 5
-------
Table 1.1 The Lemna Steering Group, 1997-1999
Ebba Tiberg, KEMI, Sweden
Rick Petrie, EPA, USA
Annette Pectersson, AB Thallasa, Sweden
Henk Jenner, N.y. KEMA, Netherlands
Bernard Clement, Universite de Savoie,
France
Juan Riego-Sintes, JRC, Italy
Jerry Smrchek, USEPA/OPPT, USA
Melanie Dixon, Environment Agency. UK
Woodrow Wang, Geological Survey, USA
Ian Sims. WRc. UK
Jane Staveley, Carolina Ecotox, USA
Lennart Weltje, Deift University of
Technology, Netherlands
Paul Whitehouse, WRc, UK (chair)
Marie-Chantal Huet, OECD. Paris
Mait Hutchings, Zeneca, UK
When Phase 1 had been completed a second meeting of the Steering Group was convened at
WRc Medmenham in the UK in November 1997. This led to a revised draft of the test
Guideline which was subjected to a technical commenting round as a result of which further
refinements were made.
As with all OECD test Guidelines the resulting draft OECD Lemna test Guideline provides
some latitude in methodology and this inevitably contributes to variability in estimates of
toxicity. An interlaboratory ring-test was undertaken in 1998-9 to assess the performance of
the draft method under realistic conditions. Its aim was to assess key performance
characteristics of the draft test method, especially compliance with the critical quality criteria.
and to assess the variability of the draft Guideline, specifically the repeatability of the method
within laboratories and reproducibility between laboratories. This was referred to as ‘Phase 2
and is descnbed in detail in Section 3. Thirty seven laboratories from 15 OECD member
countries took part in the ring-test (Table 1.2). The names and affiliations of those concerned
are included in Appendix A.
The ring-test was co-ordinated by WRc with financial support for this activity from the UK
Environment Agency and USEPA Office of Prevention, Pesticides and Toxic Substances.
R&D Technical Report EMA 003 6
-------
Table 1.2 Ring-test participation
Country No. participants
Austria 1
Canada 1
Denmark 1
France 5
Finland I
Germany 6
Hungary 2
Italy 2
Japan 3
Netherlands 1
South Korea 1
Sweden
Switzerland 3
United Kingdom 4
USA 4
Total 37
1.3 Scope of the OECD test guideline
This report descnbes the development of the draft Guideline and it’s subsequent ring-testing
It does not detail all aspects of the test method but a copy of the draft Guideline is included in
Appendix B.
As an OECD test Guideline, emphasis was placed on the development of a method for the risk
assessment of chemicals. However, it was made clear by several national regulatory
authorities and members of the Steering Group that the method would also be valuable in the
context of effluent toxicity testing. Whilst this application is not specifically discussed in the
draft OECD test Guideline, Appendix E describes modifications to the Guideline that are
recommended for assessing the toxicity of effluents, leachates and receiving waters.
R&D Technical Report EMA 003 7
-------
R&D Technical Report EMA 003 8
-------
2. PHASE 1: RESEARCH TO DEVELOP THE DRAFT
GUIDELINE
2.1 Introduction
The Steering Group agreed that ‘Phase I’ studies should address the following issues:
Topic Objective
(a) Plant material To determine whether it was necessary and desirable to
specify spec fic clones or genolypes for testing and, if so, to
identify suitable clones. Also, to recommend procedures for
the culturing of Lemna.
(b) Test media To identify suitable growth media for L. gibba and L. minor
(c) Test vessels To identify suitable test vessels and volumes
(d) Test endpoints To identify technically and practically sound test endpoints
(e) Statistical analysis To recommend approaches for the expression of toxicity data
and procedures for statistical analysis
At the Steering Group meeting in Ispra in March 1997, a work programme was agreed and
responsibilities for carrying out different tasks were assigned. The results from these ‘Phase 1
studies were considered in detail when the Steering Group reconvened at WRc, UK in
November 1997. The key points arising from investigation of each of the issues (a - e) are
discussed below.
2.2 Plant material
There is an extensive literature describing the culturing of Lemna species and their use in
toxicity testing of chemicals and effluents. This is comprehensively reviewed by Wang (1990;
1991) and so is not reviewed further here.
Recent research carried out at the University of Connecticut and elsewhere has shown that
different clones of Lemna minor may be recognised (Les et al, unpublished; Cole and Voskuil,
1996; Vasseur et al, 1993) and that they can differ in their sensitivities to chemicals (Cowgill
et a!, 1991; Mazzeo ef al, 1998; Bengtsson et a!, unpublished). However, experience gained in
using different clones of L. minor suggests only small variations in sensitivity to chemicals
(Pettersson, pers. comm.). Different species of Lemna also exhibit differences in sensitivity to
chemicals although any species effect appears to be small and dependent on the test chemical
(Cowgill et a!, 1989; Cowgill et al, 1991). Whilst the identity of species maintained by
members of the Steering Group and used for testing were generally known, none of the
Steering Group knew the clone of Lemna minor that they were keeping in culture. It was felt
this was likely to be the case in most laboratories.
R&D Technical Report EMA 003 9
-------
Regulatory agencies may specify toxicity test data for either L. minor or L. gibba and so
provision for both species was considered necessary but it would not be realistic to specify
particular clones of L. minor or L. gibba in the test Guideline. However, the species of Lemna
used for testing should be known and reported. This guidance is consistent with other OECD
test Guidelines describing toxicity tests with algae and fish, where a range of species is
covered in the test Guideline. Methods for the identification of Lemna species are reported to
be available (Correll and Correll, 1972. Les et al, unpublished), based on the association of
different Lemna species with unique markers in the DNA sequence.
The conditions for acquiring and culturing Lemna for toxicity testing were considered to be
important. In particular, plant material used for ecotoxicity testing should be free from
contamination by algae. Experience of the Steering Group members had shown that conditions
which favoured growth of Lemna may also encourage the growth of unicellular algae. This
was to be avoided by surface sterilisation of new plant matenal (either acquired from another
laboratory or from the field) and detailed guidance on this procedure is provided in Annex 3 of
the draft test Guideline.
Whilst acquisition of plant material from other laboratories was the preferred source.
collection from the field was considered an acceptable alternative as long as contaminating
organisms were removed and the plant material treated to remove algal contamination. After
debate, it was felt that a requirement to use only a truly axenic culture which was free of
microbial contamination was not realistic and, for this reason, the need for fully aseptic
conditions during culture of Lemna was considered unnecessary. Nevertheless, certain steps to
minimise the risk of algal contamination are advised. These include the use of autoclaved
glassware for culturing and testing, and sub-culturing of Lemna cultures under sterile
conditions (e.g. positive pressure sterile cabinets).
Under good growing conditions it is usually necessary to sub-culture approximately every 10
days. However, if tests are performed only infrequently, cultures of Lemna may be maintained
for long periods without the need for sub-culturing by placing them in an illuminated
incubator at 4 - 5°C. Lemna fronds may be easily transferred between laboratories by post or
courier on damp filter paper in a sturdy container. Under these conditions, a new culture may
be established even after several days in the dark. However, as with plant material that has
been kept under cold conditions, a period of acclimation to rapid growing conditions (of the
order of two or three weeks) is advised before fronds are used for toxicity testing.
2.3 Growth media
Several media have been reported for growing Lemna in culture and this subject was reviewed
by Wang (1987). Most media are based on Hoaglands medium for growing plants
hydroponically although modified algal growth media have also been employed. The medium
recommended in the draft Swedish guideline (‘modified SIS’) is noteworthy in that it contains
lower concentrations of nitrate than, for example, media based on Hoaglands medium.
The effects of different media on growth rate of Lemna can be difficult to separate from the
effects of light intensity or renewal frequency. However, it is clear that the optimum pH for
growth of L. minor and L. gibba lies between pH 5-7. EDTA is normally incorporated to
regulate availability of trace nutrients although its possible chelating effects (and hence effects
R&D Technical Report EMA 003 10
-------
on bioavailability of metal toxicants) are well-known (Sims et al, 1997). The draft Guideline
recommends that when testing metals with L. minor (or other substances whose stability or
bioavailability may be affected by pH), consideration should be given to pH control using
MOPS buffer.
It was agreed that the same medium should be used for culturing and testing and, if possible,
the same medium should be used for both L. minor and L. gibba (although subsequent studies
found this was not practical).
Rapid growth in controls was emphasised to (a) ensure that plants were free from stress due to
nutrient limitation and (b) promote precision - unpublished studies by Wang show smaller
coefficients of variation for control doubling time in media which promote rapid growth.
Indeed, the only test validity criterion (i.e. that must be satisfied for an acceptable test) related
to plant growth rate. This requires that the doubling time in the controls should not exceed 2.5
days. Subsequent ring-test data were analysed to investigate whether or not doubling time was
correlated with sensitivity to the test chemicals used (Section 3). It was also generally
acknowledged that rapidly growing cultures are associated with a high proportion of colonies
consisting of 3 - 4 fronds. A high incidence of single or only double fronds is indicative of
environmental stress, such as nutrient limitation, and will often be associated with reduced
growth rates.
The following criteria were agreed for the selection of suitable growth media:
• predominantly inorganic although EDTA should be present to maintain
concentrations of micro-nutrients, particularly iron. The concentration of EDTA
should be as low as possible (to minimise possible interactions with test substances)
and in a 1:1 molar ratio with the following di- and trivalent cations (Zn 2 , Cu 2 and
Fe 3 );
• supports exponential growth over 7 days without renewal;
• supports a high growth rate (a doubling time of no more than that specified in the
draft Guideline - 2.5 days);
• nitrate, phosphate and iron are not present in excess so that possible interactions with
test substances are minimised;
• pH should favour Lemna growth but be environmentally realistic (between pH 5-9);
• buffering capacity - pH should not vary excessively over a 7-day period (± 1.5 pH
units from the starting pH and preferably without the addition of buffer);
• only one medium should be permitted for each Lemna species to encourage
reproducibility of toxicity tests performed in different laboratories with the same
species.
Considerable effort was subsequently invested by AB Thalassa and Zeneca in an evaluation of
suitable growth media for L. minor and L. gibba based on the criteria listed above. None of
those available at the time clearly met all the criteria. However, two media, the modified
R&D Technical Report EMA 003 11
-------
Swedish medium (SIS) specified in the Swedish test guideline (NAEP, 1990) and a variant of
Hoaglands medium (LMI I), were short-listed for more detailed practical assessment in eight
laboratories. This took the form of measurements of growth in control vessels while
monitoring the change in pH under test conditions.
Table 2.1 summarises the results of this interlaboratory comparison. It is evident that growth
rates of both species were acceptable in both media although data for L. gibba are limited to
just two studies. Comparisons of Lemna growth in static and semi-static expenments by WRc
and University de Savoie (renewals every 3 days) indicated modest (ca. 10%) increases in
growth rate when the nutrient media were renewed but these differences were not statistically
significant. A separate experiment into the influence of hght intensity (AB Thalassa) showed
significantly higher growth in SIS medium at 120 tEm 2 s ’ than at 60 .tEm 2 s with doubling
times of 39 and 50h, respectively. Variations in light intensity by laboratories participating in
the interlaboratory assessment of growth media are likely to have contributed to the observed
variations in growth rates.
Table 2.1 Interlaboratory comparison of Lemna growth in different media
Species
Medium
Growth rate
Doubling
No.
of
time (d)
experiments
(increase in frond number d’)
mean range
L.
minor
SIS
0.33 0.23-0.43
1.6-3.1
11
L.
minor
LMI1
0.32 0.22-0.41
1.7-3.2
11
L.
gibba
SIS
0.32 0.29 - 0.36
1.9 - 2.4
2
L.
gibba
LM/1
0.35 0.31 - 0.39
1.8 - 2.2
2
•
Although there was little to choose between the SIS and LM/1 media, the former was
preferred by virtue of the lower levels of nutrients and the greater ease of preparation of
certain constituents. Changes in pH over time during toxicity testing were evident, particularly
when test media were not renewed. However, these variations fell within the ± 1.5 pH units
stipulated in the criteria and so were deemed acceptable. At a subsequent meeting of the
Steering Group, it was agreed that buffer control of pH would not routinely be required but
that 2.3mM 3- [ N-morpholino] propanesulfonic acid buffer (MOPS; pKa = 7.20) could be used
in SIS medium if necessary e.g. because of pH effects on bioavailability, speciation or
hydrolytic stability of the test substance. To prevent precipitation of nutrients in the growth
medium the MOPS stock solution should be pH adjusted with sodium hydroxide before
addition.
Although based on a much smaller dataset, the interlaboratory evaluation of growth media
indicated that SIS medium may also be acceptable for culturing and testing L. gibba as well as
L. minor. Studies carried out at AB Thalassa (Petersson pers. corn.) indicated that 20 x AAP
R&D Technical Report EMA 003 12
-------
medium may be more difficult to prepare due to precipitation and possibly poor pH control.
However, there is more expenence of growing and testing L. gibba on a modified algal growth
medium (20x AAP) and this is the medium specified in a recently drafted ASTM Lemna test
guideline (E1415-91;1995). Since control growth on 20x AAP appears to be acceptable and
the available data for alternative media are so limited, 20x AAP medium was recommended
for L. gibba rather than either the SIS or LM/l media. Nevertheless, the optimum growth
medium for L. gibba remains an area of some uncertainty. The compositions and preparation
of the SIS and 20x AAP growth medium are appended to the draft test Guideline.
When initialising cultures with a new medium, growth may be poor initially and an
acclimation period of at least one month is advisable.
2.4 Test vessels and volumes
In addition to glass test vessels, plastic vessels would be acceptable where the test substance
was unlikely to be sorbed to such materials. The depth of growth medium should not restrict
root growth and the surface area available should not result in overlapping fronds at test
termination. Minimum test volumes were specified in the draft Guideline because expenments
had indicated that nutrient limitation, resulting in reduced growth rate, was evident in small
test volumes. A minimum volume of 100 ml was selected, based on the experience of the
Steenng Group.
2.5 Test endpoints
2.5.1 Principle of the test method
The principle behind the draft OECD test Guideline is a growth inhibition test in which
chemical effects on growth parameters of a group of plants are compared with those of control
plants. Chemical toxicity may become evident through effects on a number of parameters,
although for routine use, changes in frond number, frond area, (fresh or dry) weight and
chlorophyll content are the most conveniently assessed. Wang (1990) considers toxicity
endpoints that may be employed in Lemna growth tests. The following summarises the
deliberations of the Steering Group in selecting suitable endpoints for inclusion in the draft
Guideline.
2.5.2 Frond number
Growth may easily be determined in terms of changes in total frond number; this endpoint is
readily assessed without recourse to expensive apparatus and is non-destructive so that
measurements may be made on the same treatment on several occasions during the course of a
study. This is an important feature when growth inhibition is expressed in terms of average
specific growth rate (Section 2.6). The use of total frond number also effectively integrates the
effects of chemicals on the survival of fronds, the time taken to produce new fronds and the
extent of budding (vegetative reproduction). It may thus be argued that changes in frond
number provide a direct measure of a population-level endpoint. As such, this should provide
R&D Technical Report EMA 003 13
-------
a high degree of biological relevance to the risk assessor when establishing an acceptable
concentration for a chemical in a waterbody.
A potential disadvantage of total frond number as a test endpoint concerns the definition of a
frond, particularly when the plant divides and gives rise to a small distortion which may or
may not be regarded as a new frond. Consequently, some systematic error is conceivable as a
result of different operators achieving slightly different estimates of frond number from
identical groups of plants.
2.5.3 Frond area
The total area of fronds determined during the course of the experiment is a potentially useful
endpoint because it is also non-destructive and so may be used to denve summary statistics
which are based on average specific growth rate. This parameter has the added advantage that
- unlike frond number as an endpoint - chemical effects on the size of individual fronds can be
taken into account. Of course, in this instance, frond numbers have also to be established in
order to calculate the mean frond area. Whilst proprietary products are available which permit
total frond areas to be determined quickly and accurately (by projecting the silhouette of the
fronds in a test vessel onto a video camera and digitising the resulting image, whose area can
then be determined), such facilities are not widely available.
2.5.4 Dry/fresh weight
Whilst there are sound arguments in favour of dry weight and fresh weight as test endpoints,
these are necessarily destructive parameters and so can be determined on only one occasion.
As a result, initial (‘t 0 ’) estimates of these parameters must be made on separate sets of plant
material and, more importantly, toxicity may only be expressed in terms of final biomass
(Section 2.6). There is evidence that fresh/dry weights, frond number and area are highly
correlated (Cowgill and Milazzo, 1989).
2.5.5 Chlorophyll content
Chlorophyll content provides a useful indication of chemical toxicity and would probably
correlate with fresh or dry weight in most instances. As with measurements of fresh and dry
weight, chlorophyll can only be determined by sacrificing plant material and so it can be
determined on only one occasion. As a result, initial (‘to’) estimates of chlorophyll content
must be made on separate sets of plant material. Again, toxicity may only be expressed in
terms of final biomass.
R&D Technical Report EMA 003 14
-------
2.6 Expression of toxicity and data analysis
2.6.1 Expression of toxicity
Chemical effects on growth may be expressed either in terms of (a) final biomass’ or (b)
average specific growth rate (p.).
(a) Final biomass may be determined on the basis of any of the endpoints described above
and simply involves a comparison of biomass in each treatment and the controls at the
end of the test period.
(b) Average specific growth rate is based on a comparison of growth rates in each treatment
and the controls over the entire duration of the test. Because assessments need to be
made on several occasions during a test a non-destructive endpoint is preferred.
Therefore, frond number is clearly the most convenient parameter if toxicity is to be
expressed in this way, although total frond area is also suitable. The use of dry or fresh
weight can be used, but these will necessitate the use of additional replicates that can be
sacrificed during the test to enable these measurements to be made.
Evidence from the ring-test and elsewhere clearly shows that summary statistics expressed in
terms of biomass (EbCSO) are generally lower than those based on average specific growth rate
(ErC5O). This is a consequence of the mathematical derivation of these summary statistics and
is explained by Huebert and Shay (1993) and by Nyholm (1990). However, estimates of
toxicity based on final biomass are more dependent on exposure duration than those based on
specific growth rate, and toxicity based on final biomass as the test endpoint only describes
the effects occurring at a particular time. This issue has been thoroughly explored in the
context of algal toxicity data (Nyholm, 1985, 1990) and is pertinent here because exactly the
same principles apply to the expression of chemical toxicity on Lemna growth. Indeed, this
very issue has been explored in relation to studies with Lemna by Huebert and Shay (1993).
They conclude that average specific growth rate is a technically supenor expression of toxicity
compared to final biomass because it is less dependent on exposure period and consequently
resulting expressions of toxicity should be more useful in regulatory decision-making. Indeed,
Huebert and Shay (1993) argue that comparisons of toxicity based on final biomass cannot be
made between studies carried out in different laboratories or at different times and are
unsuitable for regulatory decision-making. However, when the test duration is fixed (as it
would be in the case of an OECD test guideline), this criticism does not really apply.
This issue was debated at length by the Steenng Group. Ultimately, which summary statistic is
used is a philosophical decision about whether we are primarily concerned with chemical
effects on the rate at which a Lemna population grows or on the size of that population at a
particular time. The greater robustness of the average specific growth rate was, on balance,
preferred but it was agreed that regulators should be provided with estimates of both EbC5o
and ETC5O. Consequently, methods for estimating both summaries are included in the test
guideline.
sometimes referred to as fina1 yield’
R&D Technical Report EMA 003 15
-------
It follows that:
(a) test endpoints based on non-destructive parameters (frond number or frond area) were
required and it was not sufficient to rely solely on those which permit only an
assessment of final biomass (i.e. dry weight, fresh weight, chlorophyll content).
(b) Frond number is preferred over frond area because it is easily determined in any
laboratory.
(c) Useful information may also be obtained by reporting other test endpoints and so the test
Guideline allows the operator to select from any of the following to be reported in
addition to estimates of average specific growth rate based on frond number: frond area,
dry weight, fresh weight or chlorophyll content.
(d) The use of average specific growth rate is only stnctly valid if exponential growth can
be demonstrated in the controls. The test conditions specified in the draft Guideline are
intended to achieve this. In the event that exponential growth (or near-exponential
growth) is not achieved in the controls, estimates of growth inhibition may be based on
the area under the growth curve. However, it should be observed that the existence of a
lag phase or plateau in growth are indicative of sub-optimal growth conditions. It is to
minimise the risk of nutrient limitation (Haftka and Weltje, 1999) that a 7-day growth
test is favoured over static tests of a longer duration.
2.6.2 Data analysis
The draft test Guideline details methods for calculating % growth inhibition for (a) average
specific growth rates, (b) area under the curve and (c) final biomass. Guidance on the
estimation of the lowest observed effect concentration (LOEC) and hence the no observed
effect concentration (NOEC) is provided 2 , and also point estimates such as the EC 50 . Suitable
statistical models of the Lemna dose response include the logistic model and the cumulative
normal model (Bruce and Versteeg, 1992). The variability among replicates is not expected to
be constant across concentrations and so a weighted analysis is recommended so that more
emphasis is placed on the observations displaying less vanability. Since higher variability is
expected to be associated with rapid growth, the weightings used are the inverse of the
predicted values. Non-parametric approaches may also be used if there is a poor fit of the
experimental data to these models. In this case, a suitable approach would be linear
interpolation with bootstrapping (‘ICp’), as described by Norberg-King (1988).
In growth tests, it is possible that stimulation of growth can result from exposure to low
concentrations of some chemicals (‘hormesis’). If this occurs, the draft Guideline advises that
a suitable hormesis model be used.
2 The shortcomings of hypothesis testing e g for estimation of a LOEC and NOEC are now generally
acknowledged (Chapman eta!, 1996) These stem from their dependence on the disposition of test
concentrations, the degree of within-test error, the skill of the experimenter, the choice of statistical test and
the Type I error rate selected in significance tests (Newman. 1995)
R&D Technical Report EMA 003 16
-------
3. PHASE 2: RING-TESTING OF DRAFT OECD TEST
GUIDELINE, DATED JUNE 1998
3.1 Objectives of ring-test
Following a technical commenting round amongst members of the Steering Group, a revised
draft Guideline was prepared in June 1998 and this formed the basis of a major interlaboratory
ring-test in the second half of 1998 and early 1999. The objectives of the ring-test were three-
fold:
1. To establish the extent to which the test method could be performed and the test validity
criteria met.
2. To evaluate the repeatability and reproducibility of the test method, based on responses
(EC 50 ) to two reference toxicants, potassium dichromate and 3,5-dichlorophenol
3. To analyse these EC 50 data in such a way that Quality Control criteria for accuracy and
precision of Lemna toxicity tests could be derived, thereby providing a means for
laboratories or regulators to judge the performance of testing.
3.2 Design of ring-test
3.2.1 Choice of reference toxicants
After considerable debate within the Steering Group it was decided that the two test
substances to be used for this ring test would be 3,5-dichiorophenol (DCP) and potassium
dichromate. These chemicals were chosen because:
1. their effects on the growth of Lemna are well-documented (Wang, 1987; Pluta, pers.
comm);
2. they have different modes of action;
3. stock solutions could be prepared without the use of solubilising agents;
4. they are stable over the test duration (7 days);
5. they represent two well-defined groups of toxicants, i.e. metals and organics.
Both materials have been widely used in nng-tests and as reference toxicants in tests with
other organisms (Wang, 1987; Bjornestad and Petersen, 1992).
R&D Technical Report EMA 003 17
-------
3.2.2 Guidance to participants
Participating laboratories were invited to perform non-overlapping repeat 7-day Lemna growth
studies in accordance with the draft Guideline. Each participant was provided with a copy of
the revised (June 1998) draft Guideline (Appendix B). Subsequently, instructions for reporting
and analysing data were supplied (Appendix C). Laboratories were encouraged to perform
tests in the same way so that variations in methodology should be no more than that allowed
by the draft Guideline.
The choice of which substance(s) and species of test organism to use was left to the
participants, although the desirability for participants to conduct repeat tests rather than test
both substances and/or both species was stressed. Repeat tests would be needed to assess
within-laboratory variability (repeatability). Consequently, participants were asked to conduct
five repeat tests if possible.
3.3 Participation in the ring-test
A total of 37 laboratories participated in the ring-test and they are listed in Appendix A. Table
1.2 groups the participants by country and illustrates the wide geographical spread
encompassed, including Europe, North America and the Far East. Table 3.1 summanses the
data received, breaking it down by test species and substance. The number of participants
conducting repeat tests was high with almost half (43%) conducting the requested five or more
repeats. Some participants tested both species of Lemna while some tested both toxicants.
Table 3.1 Data sets submitted by participants, broken down by test organism and by
toxicant, and showing the number of repeat tests performed and included in
the statistical analysis of the ring-test results.
Species
Test
substance
1
2
Num
3
her of
4
repea
5
t tests
6
7
8
Total No
data sets
submitted
Total No data
sets analysed
L. minor
chromium
1
3
4
3
7
1
1
1
87
61
L. minor
3,5-DCP
-
2
3
2
7
-
-
-
56
52
L. gibba
chromium
1
-
4
2
3
-
-
1
44
28
L gibba
3,5-DCP
1
-
1
1
-
-
-
-
8
4
It is clear from Table 3.1 that most participants opted to carry out experiments using L minor
and that potassium dichromate was used more extensively that 3,5-dichlorophenol. As
explained below (Section 3.4.1), not all the data were suitable for analysis and this explains
the difference between the number of datasets submitted and the number analysed.
R&D Technical Report EMA 003 18
-------
3.4 Statistical analysis of data
3.4.1 Estimation of EC 50 values
Participants were asked to analyse their own data using the methods described in the draft
Guideline. The results were to be supplied to WRc by the end of January 1999 on data sheets
provided for this purpose. Data were requested in the form of EC 50 s calculated from average
specific growth rates based on frond number. All participants conducted these calculations
and some reported EC 50 s based on additional endpoints. These are discussed in Section 3.5.3.
Participants were asked to express the toxicity of potassium dichromate in terms of Cr and
this appears to have been done by those using this substance. All EC 50 values are based on
nominal concentrations. NOECs and LOECs were not requested as part of the ring-test, but
many participants supplied these as well.
Those tests not meeting the test validity criterion for control doubling time and/or where the
EC 50 was extrapolated beyond the concentration range used were excluded from subsequent
statistical analysis of the nng-test results. Of the 131 data sets for potassium dichromate, 42
(32 %) were excluded from the statistical analysis on the basis of the data being extrapolated
and/or excessive control doubling times. With DCP, 64 data sets were received, 8 (12.5 %) of
which were excluded due to excessive control doubling times. None of the EC 50 values
derived from tests with DCP had been extrapolated beyond the test concentrations used.
Difficulties in selecting a suitable concentration range in tests using potassium dichromate
accounted for the high incidence of extrapolated EC 50 values for this compound. This is
discussed in Section 3.5.
3.4.2 Identification of the components of variance
Possible sources of variability, i.e. within-laboratory and between-laboratory, were partitioned
using Residual Maximum Likelihood (REML) procedures using log-transformed EC 50 data.
This is a generalisation of the General Linear Model which includes both regression and
analysis of variance (ANOVA). Whilst ANOVA can isolate several sources of variation, it
requires a balanced test design. This requirement was clearly not met in the ring-test because
the number of repeat tests and number of laboratories contributing data varied. Regression
analysis can deal with an unbalanced design but requires only a single source of error. REML
can be used to investigate several sources of variation in an unbalanced design. The statistical
package used to perform the analysis was Genstat, release 5.3.1 (Payne et al, 1993).
3.4.3 Influence of statistical method on EC 50
The draft test Guideline permits the use of several statistical models for estimating EC 50
values. To investigate whether different statistical methods used by the participants for
estimating EC 50 values had any effect on the resulting EC 50 s (i.e. a possible source of bias),
the data for tests using L. minor with potassium dichromate and DCP, and for L. gibba with
potassium dichromate were classified into two sets according to the statistical method used.
One set comprised EC 50 s calculated using plausible forms of dose-response curve (Probit,
Logit, Hill and Weibull models), whilst the other Set compnsed those derived using other
R&D Technical Report EMA 003 19
-------
more empincally based methods (linear regression, including ICp and graphical methods),
although we do not have the full details of the various methods used. The EC 50 values
obtained in these two groups were then compared by including ‘statistical method’ as a fixed
effect in R.EML.
3.4.4 Influence of doubling time on sensitivity
To investigate whether growth rate had any effect on the resulting EC 50 s, the EC 50 s derived
using L. minor with potassium dichromate and DCP (the largest datasets) were plotted against
control doubling time.
3.5 Results and Discussion
3.5.1 General comments
As far as we could ascertain, participants conducted tests with one of the two species specified
(L. minor or L. gibba), but the clones used are largely unknown. Some participants reported
using L. gibba G3.
The draft Guideline recommended that modified Swedish Standard Medium (SIS) be used for
culturing and testing L. minor, and that 20 x AAP medium be used with L. gibba. All
participants complied with this requirement, some doing additional work with other media.
The draft Guideline required growth to be measured in terms of frond number and this
requirement was met by all participants. In addition, final biomass, e.g. frond area, dry or fresh
weight, could also be utilised. All participants measured frond number, some also measured
biomass. Three participants utilised chlorophyll content as an endpoint while one measured
root elongation.
The draft OECD Guideline recommended that temperature and illumination of the test area
and the pH of the growth medium in the controls be measured at the start and end of the test.
All participants complied with this minimum requirement, several making more extensive
measurements of temperature and illumination. Three participants reported that at least one
replicate test was conducted under low illumination, while four participants reported low test
temperatures. All instances of low temperature resulted in poor control growth, i.e. the test
validity cnterion for control doubling time was not met, while two of those reporting low light
intensities failed the same criterion. As noted in Section 3.4.1, data generated from these tests
were not included in the statistical analysis. This contributed to the difference between the
number of datasets submitted and analysed (Table 3.1). It is interesting to note that a higher
proportion of datasets where potassium dichromate was used as the reference toxiCant were
excluded than occurred with tests using DCP. This is due to the flat dose response seen in
studies with potassium dichromate which made it difficult for participants to correctly identify
a suitable range of test concentrations. Consequently, there was a high incidence of
extrapolation to estimate EC 50 values with this substance, i.e. the EC 50 frequently lay outside
the tested range of concentrations.
R&D Technical Report EMA 003 20
-------
The most commonly encountered problem regarding excedances of environmental conditions
was that of excessive change in medium pH. A total of nine participants (24%) expenenced an
increase in pH of the control medium in at least one repeat test that was greater than that
allowed by the draft Guideline. At least one of these was associated with contamination of the
test vessels with unicellular algae. These data were retained in the subsequent analysis unless
there was evidence that control growth rate was also impaired. The draft test Guideline
recommends that the pH should not deviate by more than 1.5 pH units. Within this range,
effects on growth rate due to pH are probably nor important. However, a narrower range is
preferable when testing metals because their speciation, and hence toxicity, is highly
dependent on the pH of the medium.
One participant working with 20 x AAP media and L. gibba reported precipitation of the
macronutrients in the stock solution, while three participants reported that the presence of
MOPS buffer, used in the SIS medium when testing potassium dichromate, had a negative
effect on control growth. One participant used acetone as a camer with DCP.
Many statistical methods were employed by participants for analysing the data. Of these, the
most frequently used was Probit analysis (35 %). This was followed jointly by ICpfLogit
techniques (25 %) and regression (linear and non-linear) (25 %). A total of 10 % employed
graphical methods and two participants (5 %) supplied no information regarding the statistical
methods they used.
3.5.2 Estimation of EC 50 values
All the EC 50 values for the four species/toxicant combinations are included as Appendix D.
Data for L. minor/chromium, L. minor/DCP and L. gibbalchromium are summansed in Figure
3.1 in the form of box-whisker plots. Insufficient data are available for L. gibbafDCP to
permit this type of analysis. The vertical line within the box represents the median, the
horizontal extent, horizontal box indicates the interquartile range, and the ‘whiskers’ span the
range between the 5 and 95 percentile values. Outlying data are represented by asterisks.
The left-hand set of plots shows the data expressed on an absolute concentration (mg 1.1) scale.
To remove the influence of differences in sensitivity to the two test chemicals, the plots on the
right are expressed in terms of the deviation in EC 50 from the median value. This has the effect
of normalising variability in EC 50 values in relation to the median.
R&D Technical Report EMA 003 21
-------
‘ac
o io 20 0 500 1000
.1 L.m,nor/ chromium
L.minorl chromium (mgi ) (% deviation from median)
*fJJ—ixx
I I I —
0 5 10 0 500 ioool
• .1 I L.minor/ DCP
L.minorl DCP (mgi ) (% deviation from median)
0 20 40 60 0 500 1000
L. gibba / chromium
L. gibba / chromium (mgF’) (%deviation from median)
Figure 3.1 Box-whisker plots of EC 30 values (as mg 1’ and % deviation from median
EC 50 )
As expected, distribution of EC 50 values shows a marked skew with respect to concentration.
For this reason, subsequent REML analysis was performed using log-transformed EC 50 data. It
is clear that tests carried out using potassium dichromate led to greater variability in the EC 50
estimates than tests in which DCP was used. This is shown by the considerably larger inter-
quartile ranges in tests where potassium dichromate was used as the reference toxicant. With
L. minor, a total of 66 % of the EC 50 s for chromium and 93 % for DCP were within a factor of
5 of each other (1 - 5 mg Cr or DC? l’). For L. gibba and potassium dichromate, 70 % were
within a factor of 5 of each other (6 - 30 mg Cr l ), similar to that seen in the studies using
L. minor. The data-set for DCP to L. gibba was too small to enable the same comparison to be
made, but the variability in EC 50 s within the one laboratory that accomplished repeat testing
(Lab. 34) was low (coefficient of variation = 6.4%).
L. minor was clearly the more sensitive of the two species to potassium dichromate (median
EC 50 of 2.5 mg F’ compared with 16.8 mg F’ for L. gibba). This may be due to the higher
R&D Technical Report EMA 003
-------
concentration of EDTA present in the 20 x AAP medium used with L. gibba reducing the
availability of chromium to the test plants. Unfortunately, the data set for the toxicity of DCP
to L gibba is too small to enable its comparison with the corresponding L. minor data set.
3.5.3 Comparison of test endpoints
As requested, all participants used frond number to calculate EC 50 s based on average specific
growth-rate (E 1 C 50 ). In addition, several participants used alternative measures of growth for
calculating EC 50 s (Tables 3.2 - 3.5).
Table 3.2 Number of participants reporting EC 50 s for different endpoints with Lemna
minor and potassium dichromate.
Frond
number
Dry
weight
Test endp
Fresh
weight
oint
[ Chlorophyll]
Frond
area
Average
specific
growth rate
21
—
—
—
3
Areaunder
curve
7
0
Final biomass
1
6
0
1
2
Doubling time
0
0
0
0
0
Table 3.3 Number of participants reporting EC 50 s for different endpoints with Lemna
minor and DCP.
Frond
number
Dry
weight
Test endp
Fresh
weight
oint
(Chlorophyll]
Frond
area
Average
15
—
—
—
i
specific
growth rate
Areaunder
5
0
curve
Final biomass
2
4
1
3
1
Doubling time
1
0
0
0
0
R&D Technical Report EMA 003 23
-------
Table 3.4 Number of participants reporting EC 50 s for different endpoints with Lemna
gibba and potassium dichromate.
Frond
number
Dry
weight
Test endp
Fresh
weight
oint
[ Chlorophyll]
Frond
area
Average
specific
growth rate
9
—
—
—
0
Areaunder
curve
2
—
—
—
0
Final biomass
1
2
1
0
0
Doubling time
0
0
0
0
0
Table 3.5 Number of participants reporting EC 50 s for different endpoints with Lemna
gibba and DCP.
Frond
number
Dry
weight
Test endp
Fresh
weight
oint
[ Chlorophyll]
Frond
area
Average
specific
growth rate
3
—
—
—
1
Areaunder
1
—
—
—
0
curve
Final biomass
0
1
0
0
0
Doubling time
°
0
0
0
0
The data-set for final biomass (EbC5O) based on dry weight at seven days was the most
extensive of the alternative end-points reported. Consequently, the data for L. minor with
potassium dichromate and DCP were used to compare ErC5O values based on average specific
growth-rate (calculated using frond number) with EbC 5 OS based on final biomass (calculated
R&D Technical Report EMA 003 24
-------
using dry weight) from the same experiments. Mean and median EC 50 s were calculated for
these two methods (Table 3.6).
Table 3.6 EC 50 values for L. minor calculated using average specific growth-rate
based on frond number (ErC 50 S) and final biomass based on dry weight
(E bC5O).
Test substance
KCr
(E Cso)
207
(EbC5O)
3,5
(ErC 50 )
-DCP
(EbCSO)
mean (mg F’)
7.7
0.84
2.9
1.9
median (mg F’)
2.7
0.71
2.9
1.7
In tests with potassium dichromate the method used for measuring growth inhibition had a
large influence on the resulting EC 50 values. There is a factor of nine between the mean EbC5o
and the mean ErC 50 , with the mean E C 50 being the higher. With DCP the difference between
means is much smaller (factor = 1.5), but again mean ErCSOS are higher than mean EbC 5 OS. A
similar pattern is evident for median EC 50 values.
The large differences between ErC5O and EbC5o values seen with potassium dichromate (mean
and median E 1 C 50 s being 9 and 4 times greater than their respective EbC5os) are exacerbated by
the shallow dose-response for this substance. It is clear that, when based on the same endpoint,
estimates of ErC 50 will typically be higher than estimates of EbC 5 o, especially when the slope
of the dose response is shallow (Nyholm, 1990).
3.5.4 Test repeatability and reproducibility
Section 3.5.2 showed that estimates of toxicity were subject to variability. When we compare
results arising from different laboratories we are actually seeing the combined effects of
several sources of error: the within-test error (the variability in response between replicates),
the within-laboratory error (the variability between results of tests conducted on different
occasions) and the between-laboratory error (the variability that arises when we compare
results of tests conducted in different laboratories). Using REML analysis, the ring-test data
were analysed to identify the sources of error contributing to the variability in EC 50 estimates.
This technique was used to partition the variances which resulted from:
(a) within-laboratory error (O 2 wichin.iab)
(b) between-laboratory error (a 2 between.,ab).
The repeatability of a method depends on the sum of the errors arising from within-test and
within-laboratories, and may be expressed as:
0 repeacabi lity = within—test ÷ 02 v ithsn—lab
R&D Technical Report EMA 003 25
-------
Similarly, the reproducibility of a method depends on the sum of the errors arising from
within-test, within-laboratories, and also between-laboratones, and may be expressed as:
I ,
areproducibiutv = - within — rest + O within — lab + a - between — lab
In this ring-test, it was not practical for participating laboratories to use sufficient replication
to provide an estimate of within-test error and so in the REML analysis this source
of variability is effectively ‘nested’ within the &wIthifl..iab estimate.
An initial analysis of the data clearly showed that in the case of experiments performed using
Lemna minor and potassium dichromate, a number of laboratories produced EC 50 values
which were consistently different to those reported by the other laboratories (Appendix D).
Data from two laboratories in particular deviated greatly from the other data (laboratories 19
and 31) although smaller discrepancies were also evident in the results obtained by
laboratories 10 and 28. It was felt that there may be a source of bias in these datasets to give
rise to such markedly different results which would distort the estimate of between-laboratory
variance ( 2 t,etween-iab). Consequently, REML analysis was performed on L. minor/potassium
dichromate data (a) after excluding data from laboratories 19 and 31 and also (b) after
excluding data from laboratories 19, 10, 28 and 31. Although differences between laboratories
were also evident in tests using L. minor and DCP, they were smaller and omission of outlying
data could not be justified. Insufficient data were generated during the ring-test to permit
analysis of tests in which Lemna gibba was exposed to DCP.
Table 3.7 summarises the results obtained from REML analysis and the main points are
discussed below. This table includes estimates of the ‘consensus’ mean for each
specles/toxicant combination. This is the mean of the mean EC 50 values generated by each
laboratory and may be regarded as an approximation of the ‘true’ toxicity of the toxicant to the
species in question.
Lemna minor/potassium dichromate
The ‘consensus’ mean EC 50 was 3.1 mg I-i (expressed as chromium). When data from
laboratories 19 and 31 were excluded, between-laboratory variance contributed a greater
proportion of the overall variance than that arising within laboratories (a 2 betweeniab is greater
than aWithin.iab) but this was reversed when data from laboratories 10 and 28 were also
excluded. As expected, the estimate of a r uCibih declined as more of the laboratories
generating apparently discrepant data were omitted.
Lemna minor/DCP
Much smaller variances were associated with experiments camed out with L. minor when
DCP was used as the reference toxicant. This applies to estimates of both a 2 withiniab and
0 between-iab. Consequently, both the repeatability and reproducibility of these tests were
superior to that obtained when potassium dichromate was used as the toxicant. REML analysis
shows that variations between laboratories contributed a greater proportion of the overall error
than that arising from within laboratories.
R&D Technical Report EMA 003 26
-------
Lemna gibbalpotassium dichromate
The available data confirm that L. gibba is much less sensitive to potassium dichromate than
is L. minor (mean EC 50 9.8 mg II). REML analysis indicates much greater variances than for
either of the previous combinations although this would be strongly influenced by the smaller
dataset available for this species/toxicant combination. Indeed, the dataset is probably too
small to draw meaningful conclusions about the extent and sources of vanability.
Lemna gibbalDCP
Valid data were generated by only one laboratory and so it is not possible to carry out a
meaningful analysis. However, as noted earlier, the repeatability of these tests was good with a
coefficient of variation of only 6.4%.
Table 3.7 Results of REML analysis
Statistic
L. minor and
potassium dichromate
L. minor and
3 ,5-dichlorophenol
L. gibba and
potassium dichromate
two labs four labs
excluded excluded
Numberof
laboratories included
in REIvIL analysis
14 12
14
8
Number of data
points used in REML
analysis
48 39
52
24
‘Consensus’ mean*
(log)
0.327 0.326
0.493
0.991
‘Consensus’ mean*
(mg 1’)
2.1 2.1
3.1
9.8
Standard error of
0.065 0.043
0.033
0.229
mean
Between-lab a 2
0.052 0.014
0.013
0.417
Within-lab a 2
0.02 1 0.023
0.005
0.009
Reproducibility a 2
0.073 0.037
0.018
0.427
Reproducibility a
0.271 0.191
0.135
0.654
* = geometric mean of the mean EC 50 obtained by each laboratory
R&D Technical Report EMA 003 27
-------
3.5.5 Possible sources of bias and variability
As the data included in the REML analysis were screened to exclude extrapolated EC 50 s and
tests where the control doubling time was excessive, these factors were unlikely to have
influenced the overall conclusions from this nng-test. The influence due to genetic vanabilaty
of the biological material used could not be determined (but see next paragraph) as no control
was imposed on the sources of plant material used by the participants, in line with the
requirements of the draft test Guideline.
As noted in Section 3.5.1, control doubling time appears to have been influenced by the level
of illumination and by temperature. Low illumination and/or low temperatures resulted in tests
failing to meet the 2.5-day maximum limit for doubling of control frond numbers. To
determine whether there was any relationship between doubling time and sensitivity to
toxicants, doubling time was plotted against EC 50 for the two largest data sets (L.
minor/potassium dichromate and L. minorIDCP). Upon initial inspection, Figure 3.2 suggests
there may be two populations of L. minor, differing in their sensitivities to potassium
dichromate. However, this is unlikely because data from the same laboratory (No. 26) are
located at the extremes of both groups. Conditions leading to low growth rates may reduce
sensitivity but any such relationship is highly variable. This is borne Out in Figure 3.4.
showing corresponding data for L. minor and DCP
. .
18
16
14
EC , 0 12
10
.. .
8
6 .
•.
4 .
2 • .4 ! ••%
00 05 15 2 25
Doubling time (days)
Figure 3.2 Effects of doubling time on sensitivity of Lemna minor to potassium
dichromate
R&D Technical Report EMA 003 28
-------
9
8
7
.
6
.
EC 50 : . .
:.r; ...:
0 I • I
0 05 1 15 2 25 3 35
Doubling time (days)
Figure 3.3 Effects of doubling time on sensitivity of Lemna minor to DCP
Analysis of the data for bias due to statistical method used by the participants in defining
EC 50 s showed no significant effect (p>O.05) due to the use of a model-based approach (Probit,
Logit, Hill and Weibull models) compared with other methods (linear regression).
The shallow dose response for potassium dichromate was apparent with both L. minor and L.
gibba. When examining frond numbers at the various concentrations used, it appeared that
most participants had adequately covered the effect concentration with their concentration
ranges. However, on calculating EC 50 s based on average specific growth rate a high
proportion found that they had missed the 50% effect concentration so that it became
necessary to extrapolate to estimate an EC 50 . This highlights a feature of growth-rate based
EC 50 s in that they tend to be higher than those based on biomass or frond area. It was apparent
that many participants had not anticipated such a difference and, as shown in Section 3.5.3, it
was undoubtedly exacerbated by the shallow dose-response with potassium dichromate. This
was the main reason for excluding toxicity data from REML analysis but it probably also led
to a source of variability in EC 50 values for this substance because they could not always be
estimated accurately. On balance, potassium dichromate was probably not the best choice of
test substance for use with Lemna spp. due to its particularly shallow dose-response.
3.5.6 Derivation of Quality Control criteria
Variability in the estimation of toxicity is inevitable and expected to some extent, even when
standardised methods are used (Whitehouse et al, 1996). However, if test data are to be used
for regulatory purposes it is very important that they should be free of excessive bias or
variability. This is because either of these can undermine the quality of decisions e.g. in
R&D Technical Report EMA 003 29
-------
hazard classification or risk assessment, and may lead to inappropriate risk management
decisions being made (Whitehouse et al, 1996).
The problem of bias and variability has been recogmsed for some considerable time in
analytical chemistry and this has given rise to the development of Quality Control (QC)
procedures and the definition of performance characteristics for a wide range of determinands.
However, little attempt has been made to develop QC procedures for ecotoxicity testing. Two
characteristics are important: the accuracy of measurements (how close the measurement is to
the ‘true’ value) and the precision of measurements (the degree of agreement between repeat
measurements). Determinations of toxicity pose a particular problem because the ‘true’
toxicity of a chemical to an organism can never actually be known. In practice, the ‘consensus’
mean obtained from repeat testing by many laboratories is the only realistic approach.
It is possible to define limits, i.e. QC criteria, within which we would expect a high proportion
(say, 95%) of estimated EC 50 values to fall, based on the performance achieved by laboratones
participating in the ring-test. Such criteria can then be used to assess:
(a) the extent of deviation from a ‘consensus’ mean - thereby providing a measure of the
accuracy of toxicity tests;
(b) the extent of agreement between repeat tests within a laboratory that might reasonably
be expected - thereby providing a measure of the precision with which tests are
performed.
The implication is that EC 50 values which fall outside defined limits for accuracy are subject
to an unacceptable level of error, perhaps due to a bias resulting from the source of plant
material, the way tests are performed, monitored or the data analysed. This would typically
lead to reduced accuracy but not necessarily reduced precision. Alternatively, EC 50 values
from repeat tests may span a large range, denoting poor precision. This would be indicative of
excessive random error. Indeed, both these situations may occur simultaneously. With the
exception of the estimation of the ‘consensus’ mean, this approach to QC for ecotoxicity
testing is similar to that routinely employed in chemical analysis; further details can be found
in Whitehouse et al (1996).
Derivation of QC criteria for accuracy
Control limits to indicate the maximum acceptable deviation from the ‘consensus’ mean can
be derived by applying the variance for test reproducibility ( rep 1ucibi1i ,), obtained from
REML analysis (Section 3.5.4) to the ‘consensus’ mean, with a factor denoting the level of
confidence we wish to stipulate. For 95% control limits, the following would apply:
95% control limits = ‘consensus’ mean + 1.96 x arepr ucibiIity
This assumes a normal distribution of the EC gj data which, as Figure 3.1 shows, does not
apply. For this reason log-transformed data were employed in the REM1. analysis and the
calculation of QC criteria. In Table 3.8, the values shown in bold are the calculated 95%
control limits for accuracy derived using the equation given above.
R&D Technical Report EMA 003 30
-------
Table 3.8 Control limits for accuracy in Lemna growth inhibition tests
Control limit
L.minorl potassium
dichromate
L. mznor/DCP
L gibbalpotassium
dichromate
(a) data from (b) data from
two labs four labs
excluded excluded
lower limit
-0.204 -0.049
0.229
-0.290
antilog lower limit
(mgF5
06 0.9
1 7
0.5
upper limit
0.858 0.700
0.758
2 271
antilog upper limit
(mgF’)
7.2 5.0
5.7
186 8
consensus’ mean
(mg F’)
2.1 2.1
3.1
9.8
Useful control limits can be calculated from the ring-test data for L. minor when potassium
dichromate and DCP are used as toxicants. The effect of excluding outlying data for tests with
potassium dichromate are clearly evident because more stringent criteria result when data
from four laboratories, which give rise to consistently different EC 50 estimates, are excluded
compared with just two, despite the overall reduction in the number of data points. A
pragmatic approach might be to regard the control limits derived when data from two
laboratories are excluded (0.6 - 7.2 mg I’) as the principal target for judging the accuracy of
L. minor toxicity tests with potassium dichromate. Laboratories should be capable of meeting
these targets on 19 out of 20 occasions. The control limits derived when data from four
laboratories are excluded produce more stringent criteria (0.9 - 5.0 mg F’). This second set
may therefore be regarded as an accuracy target representing ‘best practice’.
Control limits are in part a reflection of the deviation of EC 50 values from the ‘consensus’
mean but are also strongly influenced by the quantity of data available. Thus, very wide
control limits are calculated for L. gibba when potassium dichromate is used as a reference
toxicant because of the lack of data. The limits derived from this data-set are too wide to be of
any useful guidance to laboratories or to be of any regulatory use when assessing test data.
Derivation of QC criteria for precision
REML analysis of the ring-test data also provides an estimate of the underlying repeatability
of Lemna growth inhibition tests for two toxicants. Instead of comparing individual EC 50
estimates (as in the case of the control limits for accuracy), a series of tests with the reference
toxicant must be performed. The variance of the EC 50 values from the series of repeat tests
(S 2 ) is compared with the underlying variance that would be expected, based on the ring-test
R&D Technical Report EMA 003 31
-------
(rrepea abiIity). This comparison may be done using a test and an appropriate significance
level (a), as shown below:
c 2 2 2
I a repeatability
-------
4. CONCLUSIONS
In a review of test guideline requirements in 1995, an aquatic higher plant toxicity test
method was highlighted as a priority for development by the OECD. In 1997 a Steering
Group was established and research needs identified which would enable a test
Guideline to be drafted. These were concerned with the choice of test species, test
media, endpoints and data analysis and were addressed between 1997 and 1998 through
an experimental programme by members of the Steering Group. The results and
conclusions drawn from these studies are discussed. The draft method was then
subjected to international ring-testing in 1998-99.
2. Significant features of the test guideline are that it permits the use of two Lemna species,
L. minor and L. gibba. Essentially, it involves determination of the effects of substances
on the growth of a population of test plants compared to that in control vessels.The test
guideline details methods for expressing toxicity in terms of average specific growth
rate and also final biomass.
3. The main purpose of the OECD test Guideline is for hazard assessment of chemicals.
However, the Lemna growth inhibition test also has value for assessing the phytotoxicity
of effluents, leachates and receiving waters because it is less prone to interference by
properties of these samples than tests using unicellular algae. Specific guidance on the
testing of such samples is appended.
4. The ring-test was designed to assess key performance characteristics of the draft test
method, especially compliance with the critical quality criteria, repeatability of the
method within laboratories and reproducibility between laboratories. Thirty-seven
laboratories took part in the ring-test and more than half conducted the requested five or
more repeat tests. All participants used one or both of the specified reference toxicants
(potassium dichromate and 3,5-dichlorophenol).
5. More laboratories submitted data for L. minor than for L. gibba and potassium
dichromate was used more extensively in the ring-test than was 3,5-dichlorophenol.
Adherence to the draft Guideline in the nng-test was good. All participants supplied
EC 50 data calculated from average specific growth rates based on frond number, as
required by the Guideline, and most met the test acceptability criterion for control
doubling time. Those tests that failed this criterion were usually conducted under
conditions of low illumination and/or low temperature. Only 8% of participating
laboratories experienced problems with low levels of illumination and 11% with low
temperatures. All but one of these tests failed to meet the Guideline validity criterion for
control doubling time, thereby reinforcing the link between these parameters and growth
rate.
6. The commonest problem with respect to maintenance of water quality conditions was an
increase in pH of the control medium greater than that advised in the draft Guideline. A
total of 24% of participants experienced this in at least one repeat test and in at least one
case it was associated with contamination of the vessels with unicellular algae. This did
not seem to be a major cause of unacceptable growth rates, although it may have
R&D Technical Report EMA 003 33
-------
affected speclatlon of chromium and, may therefore have contnbuted to the variability
seen in test results.
7. The ring-test data show greater variability in tests when potassium dichromate was used
than when 3,5-dichiorophenol was used as the reference toxicant. Much of the
variability seen in tests with potassium dichromate could be attributed to the shallow
dose-response associated with this substance and consequent difficulties in determining
an EC 50 with accuracy; in a sigrnficant number of cases, it had been necessary for
laboratories to extrapolate beyond the range of test concentrations. This was not a
problem when 3,5-dichlorophenol was used as the reference toxicant.
8. Although the data suggest that L. minor was the more sensitive of the two species to
potassium dichromate, the 20 x AAP medium used with L. gibba contained over four
times more EDTA than the SIS medium used with L. minor. EDTA is known to
ameliorate metal toxicity and so it is not possible to draw any conclusions regarding the
relative sensitivities of these two species to this substance from the ring-test data.
9. Some participants expressed toxicity in terms of final biomass as well as average
specific growth rate. In tests with potassium dichromate and L. minor, the choice of
endpoint had a large influence on the resultant EC 50 values. The mean EbC5O was much
lower than the corresponding E C 5 o. With DCP the difference between means based on
final biomass and growth rate was much smaller, but again mean ErC5OS were higher
than mean EbC5os. This is consistent with previous research and the large difference
seen in tests with potassium dichromate is typical of a substance giving a shallow dose-
response.
10. The ring-test data were analysed to identify the contributions to variability in EC 50
values resulting from variability within- and between-laboratones. Those EC 50 values
from tests where the doubling time for control frond number was exceeded or where the
EC 50 was extrapolated beyond the concentration range tested were excluded from this
analysis. As expected, both sources of variability were greater in expenments using
potassium dichromate but the between-laboratory variability generally accounted for a
greater proportion of the observed variability than the variability found within-
laboratories. In tests where L. minor and potassium dichromate had been employed, two
laboratories were responsible for a high proportion of the between-laboratory variability,
suggesting their EC 50 estimates were rather extreme. The choice of statistical technique
used to estimate EC 50 values did not influence the resultant summary statistics to any
appreciable extent.
11. Using the estimates of variance within-laboratories and between-laboratories, Quality
Control criteria have been derived for accuracy (expressed in terms of the deviation
from the ‘consensus’ mean EC 50 ) and for precision (the range of EC 50 values generated
within a laboratory). For EC 50 values in tests using L. minor/potassium dichromate, the
recommended accuracy cntena are 0.6 - 7.2 mg F’ and for L. minorl 3,5-dichlorophenol,
the corresponding criteria are 1.7 - 5.7 mg F’. When used in conjunction with reference
toxicant data, these criteria may be used by laboratories to assess the accuracy and
precision with which they perform Lemna growth inhibition tests. They may also be
used by regulators when assessing the quality of data to be used for chemical
R&D Technical Report EMA 003 34
-------
classification and risk assessment purposes so that excessive bias and vanability can be
identified.
12. Quality Control critena cannot yet be established for L. gibba due to a lack of suitable
data. Of the two toxicants used, 3,5-dichlorophenol is to be preferred because of the
difficulties in interpreting the very shallow dose-response associated with the use of
potassium dichromate.
13. Further work should concentrate on tests using L. gibba and 3,5-dichlorophenol,
performed according to the OECD draft Guideline so that Quality Control cntena can be
derived for this species also.
R&D Technical Report EMA 003 35
-------
R&D Technical Report EMA 003 36
-------
REFERENCES
ASTM (1991). Conducting static toxicity tests with Lemna gibba. Guide E 1415-91. Annual
book of ASTM standards. Section 11 Water and environmental technology. Vol. 11.04.
ASTM, Philadelphia, U.S.
Bjornestad, E. and Petersen, 0. (1992). Pans Commission Ring-Test, Intercalibration and
comparison of tests and test laboratones for evaluation and approval of offshore chemicals
and drilling mud. Report of the Water Quality Institute, Horsholm, Denmark.
Bruce, R.D. and Versteeg, D.J. (1992). A statistical procedure for modelling continuous
toxicity data . Environmental Toxicology and Chemistry 11: 1485 - 1494.
Chapman, P., Crane, M., Wiles, J., Noppert, F. and Mclndoe, E.C. (1996). Improving the
quality of statistics in ecotoxicity tests. Ecotoxicology 5: 169 - 186.
Cole and Voskuil (1996). Canadian Journal of Botany 74: 220 - 230.
Correll, D.S. and Correll, H.B. (1972). Aquatic and Wetland Plants of Southwestern United
States. Environmental Protection Agency, Washington, DC.
Cowgill, U.M. and Milazzo, D.P. (1989). The cultivating and testing of two species of
Duckweed. Aquatic Toxicology and Hazard Assessment, Vol. 12. ASTM STP 1027. U. M.
Cowgill and L.R. Williams, Eds., American Society for Testing and Materials, Philadelphia.
Cowgill, U.M., Milazzo, D.P. and Landenberger, B.D., (1991). The sensitivity of Lemna gibba
G-3 and four clones of Lemna minor to eight common chemicals using a 7-day test. Research
Journal of the Water Pollution Control Federation 63:99 1 - 998.
Cowgill, U.M., Mjlazzo, D.P. and Landenberger, B.D. (1989). A comparison of the effect of
triclopyr triethylamine salt on two species of duckweed (Lemna) examined for a 7- and 14-day
test period. Water Research 23: 617 - 623.
USEPA (1996). OPPTS 850.4400 Aquatic Plant Toxicity Test Using Lemna spp., “Public
draft”. EPA 712-C-96-156. 8pp.
Haftka, J.J.H. and Weltje, L. (1999). Bioavailability and accumulation of lanthanides in
duckweeds (Lemnaceae). Paper presented at 9th Annual Meeting of SETAC-Europe, 25 - 29
May (1999), Leipzig, Germany.
Huebert, D.B. and Shay, J.M., (1993). Considerations in the assessment of toxicity testing
with duckweeds Environmental Toxicology and Chemistry 12: 481 - 483.
Kristensen, P., Torslov, J., Bjornestad, E., Peterson, 0., Samsoe-Peterson, L., Roghair, C. J.,
Wolters-Balk, M.A H., van der Wal, J., de Bruijn, J.H M., van de Guchte, C. and
Hooftman, R.N. (1998). Detailed review paper on aquatic testing methods for pesticides and
industrial chemicals. OECD Environmental Health and Safety Publications. Series on Testing
and Assessment. No. 11. Organisation for Economic Co-operation and Development,
Environment Directorate, Paris.
R&D Technical Report EMA 003 37
-------
Mazzeo, N., Dardano, B. and Marticorena, A. (1998). Interclonal variation in response to
simazine stress in Lemna gibba (Lemnaceae). Ecotoxicology 7: 151 - 160.
McCLure, T.W. and Alison, R.E. (1966). A Chemotaxonomic Study of Lemnaceae. American
Journal of Botany. 53: 849-860.
NAEP (1990). Toxicity testing method with Lemna minor. National Agency for
Environmental Protection, Sweden. Report 3755.
Newman, M.C. (1995). Quantitative methods in Aquatic Toxicology. Lewis Publishers, Boca
Raton, Fl.
Norberg-King, T. (1988). An interpolation estimate for chronic toxicity: The ICp approach.
Unpubh shed manuscnpt, U.S. Environmental Protection Agency, Environmental Research
Laboratory-Duluth, Duluth, MN.
Nyhoim, N. (1985). Response variable in algal growth inhibition tests - biomass or growth
rate? Water Research 19: 273 - 279.
Nyholm, N. (1990). Expression of results from growth inhibition tests with algae. Archives of
Environmental Contamination and Toxicology 19: 518 - 522.
Payne, R.W., Lane, P.W., Digby, P.G.N., Harding, S.A., Leech, P.K., Morgan, G.W..
Todd, A.D., Thompson, R., Tunnicliffe, W.G., Weiham, S.J. and White, P.R. (1993). Genstat
5 Release 3 Reference manual. Clarendon Press, Oxford, UK.
Sims, I., van Dijk, P., Gamble, J., Grandy, N. and Huet, M.C. (1997). Report of the final ring
test of the Daphnia magna reproduction test. OECD Environmental Health and Saftry
Publications. Series on Testing and Assessment, No. 6. Organisation for Economic Co-
operation and Development, Environment Directorate. Paris, France.
Vasseur et al (1993). American Journal of Botany 80: 974 - 979.
Wang, W. (1987). Chromate ion as a reference toxicant for aquatic phytotoxicity tests.
Environmental Toxicology and Chemistry 6: 953 - 960.
Wang, W. (1990). Literature review on duckweed toxicity testing. Environmental Research
52: 7 - 22.
Wang, W. (1991). Literature review on higher plants for toxicity testing. Water, Air and Soil
Pollution 59: 381 - 400.
Wang, W. and Freemark, K. (1995). The use of plants for environmental monitoring and
assessment. Ecotoxicology and Environmental Safety 30: 289 - 301.
Whitehouse, P., Crane, M., Redshaw, C.J. and Turner, C. (1996). Aquatic toxicity tests for the
control of effluent discharges in the UK - the influence of test precision Ecotoxzcologv 5:
155 - 168.
R&D Technical Report EMA 003 38
-------
APPENDICES
APPENDIX A NAMES AND AFFILIATIONS OF PARTICIPANTS IN THE RING TEST
OF THE OECD LEMNA GROWTH TEST
APPENDIX B DRAFT OF THE OECD LEMNA GROWTH INHIBITION TEST
(JUNE 1999)
APPENDIX C GUIDANCE TO RING-TEST PARTICIPANTS AND PRO-FORMA
DATASHEETS
APPENDIX D EC 50 VALUES ESTIMATED FROM RING-TEST
APPENDIX E GUIDANCE ON THE TESTING OF EFFLUENTS, LEACHATES AND
RECEIVING WATERS IN THE LEMNA GROWTH INHIBITION TEST
R&D Technical Report EMA 003 39
-------
R&D Technical Report EMA 003 40
-------
APPENDIX A NAMES AND AFFILIATIONS OF
PARTICIPANTS IN THE RING TEST OF THE
OECD LEMNA GROWTH TEST
Name Company Address
Mr M Brixham Environmental Bnxham
Hutchings Laboratory Devon,
TQ5 8BA
Dr Anette AB Thalassa P0 Box 1632
Pettersson S-75 1 46 Uppsala
Sweden
Ms Melanie Environment Agency 4 The Meadows
Dixon Waterberry Dnve
Waterlooville, Hants
P07 7XX
Ian Sims WRc plc Henley Road
Medmenham
Marlow, Bucks, SL7 21-ID
Mr J A ABC Laboratories 7200 E ABC Lane
Kranzfelder Missouri Columbia
M065 202
USA
Dr Anne Rhone-Poulenc Agro BP 153
McElligott 355 Rue Dostoievski
06903 Sophia-Antipolis Cedex
FRANCE
Dr Rinus NOTOX B V Hambakenwetering 3
Bogers P0 Box 3476
5203 DL ‘s Hertogenbosch,
The Netherlands
Dr Tim Wildlife International Ltd 8598 Commerce Drive
Springer Easton
MD 21601,USA
Eija Schultz Finnish Environment Hakuninmaantie 4-6
Institute Laboratory FIN-40300 Helsinkj
FINLAND
R&D Technical Report EMA 003 41
-------
Dr M Latif Laboratory of Veterinaerplatz
Ecotoxicology Josef Baumann-Gasse 1
Institute of Hydrobiology A-1210 Vienna AUSTRIA
University of Veterinary
Medicine Vienna
Ms Mary Saskatchewan Research 15 Innovation Boulevard
Moody Council Saskatoon
Saskatchewan
SiN 2X8 CANADA
Dr J Handley SafePharm Laboratories P0 Box 45
Ltd Derby
DEl 2BT
Dr Bernard Université de Campus scientifique
Clement Savoie/ESIGEC F-73376 Le Bourget du Lac Cedex
FRANCE
M. J F Ferard Centre des Sciences de I rue des Recollets
l’Environnement 5700 Metz
FRANCE
Dr i-I Elf Atochem SA 95 rue Danton
Thiébaud Centre d’application de BP 108
Levallois 92303 Levallois-Perret
Cedex FRANCE
M. P Pandard Laboratoire Parc Technologique ALATA
d’Ecotoxicologie BP 2
INERIS 60550 Verneuil en Halatte
FRANCE
Dr R Baudo Istituto Nazionale di Largo V Tonolli, 50-52
Idrobiologia - CNR 28922 Verbania Pallanza (NO)
ITALY
Dr I Kiss University of Veszprém Egyetem utca 10
H-8200 Vezprém
HUNGARY
Dr P Gnemi Istituto di Ricerche Via Ribes, 1
Biomediche Colleretto Giacosa 10010 (TORINO)
LCG Bioscience ITALY
Dr D Flatman Huntingdon Life Sciences Huntingdon Research Centre
P0 Box 2
Huntingdon
Cambs PEI86ES
R&D Technical Report EMA 003 42
-------
Dr Sung Kyu Korea Research Institute 100 Jang-dong Yusong-ku
Lee of Chemical Technology P0 Box 107, Taejon. #305-606
(KRICT) Env.Toxicology Republic of Korea
Laboratory
Dr T Gries Springborn Laboratories Seestrasse 21
(Europe) SA CH-9326 Horn
Switzerland
Dr Ulnch RCC Umweltchemie AG Zelgliweg 1
Memrnert CH-4452 Itigen/Basal
Switzerland
Dr R Grade Novartis Crop protection R-1066.P.06
AG CH-4002 Basel
Switzerland
Dr B Olah Toxicological Research H-820 1 Veszprëm
Centre POB 348
HUNGARY
Dr S National Institute for 16-2 Onogawa
Hatakeyama Environmental Studies Tsukuba 305-0053
Ibaraki
JAPAN
Dr M Yasuno University of Shiga 2500 Yasaka-cho
Prefecture Hikoneshi
Shiga prefecture 522
JAPAN
Mr T Nozaka Kurume Research 19-14 Chuo-machi
Laboratories Kurume, Fukuoka 830-0023
Chemical Biotesting JAPAN
Center
Chemicals Inspection &
Testing Institute (C1TI)
Lise Sams e- VKI Agern Allë 11
Petersen DK-2970 Hørsholm
DENMARK
Mr A Sjolin Toxjcon AB Rosenhallsvagen 23
S-261 92 Landskrona
SWEDEN
R&D Technical Report EMA 003 43
-------
Dr W Z Pluta Umweltbundesamt, Schichauweg 58
WaBoLu D - 12307 Berlin
GERMANY
Dr W Kopf Bayerisches Landesamt fur Kaulbachstr.37
Wasserwirtschaf D-80539 Munchen
Institute fur Germany
Wasserforschung
Dr Nusch Ruhr River Association Chem. & Biol. Laboratory
P0 Box 103292
48032 ESSEN
Germany
Dr M Bayer AG Landwirtschaftszentrum Monheim
Dorgerloh PF-E/OE, Building 6620
D-5 1368 Leverkusen-Bayerwerk, FRG
GERMANY
Dr R UFZ Centre for Permoserstr. 15
Altenburger Environmental Research D-043 18 Leipzig
GERMANY
Dr. M RWTH Aachen Womnger Weg 1
Hammers- Biologie V 52074 Aachen
Wirtz GERMANY
Ms S Leva DuPont Agricultural BMP-15 Phips Mill
Enterpnse Wilmington
Delaware 19880-0015
USA
R&D Technical Report EMA 003 44
-------
APPENDIX B DRAFT OECD LEMNA GROWTH
INHIBITION TEST (JUNE 1999)
INTRODUCTION
This test Guideline is designed to assess the toxicity of substances to freshwater aquatic
plants of the genus Lemna (duckweeds). It is based on existing guidelines and standards
published by the American Society for Testing and Matenals (I), the United States
Environmental Protection Agency (2), Association Française de Norrnalisation (3) and
the Swedish Standards (4), but includes modifications of those methods to reflect recent
research and consultation on a number of key issues.
INITIAL CONSIDERATIONS
2. This Guideline describes toxicity testing using Lemna gibba and Lemna minor, both of
which have been extensively studied and are the subject of the standards referred to
above. The taxonomy of Lemna spp. is difficult, being complicated by the existence of a
wide range of phenotypes. Although genetic variability in the response to toxins can
occur with Lemna, there is currently insufficient data on this source of variability to
recommend a specific clone for use with this Guideline. Short descriptions of duckweed
species that have been used for toxicity testing are given in Annex 1.
3. Details of testing with renewal (semi-static) and without renewal (static) of the test
solution are described. The semi-static method is recommended for substances which
are rapidly lost from solution as a result of e.g. volatilisation, photodegradation.
precipitation or biodegradation.
4. Definitions used are given in Annex 2.
PRINCIPLE OF THE TEST
5. Plants of the genus Lemna are allowed to grow as monocultures in different
concentrations of the test substance over a period of 7 days. The objective of the test is
to quantify substance-related effects on vegetative growth over this period based on
assessments of frond number, and also on an assessment of dry weight, fresh weight or
total frond area.
6. To quantify substance related effects, growth in the test solutions is compared with that
in the controls and the concentration bringing about a specified (e.g. 50%) inhibition of
growth is determined and expressed as the EC e.g. EC 50 . In addition, the lowest
observed effect concentration (LOEC) and the no observed effect concentration (NOEC)
may be determined.
INFORMATION ON THE TEST SUBSTANCE
7. An adequately sensitive analytical method for the quantification of the substance in the
test medium should be available.
R&D Technical Report EMA 003 45
-------
8. Useful information on the test substance includes its’ structural formula, purity, water
solubility, stability in water and light, pKa, ICow, vapour pressure and biodegradability.
Water solubility and vapour pressure can be used to calculate Henry’s Law constant,
which will indicate if significant losses of the test substance during the test period are
likely. This will help indicate whether steps to control such losses should be taken.
Where information on the solubility and stability of the test substance is uncertain, it is
recommended that these are assessed under the conditions of the test, i.e. growth
medium, temperature, lighting regime to be used in the test. -
9. When pH control of the test medium is particularly important, e.g. when testing metals
or substances which are hydrolytically unstable, the addition of MOPS buffer to the
growth medium used for L. minor is recommended (Paragraph 20).
VALIDITY OF THE TEST
10. For the test to be valid, the doubling time of frond number in the control must be less
than 2.5d (60 h), corresponding to approximately an 8-fold increase in 7 days. Using the
media and test conditions described in this Guideline, this criterion can be attained using
a static test regime. Calculation of the doubling time is shown in Paragraph 44.
DESCRIPTION OF THE METHOD
Test facilities
11. All equipment in contact with the test solutions should be made of glass or other
chemically inert material. Glassware used for culturing and testing purposes should be
chemically clean and sterile.
12. The cultures and test vessels should be maintained in an environmental growth chamber,
temperature-controlled illuminated incubator or room with constant illumination and
temperature (see paragraphs 32-33).
13. The test vessels should be wide enough for the fronds in the control vessels to grow
without overlapping at the end of the test. It does not matter if the roots touch the
bottoms of the test vessels, but a minimum depth of 20 mm and minimum volume of
100 ml in each test vessel is advised. The choice of test vessels is not critical as long as
these requirements are met. Erlenmeyer flasks, crystallising dishes or glass petri dishes
of appropriate dimensions have all proved suitable. However, covers should be provided
to minimise evaporation and accidental contamination.
Test plants
14. The organism used for this test is either Lemna gibba or Lemna minor. Plant material
may be obtained from a culture collection, another laboratory or from the field. If
collected from the field, plants should be maintained in culture for a minimum of eight
weeks prior to use. If obtained from another laboratory or a culture collection they
should be similarly maintained for a minimum of three weeks. The source of plant
material and the species and clone (if known) used for testing should always be
reported.
R&D Technical Report EMA 003 46
-------
15. Monocultures visibly free from contamination by other organisms such as algae and
protozoa should be used.
16. Healthy L. minor Consists of colonies comprising between 2 and 5 fronds whilst healthy
colonies of L. gibba may contain up to 7 fronds. Throughout this Guideline the number
of fronds rather than the number of colonies is used to denote the quantity of plant
material.
Cultivation
17. To reduce the frequency of culture maintenance, e.g. when no Lemna tests are planned
for a period, plants can be held under reduced illumination and temperature (4°C).
Details of culturing are given in Annex 3. Obvious signs of contamination by algae or
other organisms will require surface stenlisation of a sub-sample of Lenina fronds,
followed by transfer to fresh medium (see Annex 3). In this eventuality the
contaminated culture should be discarded.
18. The quality and uniformity of the plants used for the test will have a significant
influence on the outcome of the test and should therefore be selected with care Young,
rapidly growing colonies without visible lesions should be used. Good quality cultures
are indicated by a high incidence of colonies comprising at least two fronds. A large
number of single fronds is indicative of environmental stress, e.g. nutnent limitation,
and plant material from such cultures should not be used for testing
At least 7 days before testing, sufficient colonies are transferred aseptically into fresh
sterile medium and cultured for 7 - 10 days under the conditions of the test.
Media
19. Different media are recommended for L. minor and L. gibba., as described below.
20. Growth medium for L. minor:
A modification of the Swedish standard Lemna (SIS) growth medium is recommended
for culturing and testing with L. minor. The composition of this medium is given in
Annex 4. Stock solutions are prepared in distilled water, according to the compositions
given in Annex 4, using reagent grade chemicals. Stock solution VII (MOPS buffer: 4-
morpholinepropane suiphonic acid, CAS No: 1132-61-2) is only required for certain test
substances (see Paragraph 9). Stock solutions I - V are stenuised by autoclaving (120°C,
15 minutes) or by membrane filtration (approximately 0.2 j.tm pore size). Stock VI (and
optional VII) are sterilised by membrane filtration only; these should not be autoclaved.
Sterile stock solutions should be stored under cool and dark conditions. Stocks I - V
should be discarded after 6 months whilst stocks VI (and optional VII) have a shelf-life
of 1 month.
To prepare 1-I of SIS medium, the following are added to 900 ml of distilled water:
10 ml of stock solution I
5 ml of stock solution II
5 ml of stock solution III
R&D Technical Report EMA 003 47
-------
5 ml of stock solution IV
1 ml of stock solution V
5 ml of stock solution VI
If buffer is required for additional pH control. 1 ml of stock solution VII (optional) is
also added. This stock solution should be pH adjusted to 6.5 ± 0.2 with a saturated
solution of sodium hydroxide before adjusting to the final volume.
The pH of the medium is adjusted to 6.5 ± 0.2 with either 0.1 or 1M HCI or NaOH, and
the volume adjusted to 1-I with distilled water.
21. Growth medium for L. gibba:
20X - AAP medium is recommended for L. gibba. To prepare 20X -AAP medium, the
two stock solutions listed in Annex 4 are prepared using reagent-grade chemicals and
20 ml of each added to approximately 850 ml distilled water. The pH is adjusted to
7.5+0.1 with either 0.1 or 1M HC] or NaOH. and the volume adjusted to 1-I with
distilled water. The medium is then filtered through a 0 2 .tm (approx.) membrane filter
into a sterile container.
22. The prepared media have a shelf-life of approximately 6-8 weeks for use as a stock
culture medium, if stored in the dark to preclude possible (unknown) photochemical
changes. Growth medium intended for testing should be prepared 1-2 days before use to
allow the pH to stabilise. It is advisable to check the pH of the medium pnor to use. If
the pH lies outside the specified range, it may be readjusted by the addition of NaOH or
HCI as described above.
Preparation of test solutions
23. Test solutions of the chosen concentrations are usually prepared by dilution of a stock
solution. Stock solutions are normally prepared by dissolving the substance in the test
medium, although for some substances, e.g. pesticides, a foliar application (spray) of the
test substance directly onto the fronds may be applicable if this is considered to be the
most likely exposure scenario (5, 6).
24. The solubility of the test substance should not normally be exceeded by any test
substance concentration. For poorly soluble substances it may be necessary to use an
organic solvent as an aid to realising the substance’s solubility in water. The use of
organic solvents or dispersants may be required in some cases in order to produce a
suitably concentrated stock solution, but every effort should be made to avoid the use of
such materials. In some cases, it may be possible to achieve only a stable dispersion of
the test substances with the aid of dispersants. Commonly used solvents which do not
cause phytotoxicity at concentrations up to 100 mg F’ include acetone,
dimethylforrnamide and triethyleneglycol. If a solvent or dispersant is used, its’ final
concentration should be reported and kept to a minimum ( 0 01%, i.e. 100 mg/I), and
all treatments should contain the same concentration of solvent or dispersant. There
should be no phytotoxicity resulting from the use of auxiliary solvents or dispersants.
R&D Technical Report EMA 003 48
-------
PROCEDURE
Test and control groups
25. Pnor knowledge of the toxicity of the test substance to Lemna, e.g. from a range-finding
test, will help in selecting suitable test concentrations.
26. Normally, there should be at least five test concentrations arranged in a geometric series.
with a separation factor preferably not exceeding 3.2. Justification should be provided if
fewer than five concentrations are used. At least three replicates should be used at each
test concentration.
27. In setting the range of test concentrations, the following should be borne in mind:
• If the aim is to estimate the LOECINOEC, the lowest test concentration should be
low enough so that growth is not significantly less than that in the control In
addition, the highest test concentration should be high enough so that growth at that
concentration is significantly lower than that in the control. If this is not the case, the
test will have to be repeated using a different concentration range (unless the highest
concentration is at the limit of solubility).
• If the EC for growth inhibition is to be estimated, it is advisable that sufficient
concentrations are used to define the EC with an appropriate level of confidence. If
the EC 50 is to be estimated, the highest test concentration should be greater than the
EC 50 . If this is not the case, it will still be possible to estimate the EC 50 but the
associated confidence interval will be very wide and it may not be possible to
satisfactorily assess the adequacy of the fitted model.
28. Every test should include controls consisting of the same nutrient medium, number of
fronds, environmental conditions and procedures as the test vessels but without the test
substance. If an auxiliary solvent or dispersant is used, an additional control treatment
with the solvent/dispersant present at the same concentration as that in the vessels with
the test substance should be included. The number of replicate control vessels should be
the same as that used for the test concentrations.
R&D Technical Report EMA 003 49
-------
I xposure
29. Colonies consisting of 2-4 fronds are transferred from the inoculum culture under
aseptic conditions. Each test vessel should contain a total of 9-12 fronds. The number of
colonies and fronds should be identical in each test vessel. Research has shown that
when three replicates/ per treatment, each containing 9-12 fronds are used, differences in
growth of approximately 10-15% may be discerned between treatments under the
conditions described in this Guideline.
30. A randomised location of the test vessels should be used to minimise the possible
influence of spatial differences in light intensity or temperature in the incubator. A
blocked design or repositioning of the vessels at random on the occasions when
observations are made, or more frequently, is recommended.
31. If a preliminary stability test shows that the test substance concentration cannot be
maintained (i.e. the measured concentration falls below 80% of the measured initial
concentration) over the test duration (7 days), a semi-static test regime should be used.
In this case, the colonies should be transferred to new test solutions on at least two
occasions during the test (e.g. days 3 and 5). The frequency of medium renewal will
depend on the stability of the test substance; more frequent renewals may be necessary
to maintain concentrations of highly unstable or volatile substances. In some
circumstances, a flow-through procedure may be required (7).
Test conditions and duration
32. Continuous wann or cool white fluorescent lighting (400 to 700 nm) should be used to
provide a light intensity in the range of 85-125 iE m 2 s_i (6500 - 10000 lux), as
measured at the surface of the test solution. Any differences in light Intensity over the
test area should not exceed ± 15 %. The method of light measurement, in particular the
type of receptor (collector), will affect the measured value. Sphencal receptors (which
respond to light from all angles above and below the plane of measurement), and
“cosine” receptors (which respond to light from all angles above the plane of
measurement) are preferred to unidirectional receptors and will give higher readings for
a multi-point light source of the type described here.
33. The temperature in the test vessels should be 24 ± 2 °C.
34. The pH of the control medium should not increase by more than 1.5 units during the
test. When testing metals a narrower range of pH values is advisable (see Para. 9).
35. The test is terminated 7 days after the plants were inoculated into the test vessels.
Observations
36. At the start of the test, frond and colony numbers in the test vessels are recorded. Frond
numbers and the appearance of the colonies must be observed at least every 3 days, i.e.
on at least 2 occasions during the 7 day exposure period, and at test termination (day 7).
Thus, in total, four measurements of frond number are made dunng the test. Any change
in plant development, frond size, appearance, necrosis or chlorosis should be noted as
R&D Technical Report EMA 003 50
-------
well as additional observations of root length, of the test media (e.g. presence of
undissolved matenal) or other abnormalities.
37. In addition to determinations of frond number during the test, effects of the test
substance on final biomass are also assessed, based on one of the following parameters:
(a) dry weight,
(b) fresh weight or
(c) total frond area
The chosen parameter should be determined at the start of the test on a sample of fronds
identical to that used to inoculate the test vessels and also at the end of the test on the
plant matenal in every test and control vessel.
Dry weight, fresh weight and total frond area may be determined as follows:
(a) The dry weight of colonies from each test vessel is determined by collecting plants
from the test vessels, blotting them dry and drying at 60°C to a constant weight.
Any root fragments should be included. The dry weight should be expressed with
an accuracy of at least 0.1 mg.
(b) Colonies are transferred, without damage, to a pre-weighted polystyrene (or other
inert material) tube (A) with small (1 mm) holes in its rounded bottom. Tube A is
then placed on top of a hollow cylindrical tube (B) within a centrifuge tube (C).
Tube C is closed with a cap, then the whole is centrifuged at 3000 rpm for 10
minutes at room temperature. Then tube A, containing the now dned colonies, is
re-weighed and fresh weight calculated by subtracting the weight of the empty
tube A.
(c) The total frond area of colonies may be determined by image analysis. A silhouette
of the test vessel and plants can be captured using a video camera (i.e. by placing
the vessel on a light box) and the resulting image digitised. By calibration with flat
shapes of known area, the total frond area in a test vessel may then be determined.
Care should be taken to exclude interference caused by the rim of the test vessel.
An alternative but more laborious approach is to photocopy test vessels and plants,
cut out the resulting silhouette of colonies and determine their area using a leaf
area analyser or graph paper.
Frequency of analytical determinations and measurements
38. If a static test design is used, the pH of each treatment should be measured at each
observation (i.e. at the beginning and end and at least 2 occasions dunng the test). If a
semi-static test design is used, the pH should be measured in each batch of ‘fresh’ test
solution pnor to each renewal and also in the corresponding ‘spent’ solutions.
39. Light intensity at points the same distance from the light source as the Lenzna fronds
should be measure at least once during the test and the temperature in a surrogate vessel
Should be recorded at least daily.
R&D Technical Report EMA 003 51
-------
40. In semi-static tests where the concentration of the test substance is expected to remain
within ± 20% of the nominal concentration (i.e. within the range 80 - 120%) it is
recommended that, as a minimum, the highest and lowest test concentrations be
analysed when freshly prepared and at the time of renewal (where appropnate) on one
occasion (i.e. analysis should be made on a sample from the same solution, when freshly
prepared and at renewal).
For tests where the concentration of the test substance is not expected to remain within
± 20% of the nominal, it is necessary to analyse all test concentrations, when freshly
prepared and at renewal. However, for those tests where the measured initial
concentration of the test substance is not within ± 20% of nominal but where sufficient
evidence can be provided to show that initial concentrations are repeatable and stable
(io.e. within the range 80 - 120% of initial concentrations), chemical determinations
could be reduced to the highest and lowest test concentrations In all cases,
determination of test substance concentrations prior to renewal need only be performed
on one replicate vessel at each test concentration (or the contents of the vessels pooled
by replicate).
41. If a flow-through test is used, a similar sampling regime to that described for semi-static
tests is appropriate, but measurement of ‘spent’ solutions is not appropriate in this case.
In this type of test, the flow-rate of diluent and test substance should be checked daily.
42. If there is evidence that the concentration of the substance being tested has been
satisfactorily maintained within ± 20 % of the nominal or measured initial concentration
throughout the test, analysis of the results can be based on nominal or measured initial
values. If the deviation from the nominal or measured initial concentration is greater
than ± 20 %, analysis of the results should be based on the time-weighted mean (see
Annex 5).
LIMIT TEST
43. Under some circumstances, e.g. when a preliminary test indicates that the test substance
is non-toxic up to 100 mg 1 1, a limit test, involving a comparison of responses in a
control group and one treatment group (100 mg ri), may be undertaken. It is strongly
recommended that this is supported by analysis of the exposure concentration. All
previously described test conditions and validity criteria apply to a limit test, with the
exception that the number of treatment replicates should be doubled. Growth in the
control and treatment group may be analysed using a statistical test to compare means,
e.g. a Student’s t-test.
DATA ANALYSIS AND REPORTING
Treatment of results
44. To determine performance against the validity criterion of doubling time (Paragraph 10),
the doubling time is calculated from the doubling time (Td) of the fronds in the control
vessels using the following formula:
Td = ln 2 /ji.
R&D Technical Report EMA 003 52
-------
where t is the average specific growth determined as described in paragraph 47.
45. The purpose of the test is to determine the effects of the test substance on the vegetative
growth of Lemna. This Guideline descnbes methods for determining the effects of a test
substance on:
(a) ‘Average specific growth rate’. This is calculated on the basis of changes in frond
number determined dunng the course of the 7 day exposure penod in controls and
in each treatment group. It is sometimes referred to as ‘relative growth rate’ (8).
(b) Area under the growth curve. This is also calculated on the basis of frond number
in controls and each treatment group but integrates frond number and exposure
penod.
(c) Final biomass. This is calculated on the basis of changes in dry weight, fresh
weight or total frond area of fronds in the controls and in each treatment group
during the test.
46. It is recommended that toxicity estimates are based on at least (a) average specific
growth rate or (b) area under the growth curve, along with an estimate of toxicity based
on (c) final biomass using one other growth parameter (dry weight, fresh weight or total
frond area).
Note: If the doubling time cntena are met (Paragraph 44) but there is evidence that
growth in the controls is not exponential, then it is preferable to base estimates of
toxicity on area under the curve (b) rather than average specific growth rate (a).
The number of fronds as well as any other recorded parameter of growth, i.e. dry weight.
fresh weight or frond area, are tabulated together with the concentrations of the test
substance for each measurement occasion. Subsequent data analysis e.g. to estimate a
LOEC, NOEC or EC should be based on the values for the individual replicates and not
calculated means for each treatment group.
47. (a) Average specific growth rate :
To determine the average specific growth rate for each test concentration and controls,
frond numbers for each replicate in the controls and each treatment at each observation
time are plotted against time as a semilogarithmic graph to produce growth curves. If
there is exponential growth in the controls (or growth is close to an exponential pattern),
the average specific growth rate ( .t) can be den ved from the slope of the regression line
in a plot of In N versus time. The slope (p) can be calculated from:
= ln(N 1 ) - In (N 0
t
where:
- N 1 is the number of fronds observed in the test or control vessel at time t;
- N 0 is the number of fronds observed in the test or control vessel at the
beginning of the test;
R&D Technical Report EMA 003 53
-------
- t is time;
- .t is the average specific growth rate.
Percent inhibition of growth rate, ‘r, may then be calculated for each test concentration
according to the following formula:
%Ir= ( c ] x 100
C , ’
where:
- %Ir is the percent inhibition in average specific growth rate:
- C,’ is the mean value for in the control:
- T,’ is the mean value for in the treatment group
48. (b) Area under the curve :
The area under the growth curves can be calculated for each control and treatment
replicate according to the following equation:
A = jJ xti + N -2Nox(t 2 -t 1 ) + N a ox(tn-tn.i)
2 2 2
where:
- A is the area under the growth curve;
- N 0 is the number of fronds observed in the test or control vessel at the start of
the test (to);
- N 1 is the number of fronds observed in the test or control vessel at time t,;
- N is the number of fronds observed in the test or control vessel at time t ;
- t, is the time of first measurement after beginning of test:
- t is the time of the measurement after beginning of test.
The area should be calculated for the entire test period, or a rationale for selecting only a
portion of the growth curve provided. For each test concentration and control, a mean
area is calculated, with variance estimates.
Percent inhibition of area under the curve, ‘a. may be calculated for each test
concentration according to the following formula:
%1a I Ii i X 100
CA
where:
- CA is the mean value for area under the curve in the control group;
- TA is the mean value for area under the curve in the treatment group.
R&D Technical Report EMA 003 54
-------
49. (c) Final biomass :
Effects on final biomass may be determined on the basis of either dry weight, fresh
weight or surface area of the fronds present in each test vessel at the start and the end of
the test. The mean percent inhibition in final biomass (%Ib) may be calculated for
controls and each treatment group as follows:
%Ib=i -b) x 100
b
where:
- %I is the percent reduction in biomass;
- bc is the final biomass minus the starting biomass for the control group:
- b is the final biomass minus the starting biomass in the treatment group.
50. Concentration-growth curves relating mean percentage inhibition of the growth
parameters (Jr. L. or ‘b ’ calculated as shown in Paragraphs 47-49) and the log
concentration of the test substance should be plotted.
Estimation of toxicity
51. To estimate the LOEC, and hence the NOEC, ANOVA is used to calculate the mean
average specific growth rate, area under the curve or final biomass and pooled residual
standard deviation across replicates for each test concentration. The resulting mean for
each test concentration is then compared with the control mean using an appropriate
multiple comparison method e.g. Dunnett’s or Williams’ tests.
52. A test for normality of the data is advised e.g. by calculating the Shapiro-Wilk’s statistic
and, if the replicate data reveal a normally distributed error structure, a test for
homogeneity of variances across treatment groups is recommended e.g. using Bartlett’s
or Levene’s test. If the variances are not homogeneous, it may be necessary to carry out
a transformation of the data prior to carrying out ANOVA. The log transformation is
recommended for average specific growth rate and area under the curve, and the square
root transformation for final biomass. Non-parametric analysis, e.g. Wilcoxon Rank
Sum Test, is only required when the assumptions of normality and homogeneity of
variances are severely violated.
53. If a one-tailed test to compare means is used, rejection of the null hypothesis implies
that the mean of the treatment group is less than the mean of the control group. If a two-
tailed test is used, rejection of the null hypothesis implies that the mean of the treatment
group could be either less (inhibition) or more (stimulation) than the mean of the control
group. The type of means comparison test should therefore be described and also
whether a one-tailed or two-tailed procedure was employed.
In addition, the size of the effect which can be detected using ANOVA (the least
significant difference) must be reported.
54. Estimates of EC (e.g. EC 50 ) should be based on frond number and at least one other
growth parameter i.e. final dry weight, fresh weight or total frond area.
R&D Technical Report EMA 003 55
-------
An EC can be obtained by non-linear regression analysis of the concentration-response
curve using an appropriate mathematical function. Suitable functions, which are fitted to
the replicate level % inhibition data, are:
(a) the logistic curve;
(b) the cumulative normal model (9);
(c) linear interpolation with bootstrapping (ICp) (10).
The formulae for these models can be solved for any percentage x e.g. 50%, 20%, 10%.
The linear interpolation model is most suitable when the responses to the test substance
are not monatomic.
All three models assume that 0% inhibition corresponds to a replicate value equal to the
control mean and that 100% inhibition corresponds to a replicate value of zero.
Consequently, including treatment levels that have negative values for any of the
endpoints (i.e. hormesis) may distort the results for either model. This can result in an
underestimate of EC values, particularly with the cumulative normal model. If hormesis
is substantial, resulting in a poor fit with these models, or a substantial portion of the
dataset must be left Out lii order to achieve a reasonable fit, then a hormesis model may
need to be considered (11).
Variance within treatments may not be constant across concentrations, in which case
consideration may need to be given to a weighted analysis in which greater weight is
accorded to treatments having less vanability. Where possible, two-sided 95%
confidence limits around the estimated EC should be quoted. Therefore, the fitted curve
should be parametensed so that the EC of Interest and its standard error can be
estimated directly. Goodness of fit should also be assessed, either graphically or by
dividing the residual sum of squares into ‘lack of fit’ and ‘pure error components’ and
performing a significance test for lack of fit.
R&D Technical Report EMA 003 56
-------
Reporting
55. The test report should include the following information:
Test substance:
— physical nature and physicochemical properties, including water solubility limit and
percent purity;
— identification data, e.g. CAS No.
Test species:
— scientific name and clone (if known) and source.
Test conditions:
— test procedure used (static, semi-static or flow-through);
— test duration;
— test medium;
— description of the experimental design test vessels and covers, solution volumes.
number of colonies and fronds per test vessel at the beginning of the test;
— test concentrations and number of replicates per concentration;
— methods of preparation of stock and test solutions including the use of any solvents
or
dispersants;
— temperature during the test;
— light source and light intensity;
— pH values;
— test substance concentrations and the method of analysis with appropriate quality
assessment data (validation studies, standard deviations or confidence limits of
analyses);
— methods for determination of frond number and other parameters, e.g. dry weight.
fresh weight or frond area;
— all deviations from this guideline.
Results:
— raw data: number of fronds and other parameters in each test and control vessel at
each
observation and occasion of analysis;
— means and standard deviations for each measured growth parameter;
— growth curves for each concentration;
— doubling time in the control based on the frond number;
— estimates of toxic endpoints e.g. EC 50 , LOEC and NOEC, and the statistical methods
used for their determination;
— any stimulation of growth found in any treatment;
— any change in colony development or appearance as well as observations of test
solutions;
Discussion of the results, including any influence on the outcome of the test resulting
from deviations from this Guideline.
R&D Technical Report EMA 003 57
-------
R&D Technical Report EMA 003 58
-------
LITERATURE
(1) American Society for Testing and Materials (1991). Standard Guide for Conducting
Static Toxicity Test With Lemna gibba G3. E 1415-91. lOpp.
(2) United States Environmental Protection Agency (1996). OPPTS 850.4400 Aquatic Plant
Toxicity Test Using Lemna spp., “Public draft”. EPA 712-C-96-156. 8pp.
(3) Association Francaise de Normalisation (1996). XP T 90-337: Determination de
l’inhibition de croissance de Lemna minor. lOpp.
(4) Swedish Standards Institute (1995). Water quality - Determination of growth inhibition
(7-d) Lemna minor, duckweed. SS 02 82 13. lSpp. (In Swedish).
(5) Lockhart W. L., Billeck B. N. and Baron C. L. (1989). Bioassays with a floating plant
(Lemna minor) for effects of sprayed and dissolved glyphosate. Hydrobiologia.
118/1 19, 353 - 359.
(6) Environment Canada (1993) Proposed Guidelines for Registration of Chemical
Pesticides: Non-Target Plant Testing and Evaluation. Canadian Wildlife Service,
Technical Report Series No. 145.
(7) Walbridge C. T. (1977). A flow-through testing procedure with duckweed (Lemna
minor L.). Environmental Research Laboratory - Duluth, Minnesota 55804. US EPA
Report No. EPA-600/3-77 108. September 1977.
(8) Huebert D.B. and Shay J.M. (1993) Considerations in the assessment of toxicity using
duckweeds. Environmental Toxicology and Chemistry, 12, 48 1-483.
(9) Bruce R.D. and Versteeg D.J. (1992) A statistical procedure for modelling continuous
toxicity data. Environmental Toxicology and Chemistry, 11, 1485-1494.
(10) Norberg-King T.J. (1988) An interpolation estimate for chronic toxicity: The ICp
approach. National Effluent Toxicity Assessment Centre Technical Report 05-88.
USEPA, Duluth, MN.
(11) Brain P. and Cousens R. (1989) An equation to describe dose-responses where there is
stimulation of growth at low doses. Weed Research, 29, 93-96.
R&D Technical Report EMA 003 59
-------
R&D Technical Report EMA 003 60
-------
ANNEX I
Description of Lemna spp
The aquatic plant duckweed (Lemna spp.) belongs to the family Lemnaceae which has a
number of world-wide species in four genera. Their different appearance and taxonomy have
been exhaustively descnbed (1, 2). The species Lemna gibba and L minor are represented in
the temperate areas and are commonly used for toxicity tests. Both species have a floating or
immersed discoid stem (frond) and a very thin root emanates from the centre of the lower
surface of each frond. Lemna very rarely comes into flower and the plants multiply by
producing new fronds vegetatively (3). In companson with older plants the younger ones tend
to be paler, have shorter roots and consist of two to three fronds of different sizes. The small
size of Lemna, its simple structure, asexual reproduction and short generation time makes
plants of this genus very suitable for laboratory testing (4, 5).
Because of probable interspecies variation in sensitivity, only compansons of sensitivity
within a species are valid.
LITERATURE
1. Hillman, W.S. (1961). The Lemnaceae or duckweeds: A review of the descnptive and
experimental literature. The Botanical Review, 27:221-287.
2. Landolt, E. (1986). Biosystematic investigations in the family of duckweed
(Lemnaceae). Vol. 2. Geobotanischen Inst. ETH. Stiftung Rubel, ZUnch, Switzerland.
3. Bjömdahl, G. (1982). Growth performance, nutrient uptake and human utilization of
duckweeds (Lemnaceae family). ISBN 82-991150-0-0. The Agncultural Research
Council of Norway, University of Oslo.
4. Wang, W. (1986). Toxicity tests of aquatic pollutants by using common duckweed.
Environmental Pollution, Ser B, 11:1-14.
5. Wang, W. (1990). Literature review on duckweed toxicity testing. Environmental
Research, 52:7-22.
R&D Technical Report EMA 003 61
-------
Examples of Lemna species which have been used for testing:
Species Reference
Lemna aequinoctialis Ekiund, B. (1996). The use of the red alga Ceramzum
strictum and the duckweed Lemna aequinoctialis in
aquatic ecotoxicological bioassays. Licenti ate in
Philosophy hesis 19.96:2. Dep. of Systems Ecology.
Stockholm University.
Lemna major Clark, N. A. (1925). The rate of reproduction of Lemna
major as a function of Intensity and duration of light. J
phys. Chem., 29: 935-941.
Lemna minor US-EPA (1996), AFNOR (1996), SIS (1995);
Lemna gibba ASTM (1991), USEPA (1996)
Lemnapaucicostara Nasu, Y. , Kugimoto, M. (1981). Lemna (duckweed) as an
indicator of water pollution. I. The sensitivity of Lemna
paucicostata to heavy metals. Arch. Environ. Contam.
Toxicol., 10:1959-1969.
Lemnaperpusilla Clark, J. R. et al. (1981). Accumulation and depuration of
metals by duckweed (Lemna perpusilla). Ecotoxicol.
Environ. Saf., 5:87-96.
Lemna trisulca Huebert, D. B., Shay, J. M. (1993). Considerations in the
assessment of toxicity using duckweeds. Environ. Toxicol.
and Chem., 12:481- 483.
Lemna valdiviana Hutchinson, T.C., Czyrska, H. (1975). Heavy metal
toxicity and synergism to floating aquatic weeds. Verh.-
mt. Ver. Limnol., 19:2102-2111.
R&D Technical Report EMA 003 62
-------
ANNEX 2
Definitions
The following definitions and abbreviations are used in the guideline:
Clone is a group of individuals reproducing vegetatively (by mitosis) from a single ancestor
Colony consists of a number of fronds (usually 2-4). Sometimes referred to as a plant.
is the concentration of the test substance dissolved in test medium that results in a 50%
reduction in growth of Lemna within a stated exposure period.
Frond is a single Lemna “leaf-like” structure. It is the smallest unit, i.e. individual, capable of
reproduction.
Genotvoe is the genetic constitution of an individual.
Growth is an increase in the recorded parameter, e.g. frond number, dry weight. wet weight or
frond area, over the test penod.
Lowest Observed Effect Concentration (LOEC ) is the lowest tested concentration at which
the substance is observed to have a statistically significant reducing effect on growth (at p <
0.05) when compared with the control, within a given exposure time. However, all test
concentrations above the LOEC must have a harmful effect equal to or greater than those
observed at the LOEC. When these two conditions cannot be satisfied, a full explanation must
be given for how the LOEC (and hence the NOEC) has been selected.
Monoculture is a culture with one plant species.
Necrosis is dead (i.e. white or water-soaked) frond tissue.
Chiorosis is yellowing of frond tissue.
No Observed Effect Concentration (NOEC ) is the test concentration immediately below the
LOEC which, when compared with the control, has no statistically significant effect
(p <0.05), within a given exposure time.
Phenotype describes the characteristics of an organism resulting from the interaction of
genotype and environment.
Semi-static (renewal) test is a test in which the test solution is periodically replaced at
specific intervals during the test.
Static test is a test method without renewal of the test solution during the test.
Test medium is the complete synthetic culture medium on which test plants grow when
exposed to the test substance. The test substance will normally be dissolved in the test
medium.
R&D Technical Report EMA 003 63
-------
R&D Technical Report EMA 003 64
-------
ANNEX 3
Maintenance of stock culture
Stock cultures can be maintained under lower temperatures (4-10°C) for longer times without
reinoculation. The Lemna culture medium may be the same as that used for testing but other
nutrient rich media can be used for stock cultures.
Under normal conditions of temperature (24°C) and illumination (6500 -10000 lux), monthly
subculturing of stock cultures is advised. A number of young, light-green plants are removed
to new culture vessels containing fresh medium using an aseptic technique. Under the cooler
conditions suggested here, subcultunng may be conducted less frequently. Intervals of up to 3
months have been found to be acceptable.
Chemically clean and sterile glass culture vessels should be used and aseptic handling
techniques employed. In the event of contamination of the stock culture e.g. by algae, steps are
necessary to eliminate the contaminating organisms. In the case of algae and most other
contaminating organisms, this can be achieved by surface stenlisation. A sample of the
contaminated plant material is taken and the roots cut off. The material is then shaken
vigorously in clean water, followed by immersion in a 0.5% (v/v) sodium hypochlorite
solution for between 30 seconds and 5 minutes. The plant material is then rinsed with sterile
water and transferred, as a number of batches, into culture vessels containing fresh culture
medium. Many fronds will die as a result of this treatment, especially if longer exposure
periods are used, but some of those surviving will usually be free of contamination. These can
then be used to re-inoculate new cultures.
R&D Technical Report EMA 003 65
-------
R&D Technical Report EMA 003 66
-------
ANNEX 4- Media
Different growth media are recommended for L. minor and L. gibba. For L. minor, a modified
Swedish Standard (SIS) medium is recommended whilst for L. gibba, 20X AAP medium is
recommended. However, when initialising cultures on a new medium, growth may initially be
poor and an acclimation period of at least one month is advised. Compositions of both media
are given below. When preparing these media, reagent-grade chemicals should be used.
Swedish Standard (SIS) Lemna growth medium
Stock Substance
Concentration Concentration Prepared medium
solution No.
in stock in prepared
solution (g I’) medium
(mg 1.1)
Element Concentration (mg Fi)
I NaNO 3
850 85
Na, N 31.86. 1400
KH 2 PO 4
1.34 134
K,P 601.238
II MgSO 4 7H 2 0
15 0 75
Mg, S 7 39, 9 77
III CaCI 2 . 2H 2 0
7.20 36
Ca, Cl 9 82, 17 46
IV Na 2 CO 3
400 20
C 226
V H 3 B0 3
100 100
B 0173
MnCI 2 . 4H 2 0
0 200 0.200
Mn 0.056
Na 2 MoO 4 . 2H ,O
0.010 0010
Mo 0.004
ZnSO 4 7H 2 0
0.050 0050
Zn 0011
CuSO 4 .5H 2 0
0005 0005
Cu 00013
Co(N0 3 ) 2 6H 2 0
0.010 0010
Co 0002
VI FeC I 3 . 6H 2 0
0.168 084
Fe 0 168
Na 2 -EDTA . 2H 2 0
0280 1 40
- -
VII MOPS (buffer)
488 488
- -
20X AAP growth medium
Stock solution Substance
No.
Concentration in stock
solution (g 15
Prepared medium
Element Concentration (mg 15
I (macronutrients)
NaHCO 3
15.0
Na, C 2202, 42.86
K 2 HPO 4
1.044
K 9.38
P 3.72
MgSO 4 .7H 2 0
14 7
S 38 22
NaNO 3
25.5
N 84 0
MgC I 2 .6H 2 0
12.164
Mg 58 08
CaC I 2 .2H 2 0
4.41
Ca 24.04
II H 3 Bo 3
0.1855
B 0649
(micronutnents) MnC I 2 4H 2 O
04154
Mn 23075
ZnC I 2
00033
Zn 00314
C0CI 2 .6H 2 0
00014
Co 00071
CuCI 2 2H 2 0
1 2x10-5
Cu 008 tgF 1
Na 2 MoO 4 .2H 2 0
0.0073
Mo 00576
FeC I 3 .6H 2 0
0.1598
Fe 0661
Na 2 EDTA 2H 2 0
0.3000
- -
R&D Technical Report EMA 003 67
-------
R&D Technical Report EMA 003 68
-------
ANNEX 5
Estimation of the time-weighted mean
Given that the concentration of the test substance can decline over the period between medium
renewals, it is necessary to consider what concentration should be chosen as representative of
the range of concentrations experienced by Lemna. The selection should be based on
biological considerations as well as statistical ones. For example if growth is thought to be
affected mostly by the peak concentration experienced, then the maximum concentration
should be used. However, if the accumulated or longer term effect of the toxic substance is
considered to be more important, than an average concentration is more relevant. In this case.
an appropriate average to use is the time-weighted mean concentration, since this takes
account of the variation in instantaneous concentration over time.
Figure 1 shows an example of a (simplified) test lasting seven days with medium renewal at
Days 0, 2 and 4.
12
10
8
6
4
2
0
Days
Figure 1: Example of time-weighted mean
1 2 345 67
R&D Technical Report EMA 003
69
-------
• The thin zigzag line represents the concentration at any point in time. The fall in
concentration is assumed to follow an exponential decay process.
• The 6 plotted points represent the observed concentrations measured at the start and
end of each renewal period.
• The thick solid line indicates the position of the time-weighted mean.
The time-weighted mean is calculated so that the area under the time-weighted mean is equal
to the area under the concentration curve. The calculation for the above example is illustrated
in Table!.
Calculation of Time-weighted mean
Renewal No.
Days
Conc 0
Conc 1
Ln(Conc 0)
Ln(Conc 1)
Area
1
1
10.000
4.493
2.303
1.503
13.767
2
2
11.000
6.037
2.398
1 798
16.544
3
3
10.000
4.066
2303
1.403
19781
Total Days:
7
Total Area:
TWMean
50.09 1
7.156
Days is the number of days in the renewal period
Conc 0 is the measured concentration at the start of each renewal period
Conc I is the measured concentration at the end of each renewal period
Ln(Conc 0) is the natural loganthm of Conc 0
Ln(Conc I)is the natural loganthm of Conc 1
Area is the area under the exponential curve for each renewal period. It is calculated by:
Area = Conc 0 - Cone I x Days
Ln(Conc 0) - Ln(Conc 1)
The time-weighted mean (7W Mean) is the Total Area divided by the Total Days.
It is clear that when observations are taken only at the start and end of each renewal period, it
is not possible to confirm that the decay process is, in fact, exponential. A different curve
would result in a different calculation for Area. However, an exponential decay process is not
implausible and is probably the best curve to use in the absence of other information.
However, a word of caution is required if the chemical analysis fails to find any substance at
the end of the renewal period. Unless it is possible to estimate how quickly the substance
disappeared from the solution, it is impossible to obtain a realistic area under the curve, and
hence it is impossible to obtain a reasonable time-weighted mean.
R&D Technical Report EMA 003 70
-------
APPENDIX C GUIDANCE TO RING-TEST PARTICIPANTS
AND PRO-FORMA DATASHEETS
R&D Technical Report EMA 003 71
-------
DRAFT OECD LEMNA GROWTH INHIBITION TEST
PLANNED RING-TEST, 1998
Background
Most OECD guidelines provide some latitude in methodology e.g. in the genotype of
the test organism used, environmental conditions or the test vessels. This is
necessary for practical reasons although this flexibility will contribute to variability in
estimates of toxicity. The draft OECD guideline for a Lemna growth inhibition test is
no different and currently, the extent of that variability and the components of
variability are unknown. It is not the intention of the ring-test to investigate possible
modifications to the draft guideline because work on this aspect has already been
carried out. Instead, the ring-test is designed to provide information on key
performance characteristics - paticularly variability - which affect the validity and
interpretation of test data.
Aim of ring-test
The ring-test is designed to assess the following performance characteristics:
1. The extent to which validity (doubling time) criteria are met
2. The repeatability of the test within laboratories (this can be used to describe the
Drecision of the method)
3. The reproducibility of the method between laboratories (this data may be used to
assess the accuracy with which toxicity is estimated)
Instead of different laboratories carrying out different studies, all participating
laboratories are required to carry out essentially the same method. This makes the
approach distinct from previous OECD ring-tests but similar to performance-related
ring-tests used for new chemical analytical methods. The benefits of this approach
are that the information generated can be used to determine possible errors in
decision-making and also, if required, to define limits for variability and bias in
testing. A recent example of this approach to ring-testing of biological test methods
is attached.
Although it is not the main intention of the ring-test to further develop the guideline,
it is possible that changes may be recommended where they have a strong bearing
on the precision and accuracy of test data.
The ring-test is intended as a study to assess critical performance characteristics of
the method rather than a study to optimise aspects of the methodology, Of
particular concern are the precision and accuracy of toxicity estimates .
Work Programme
72
-------
Plant material
Testing will be carried out using either Lemna minor or Lemna gibba. The aim of the
exercise is to assess variability and bias when the guideline is performed as
described i.e. under realistic conditions. Therefore, there will be no requirement to
use a particular clone but plant material should be acquired and maintained as
described in the draft guideline.
Repeat tests
It is important that participating laboratories carry out tests according to the OECD
Guideline and that tests with the same toxicant and species are repeated on
different occasions to provide an estimate of repeatability. Of course, the more
laboratories participating and the more repeat tests they carry out, the better the
resulting estimates of repeatability and reproducibility will be. However, there are
practical constraints on what laboratories can contribute. Experience gained in
similar studies by ourselves and others indicates that 4-5 repeat tests with the
same toxicant are necessary for analysis. It is unlikely that data from just 1 or 2
repeats can be used in the analysis. Clearly this is a significant undertaking for
participants although the repeat tests may be carried out over a period of several
months, perhaps alongside other tests to reduce costs. Ideally, participating
laboratories would carry out repeat tests with both recommended toxicants (see
below) but this will not usually be possible. It is much more valuable to carry out 4-5
repeats with the same species and toxicant than to carry out a smaller number of
repeat tests using both toxicants or both species.
Random error in ecotoxicity testing is typically higher when tests are performed by
people with little experience and the effect of this can be to reduce (sometimes very
markedly) the repeatability of the resulting toxicity estimates. Bias (a consistently
higher or lower estimate of toxicity compared to the ‘consensus’ mean) is more likely
to arise unwittingly, e.g. as a result of some systematic error. To enable a realistic
assessment of precision and accuracy, participants should already have experience
of carrying out toxicity tests using Lemna.
Verification of test concentrations
Ideally, estimates of toxicity would be based on measured concentrations but it is
unrealistic to expect participating laboratories to incur the expenses involved. It is
acknowledged that this represents a deviation from the guideline. Therefore, toxicity
estimates in the ring-test will normally be based on nominal concentrations although
we would strongly encourage participants to at least verify the concentration of the
stock solution used for preparing test solutions by chemical analysis.
Data to be reported
The ‘raw data’ used in the analysis of of the ring-test results are EC5O estimates
from individual toxicity tests. We would ask the following to be made available for
each repeat test:
• Test species (L..minoror 1... gibba) and source
• Results for doubling time in the controls
73
-------
• The 7-day EC5O and 95% confidence interval for average specific growth rate,
based on frond number
• A note of the method used to estimate the EC5O (i.e. graphical method, logistic
model, cumulative normal model)
• A note of any significant deviations to the draft guideline or difficulties in meeting
recommended criteria for pH, light intensity, temperature etc.
Although it is not the main intention of the ring-test to further develop the guideline,
observations on any recurring practical difficulties encountered or suggestions for
improvement are encouraged. We ask participants to retain raw (replicate) data
although this need not be reported routinely.
A pro-forma record sheet will be made available to facilitate the analysis of data
submitted by participants. This will bear a unique number identifying the laboratory.
At no point will the identity of specific data be attributed to a particular laboratory in
any communication except possibly between WRc and the laboratory concerned.
Participating laboratories are asked to carr)’ Out 4-5 repeat tests with the same
species and toxicant and to estimate the EC5O based on average specific growth rate
for each test. It is expected that these will normally be based on nominal
concentrations. Participants should have experience of carrying out Lemna toxicity
tests.
Choice of toxicants
The choice of toxicant can influence one’s view about the performance
characteristics of ecotoxicity tests. It appears that organic toxicants tend, on the
whole, to be ‘better behaved’ (i.e they give rise to smaller CVs for repeatability and
reproducibility) than many inorganics. Experience shows that there is merit in
collecting toxicity data for an inorganic and an organic toxicant.
Sometimes it is suggested that a particular substance is a good toxicant because ‘it
gives repeatable results’. However, such a toxicant is less likely to provide useful
information about variability and bias in a test than one which is sensitive to
variations in factors such as water quality parameters, experimental practice and
variations in genotype. Such factors give rise to the variations in toxicity estimates
that we are interested in.
Two toxicants are proposed for the ring-test: an organic and an inorganic salt. They
have been selected based on the following criteria:
• hydrolytically stable over period of several months (stock solution)
• low volatility from water
• existing data for toxicity to Lemna species which shows reasonably well-defined
dose-response
• low-moderate toxicity to Lemna (use of higher concentrations helps minimise
errors in preparation of solutions and analysis)
• water solubility well above toxic concentrations
• ease of analysis
• readily available to all participants
• stable/ ‘conservative’ speciation over the pH range recommended for the test
74
-------
The foilowing substances are proposed as toxicants in the ring-test, providing a
reasonable match with the criteria listed above, although no toxicity data for L. gibba
and 3,5-dichiorophenol have been located t :
Organic: 3,5 - dichiorophenol
inorganic: potassium dichromate
Some relevant information about each of these is shown below:
Toxicant
Chemical
formula
Water
solubility
Stability
Analysis
Toxicity to
Lemna
Comments
spp.
3,5-
C•HIOCI,,
Ca. 200
hydrolytically
HPLC-UV
0.6-2.6
Dose-response
dichiorophenol
mgr’
stable: may be
liable to
biodegradation
mgr’ (L.
minor)
No data for
L. gibba
appears to be
fairly shallow so
choice of test
concentrations is
important
potassium
K 2 Cr 3 O,
highly
stable. Cr(VI)
AAS
10-60 mgl
dichromate
soluble
predominates
at pH>5.
(L. minor)
45 mgl’
(L. gibba)
Usually, participating laboratories will be able to contribute data for only one
toxicant. Unless there is a particular desire to use one or the other, we encourgae
the use of potassium dichromate over 3,5-dichlorophenol.
Choice of test concentrations
As described in the guideline, the test concentrations should be chosen to bracket
the EC5O with an interval between concentrations not exceeding x3.2. Where
toxicity is reasonably well-characterised as in the case of these toxicants, a
narrower interval (say, 2-fold) is advised, If laboratories have no experience with
these particular toxicants, the data given in the Table above provide some
indications of the range of likely EC5O values. Thus, for potassium dichromate,
suitable test concentrations may be: 0, 10.0, 20.0, 40.0, 80.0 and 160.0 mgr’.
Stock and test solutions
With the limited resources available, it is not practical to distribute sub-samples of a
single stock solution to participating laboratories. Therefore, participants are asked
to prepare their own stock solutions using analytical grade 3,5-dichiorophenol or
potassium dichromate from a reputable chemicals supplier (e.g. Sigma, Merck,
Lancaster). Preparation of sufficient stock solution to last for the entire duration of
the ring-test is strongly recommended so that any between-batch variation in
concentration can be eliminated. A small quantity of acetone as an auxilliary solvent
will usually be necessary for preparing stock solutions of 3,5-dichlorophenol; it is
best to add the required volume of acetonic solution to a small volume of water and
then make up to the required volume with water. Where possible, participants are
strongly encouraged to confirm the concentration of this stock solution by analysis.
If you have any information on toxicity of 3 ,5-dichiorophenol to L. gibba, please make it
available and I will circulate it to participants
75
-------
Laboratories may also wish to consider retaining samples of the highest and lowest
concentrations from each test so that ‘investigative’ analysis is possible in the event
of an unusually high or low EC5O estimate.
When not is use, stock solutions should be stored in the dark at Ca. 4°C.
Participating laboratories are requested to use one oi both of the following toxicants:
3,5-dichiorophenol and potassium dichromate. Participants should prepare a single
stock solution for all their testing and if possible, the concentration in the stock should
be confirmed by analysis.
Co-ordination
The ring-test will be co-ordinated by WRc and the contact person for
correspondence and data will be Paul Whitehouse. His contact details are as
follows:
Paul Whitehouse
WRc plc
Henley Road
Medmenham
Marlow
Buckinghamshire
UK
SL72HD
Tel: 0044-1491-571531
fax: 0044-1491-579094
e-mail: whitehouse_p@wrcplc.co.uk
WRc will be responsible for providing guidance on methodology, pro-forma reports
and analysis of participants’ data. WRc will also be a participating laboratory in the
ring-test.
Data analysis and reporting
For a more detailed description of the way in which toxicity data will be handled, you
are referred to the enclosed paper by Whitehouse et a! (1996). Essentially, the
analysis will partition sources of variation (between- and within-laboratories). The
precison of the method (i.e. the degree of agreement between repeated
observations made within a laboratory or between laboratories) can then be
calculated. Taking the ‘consensus’ mean EC5O for a particular species/toxicant as
the ‘true’ toxicity, it is also possible to estimate the accuracy with which toxicity is
76
-------
measured (i.e. how close individual EC5O values are to the ‘true’ value). Using the
estimates of precision and accuracy determined in the ring-test as a measure of
what may reasonably be achieved, it is also possible to derive limits for precision
and accuracy, if required.
The repeatability and accuracy achieved by any particular laboratory will not be
investigated or reported.
Timetable
Completion of draft guideline:
OECD commenting round complete:
Amendments to guideline and distribution to participants:
Ring-test (practical work) start:
Ring-test (practical work) end:
Data analysis and interpretation complete:
May 1998
June 1998
June 1998
July 1998
November 1998
February 1999
77
-------
OECD LEMNA GROWTH INHIBITION TEST
Invitation to participate in ring-test
If you would like to participate in the ring-test of the draft OECD Lemna
growth inhibition test, please complete this form and return to Paul
Whitehouse at WRc Medmenham (address details are given below).
We would like to participate In the planned ring-test:
Name of organisation:
Contact name (for correspondence):
Address:
Telephone:
Fax:
e-mail:
We would like to make the following contribution:
Species:
Toxicant(s):
Thank you for your interest and we look forward to a fruitful collaboration.
Your responses will be made available only to those people involved in the
project. Please return this form by 29th May to:
Dr Paul Whitehouse
WRcplc
Henley Road
Medmenham
Marlow
Buckinghamshire. SL7 2HD
UK
Tel: 0044-1491-571531
Fax: 0044-1491-579094
email: whitehousQ..p@wrcplc.co.uk
78
-------
1998-99 OECD Lemna Growth Inhibition Ring-Test
Data Sheet
IRepeat # Date Started I I Lemna species I 1
FLaboratory 1 I lnvestioator I Toxicant I 1
Nominal
Concentration
24hrs I 48hrs I 72hrs I 96hrs I l2Ohrs I l44hrs I l68hrs
t
CONTROL
CONTROL
CONTROL
CONTROL
Frond Number
Temperature
I
I
pH (start)
pH (finish)
Initial frond #
Light Intensity
-------
Data Summary Sheet
FRepeat # I I Date Started I I Lemna species
ILaboratory I I Investigator I I Toxicant I
•‘1 control 1 I Icontrol 2 I Icontrol 3 I 1
Average specific % Inhibition ot growth
growth rate (j.i) rate (Ir)
Controls
CONG. 1
CONG. 2
CONC. 3
CONC.4
CONG. 5
CONC. 6
I Estimated EC5O umg/t) I I Model used I I
I 95% Contidence Limits I I
Deviations from draft test guideline
I growth found in any treatment 9 YES NO
Stimulation of
Change in development or appearance of contrcl colonies 9
80
-------
INSTRUCTIONS FOR CALCULATING ECSOS USING “TOXCALC”.
Manipulation of Lemna growth data in Excel, to produce slope
(p.) values and analysis using ‘Toxcaic’
Data manipulation
On a blank ‘Excel’ spreadsheet set up 10 columns thus:
( A) ( B) ( C) CD) CE) (F) ( G) (H) (I) (J )
(1
) _______________________ ________________ __________ _________ _________ ________ _________ _________ _________ _________
(2
) _______________________ _______________ __________ _________ _________ ________ _________ _________ _________ _________
(3
) _______________________ _______________ __________ _________ _________ ________ _________ _________ _________ _________
(4
) _______________________ _______________ _________ _________ _________ ________ _________ _________ _________ _________
(5
) _________________________ _________________ __________ _________ _________ _________ __________ __________ _________ __________
(6
) _______________________ _______________ __________ _________ _________ _________ _________ _________ _________ _________
(7
)
Use =LN(C3) etc to link cells in columns G, H, I and J to their corresponding cells in columns C, D, E
and F.
Input test data for frond No. in columns C, D, E & F. The transformed data will appear in columns 0, H,
I&J.
When complete rearrange transformed data by setting up a table beneath the first table thus:
(87)
(90)
(91)
(92)
(93)
(94)
(95)
(96)
(97)
(98)
Now we can start analysing the data. The first step (1) is to check that control growth is exponential. If
so, we can go on to estimate j.L for each treatment (step 2).
Step (1) Select lools, Qata analysis, Regression (ok).
81
Concentration
(units)
0
0
0
0.32
Replicate
No.
1
2
3
1
Day
Day
Day
Day
0
3
5
7
Natural log of frond No.
Day
0
Day
3
Day
5
Day
7
=LN
(C3)
=LN
(D3)
=LN
(E3)
=LN
(F3)
=LN
(C4)
=LN
(D4)
CE)
-------
In the box for Y values enter the 4 cells containing control data in column B of last table, (B90-B93) for
example.
In the box for X values enter the 4 cells containing day numbers, (A90-A94) for example.
Enter a cell number into the Select output range box where you want the results of the regression
analysis to start.
Select the Line fit plots box for CONTROL data only. Then click ok.
A straight line for the control plots confims exponential growth of Lemna. The slope parameter (ii) is
referred to as the ‘X variable’ in Excel.
Step (2) To estimate ji for each treatment replicate, select a new cell for the Select output range box each
time and REMOVE THE LTh E FIT PLOT option for non-control data.
Analysis of data
At this point you can either use ‘Toxcaic’ to calculate growth inhibition and EC 50 values or you can do
this manually, or you may use an alternative software package which offers suitable models (see draft
guideline).
If using Toxcaic to calculate growth inhibition and to estimate EC 50 values, set up another table below
regression outputs linked to the XVariables produced by the regressions for each control and treatment
replicate thus:
These data are then ready for analysis by Toxcaic.
If not using Toxcaic to perform final analysis, omit this last table and extract the Xvariables (slope
values) for use by other methods, e.g. graphical estimation.
Note :
It is advisable to edit an old file when entering data for repeat tests. Linked cells will automatically re-
calculate and re-arrange data, but regressions will need to be re-calculated for each repeat test.
Enter the same cell address in the select output range box as used previously, and overwrite old
regression outputs.
(A)
(B)
(C)
(D)
(1000)
(1001)
(1002)
(1003)
(1004)
(1005)
0
1
2
3
=(cell address containing
regression analysis
Xvariable, i.e. slope
value, for control
replicate 1, e.g B 120)
=(cell address containing
regression analysis
Xvariable, i.e. slope
value, for control
replicate 2, e.g B 141)
=(cell address containing
regression analysis
Xvariable, i.e. slope value,
for control replicate 3, e.g
B 162)
0.32
1
2
3
=(cell address containing
regression analysis
Xvariable, i.e. slope
value, for treatment 1
replicate 1, e.g B 183)
=(cell address containing
regression analysis
Xvariable, i.e. slope
value, for treatment 1
replicate 2, e.g B204)
=(ceII address containing
regression analysis
Xvariable, i.e. slope value,
for control replicate 3, e.g
B225)
82
-------
APPENDIX D EC 50 VALUES ESTIMATED FROM RING-
TEST
EC 50 s for potassium dichromate to Lemna minor, calculated using average specific
growth rate based on frond number.
Laboratory
code No.
EC 50 s of replicate tests
(mg Cr F’)
Mean
EC 50
7 [ 3.4] [ 2.4] [ 2.7] [ 3.0] 2.8 3.2 2.5 - 2.8
9 2.1 1.1 - - - - - - 1.6
10 0.87 0.9 0.51 0.74 0.72 - - - 0.75
11 1.1 1.7 1.4 1.9 1.9 - - - 1.6
12 1.1 2.2 1.1 2.2 2.5 3.6 - - 2.1
13 1.3 0.9 11* 6.8* 8* 5.8* [ 9*] [ 8*] 1.1
14 1.8 1.6 1.6 1.6 - - - - 1.7
19 29.4 29 - - - - - - 29.2
20 9.4 2.8 2.5 - - - - - 4.9
21 1.8 2.6 3.1 2.5 2.7 - - - 2.5
22 [ 6.1] [ 2.7] [ 3.8] (2.6] - - - - -
23 3 3.5 1.8 3.6 2.2 - - - 2.8
25 2.5 2.3 1.7 1.7 1.1 - - - 1.9
26 [ 18.6] [ 8.9] [ 7.4] [ 19] 17.4 - - - 17.4
28 5 12.2 6.4 5.5 - - - - 7.3
30 17.7* 52.4* 22.6* -
31 21 18 9.6 2.5 9.5 - - - 12.1
32 2 2.1 2.1 - - - - - 2.1
35 [ 9.8] - - -
36 [ 4.4] 3.4 - - - - - - 3.4
37 0.66 1* 1.1*
[ ] = doubling time exceeded
= data extrapolated
R&D Technical Report EMA 003 83
-------
EC 50 s for 3,5-dichiorophenol to Lemna minor, calculated using average specific growth
rate based on frond number
Laboratory
Code No.
EC 50 s
(
of replic
mg DCP
ate
l ’)
tests
Mean
EC 50
5 4.6
2.9
2.4
1.8
2.7
2.9
9 4.2
2.5
-
-
-
3.4
12 3
3
3
2.9
3
3.0
13 2.5
2.5
2.7
2.4
-
2.5
15 2.1
2.3
2.4
2.4
3.3
2.5
18 3
2.7
2.8
2.9
2.7
2.8
20 [ 4.3]
[ 3.6]
[ 3.2]
2.8
2.7
2.8
23 3.9
2.7
2.7
2.8
-
3.0
25 2.8
2.8
3
2.5
2.5
2.7
27 4.4
4.1
4.6
-
-
4.4
30 3.2
3.2
3.3
-
-
3.2
32 2.8
2.9
3
-
-
2.9
33 [ 4.4]
5.6
8.4
8.3
6.6
7.2
36 2.5
2.8
-
-
-
2.7
[
ii
] = doubling time exceeded
* = data extrapolated
EC 50 s for potassium dichromate to Lemna gibba, calculated using average specific
growth rate based on frond number
Laboratory EC 50 s of replicate tests Mean
Code No. (mg Cr l ) EC 50
2 [ 37] 28.6 [ 41.7] [ 35.4] [ 34.3] - - - 28.6
3 6 10.9 8.2 - - - - - 8.4
4 14 14.1 11.1 11.5 [ 18.11 [ 4.88] [ 4.93] [ 2.2] 12.7
6 26.2 18.8 28.6 21.4 28.6 - - - 24.7
8 0.32 0.49 0.25 - - - - - 0.35
16 42 44.8 43.5 39.6 - - - - 42.5
17 [ 21.7] [ 26.2] 15.4 8.6 8.8 - - - 10.9
24 2.4* 25.7* 077* 2.4* - - - - -
25 7.8 - - - - - - - -
29 19 110 - - - - - - -
37 J3* 0.21* 0.07* - - - - - -
] = doubling time exceeded
= data extrapolated
R&D Technical Report EMA 003 84
-------
EC 50 s for 3,5-dichiorophenol to Lemna gibba, calculated using average specific growth
rate based on frond number.
Laboratory Code
No.
EC 50 s of replicate tests
(mg DCP l )
Mean EC 50
1
>4
- -
-
-
34
7
6.7 6.6
6
6.6
R&D Technical Report EMA 003 85
-------
R&D Technical Report EMA 003 86
-------
APPENDIX E GUIDANCE ON THE TESTING OF
EFFLUENTS, LEACHATES AND RECEIVING
WATERS IN THE LEMNA GROWTH
INHIBITION TEST
Introduction
For the testing of effluents, leachates and receiving water samples, the conventional approach
to estimating phytotoxicity would be to carry out an algal growth inhibition test (e.g. OECD
201). For such samples, growth inhibition tests using Lemna may offer some important
advantages over algal toxicity tests. This is because toxicity is less prone to interference by
physical properties of the sample which can result in phytotoxicity being over-estimated:
(a) Toxicity is not strangly influenced by colour. Indeed, unlike algal tests, Lemna growth
tests can be performed on completely opaque samples.
(b) Toxicity is not strongly influenced by the presence of suspended solids.
(c) Toxicity is unaffected by dissolved oxygen levels
(d) Certain ‘difficult’ samples such as emulsions may be tested
In principle, the testing of effluents, leachates or receiving water samples is the same as for
single chemicals, as described in the draft OECD test Guideline (Appendix B). However, there
are some important procedural differences and these are highlighted below.
Collection, transport and storage of samples
Samples should be collected in inert containers such as glass (amber glass is preferable to
clear glass). If large volumes are to be .collected, containers made of HPDE or unlacquered
stainless steel may be used. Containers should be filled completely to minimise any air space
into which volatile components could pass.
For a static test with five test concentrations and controls, up to 21 of sample may be needed.
Larger volumes are required for semi-static or flow-through tests than for static tests (see
below) and this should be taken into account when collecting samples for testing.
Samples should be tested within 48-h of collection and kept in the dark during transport and
storage. The temperature during transport and storage should not deviate markedly from the
temperature at the time of collection (±2°C). It follows that the temperature at collection
should be recorded and ideally, it should be logged at intervals during transport and storage. If
a sample is not to be tested immediately, it should be stored in the dark at 2-8°C although
consideration should be given to the time required to bring the sample to the temperature
required for the Lemna test (24±2°C).
R&D Technical Report EMA 003 87
-------
Preparation of samples for testing
Thetmpqrature of the , sample should be mçnitored and steps taken to ensure that the test can
be . nitiatecI at the correct temperature (24+2 C).
As far as possible filtration should be avoided. Although the test Guideline specifies the pH at
which the test should be pe formed, the expelirilenter should consider whether extremes of pH
constitute a feature of the saipple’s potenti l..tqxicity and therefore whether pH should be
adjusted. Whe è pH adjustment is warranted, . this may only be necessary at high test
concentrations. Aeration is to be avoided because this may reduce dissolved concentrations of
volatile substances.
Certain products may pose particular problems e.g. because they are sparingly soluble in
water, fc rin emulsions in water or comprise components of diffenng solubility. Recommended
methods fortesting such substances are to be found in ECETOC (1996).
Test regime
Effluents, leachates and receiving water samples should be regarded as unstable and, as such,
testing should be perfoijned using a semi-static or flow-through regime Usually any chemical
changes in the sample will not be known but a minimum of three replacements is advised.
These replacements should be made regularly (e.g. after 2, 4 and 6 days) using the original
sample.
Test design
Receiving waters:
In the case of receiving water samples, it would be customary to test only an undiluted sample
because the primary concern is usually to assess the effects on plant growth of that particular
sample. Consequently, a ‘limit test’ approach is the most appropnate test design, involving
comparison of responses in a control group 3 and in the undiluted sample. This would normally
entail a comparison of mean responses using e.g. a t-test but the experimenter should give
consideration to the power of the test and hence the level of replication. Higher levels of
replication will result in greater power and ability to resolve smaller differences between
means. The minimum significant difference should always be reported.
Effluents, leachates, complex mixtures:
These may be tested using a concentration-response design as described in the draft test
Guideline, for single chemicals. There are no fundamental differences in test design between
single chemicals and effluents, leachates and complex mixtures.
In some circumstances, a reference water e.g. taken froiyi a site upstream of an effluent discharge, may be used
This would have-to be spiked’ with growth medium, as described below This approach may provide a useful
estimate of relative growth inhibition resulting from exposure to the effluent but if an absolute estimate of
effluent toxicity is required, controls based on growth medium/distilled water should be used Conditions for
the sampling, transport and storage of reference waters should be the same as for receiving water samples
R&D Technical Report EMA 003 88
-------
Test procedure
The main procedural diffeftnde between. tear yiflgt&iV L emhâ1 ro tti’ I i itiö st vith a
single substance and an effluent, leachate or’en. ’irohmenta}’ sámp-lè con étni th&f3rejiaràtion of
test solutions.
When testing single chemicals, differóTn quantities of ‘Ib t übs tance ‘ãi é. h Wmally
dissolved in the growth medium to create aconcentratioiv ènè Witfi’ f fluents, lead’iates or
environmental samples this cannot be done’ because (a) it thay-bè ne ssa t&te t i1 toi0O%
sample and (b) if saniple was merely diluted with grôwt i of
nutrients would differ between test concentrations (they would be hi h r ’f ói 21 th 1& A er test
concentrations than for the higher concentrations).
Clearly, we need to ensure that each test concentration contains an equivAlent ct hceMrâtion of
nutrients so that any differences in growth between treatnients 1 can be attributed to- &istiiuents
of the sample and not to differences in nutrient levels.
Essentially, what must be done is to:
1. Prepare test solutions containing test sample diluted td the cc e t c ricèntrátió i with
distilled water; and -
2. ‘Spike’ each test concentration with small volumes of concentrated nutrient- stock
solutions.
An example of the way a concentration series may be set up for a test involving Lemna minor
grown in modified SIS medium is shown below: - -
Required test Volume Volume
concentration of test of
(%v/v) sample distilled
(ml) water
(ml)
Volume of stock solutions -
(mJ)
Stock SoIn. No.
I IV V. VI
- Tj tal.
volume
of t st
solution
(L)
0 0 969
1.0 10 959
2.2 22 947
4.6 46 923
10.0 100 869
22.0 220 749
46.0 460 509
96.9 969 0
10 5 5 5 1 5
10 5 5 5 1 5
10 5 5 5 1 5
10 5.. j 5 5 J 5..
10 5 5 5 1 “‘
10 5 5 5 1 5
10 5 5 5 1 5
0 5 5 5
1
1
i.
Ij. i
‘f
1
1
an allowance is made for the volume of the stock s Iutions ad d to th e1( tibl ?
R&D Technical Report EMA 003 89
-------
Each sample dilution should be ‘spiked’ with the concentrated SIS stock solutions before
aliquots are placed into the replicate test vessels at that treatment. It can be seen from the
Table above that the maximum concentration that may be tested is approximately 97%(v/v).
It is possible that any nutrients contained within samples for testing will influence growth
rates but this is an unavoidable feature of any test method with a growth endpoint. Indeed, it is
more likely that hormesis will be evident (growth stimulation at low test concentrations) than
is the case for single substances. In this case, a hormesis model may be required for data
analysis and this is described in the draft test Guideline.
As effluents and surface waters are likely to contain alga] cells which could subsequently lead
to significant algal contamination of test vessels, a semi-static test design with renewal of the
test medium at 48-hour intervals is recommended. The use of clean test vessels at each
renewal helps to minimise the build-up of algal cells, reduces the possibility of nutrient
limitation and leads to better control of pH.
Reference
ECETOC (1996) Aquatic toxicity testing of sparingly soluble, volatile and unstable substances
ECETOC Monograph No. 26, pp. 1-67, Brussels.
R&D Technical Report EMA 003 90
------- |