The analysis of aquatic toxicity data

160 Pages·2004·1.94 MB·English

The analysis of
aquatic toxicity data
S.A.L.M. Kooijman and J.J.M. Bedaux
ii
The analysis of
aquatic toxicity data
S.A.L.M. Kooijman
Professor of Theoretical Biology
Vrije Universiteit Amsterdam
J.J.M. Bedaux
Lecturer in Biostatistics
Vrije Universiteit Amsterdam
VU University Press
iii
Published by the VU University Press
De Boelelaan 1083, 1081 HV Amsterdam
cS.A.L.M. Kooijman & J.J.M. Bedaux
(cid:13)
First published 1996
Printed in the Netherlands
ISBN 90-5383-477-X paperback
iv
Contents
Preface vii
1 No-effect concentrations in environmental policy 1
2 Toxic effects as process perturbations 9
by Bas Kooijman
2.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Standard descriptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Dynamic Energy Budgets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Toxicokinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.5 Effects on individuals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.6 Effects on population dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.7 Toxicity patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
2.8 Conclusions/perspectives. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3 The analysis of standardized bioassays 49
3.1 Some statistical properties of estimates of no-effect concentrations . . . . . . . . . . 50
by Bas Kooijman and Jacques Bedaux
3.2 Statistical Analysis of Bioassays based on hazard modelling . . . . . . . . . . . . . . 58
by Jacques Bedaux and Bas Kooijman
3.3 Analysis of toxicity tests on fish growth . . . . . . . . . . . . . . . . . . . . . . . . . 72
by Bas Kooijman and Jacques Bedaux
3.4 Analysis of toxicity tests on Daphnia survival and reproduction . . . . . . . . . . . . 93
by Bas Kooijman and Jacques Bedaux
3.5 No-effect concentrations in algal growth inhibition tests . . . . . . . . . . . . . . . . 114
by Bas Kooijman, Arnbjørn Hanstveit and Niels Nyholm
4 Software package DEBtox 127
by Matthijs Luger, Jacques Bedaux and Bas Kooijman
4.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
4.2 Parameter estimation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
4.3 Input. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.4 Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
4.5 Final remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
Glossary 145
v
vi
Preface
Legislation around emission of chemicals in the environment aims at minimizing
ecological effects. Because of our poor understanding of ecosystem dynamics, it is
usually unclear how effects on particular species translate into ecological effects. The
interpretation of results from experiments with mesocosms is also far from obvious.
This explains why No-Observed Effect Concentrations (noecs) from standardized
bioassaysarefrequentlyusedinenvironmentalriskassessments. Thisuseis,however,
under increased criticism due to a substantial statistical problem that is inherent to
thisconcept: theproblemofrecognizingsmalleffectsinscatterydata. Astatistically
nonsignificant effect does not imply that biologically significant effects are absent.
Small-effect concentrations have been proposed to replace noecs in risk analysis.
This approach suffers from several problems, such as: What is small?; How do small
effects on species relate to ecosystem dynamics?; To what extent does the resulting
value depend on those model details that do not have a mechanistic basis? How do
we extrapolate acute effects to chronic effects?
ThisbookdescribeshowtheNo-EffectConcentration(nec)canbeestimatedasa
model parameter from data of standardized aquatic toxicity tests: acute and chronic
survival, body growth (of fish), reproduction (of daphnia) and (algal) population
growth. Thisalternativeforthenoecdoesnotsufferfromstatisticalproblems,while
its use in risk assessments still avoids the difficulty of translating observed effects
into consequences for ecosystems dynamics. Being a model parameter, the point
estimate of the nec can be provided with a confidence interval. This allows positive
identification of a concentration range where no effects are to be expected, which is
not possible in the noec approach. If, on the other hand, the null hypothesis nec=0
(implying that the toxicant has effects in all concentrations) cannot be rejected, one
might consider additional research about the effects of the compound.
The method to estimate necs is a by-product of a new process-based characteri-
zationoftoxiceffects. Thebasicideaisthatthehazardrateandtheparametersthat
quantify the energy budget of the individual are proportional to the concentration in
the animal that exceeds the no-effect concentration. The inverse of the proportion-
ality constant, which is called the tolerance concentration, quantifies the toxicity of
the compound. The energy budget parameters are defined by the Dynamic Energy
Budget (DEB) theory, which specifies the rules that organisms use for the energy
vii
viii Preface
uptake of resources (food) and the ensuing allocations to maintenance, growth, de-
velopment and propagation. The DEB theory has been tested against a wide variety
of ecophysiological data. This book only summarizes some relevant points of the
theory, which is fully discussed elsewhere (Kooijman 1993). At high concentrations,
toxicants will affect many physiological processes simultaneously and we would need
many parameters to quantify all these effects. The various physiological processes
can be ordered with respect to their sensitivity to the toxicant, however, so that just
one process (i.e. one parameter) is affected in the lower concentration range.
The characterization of toxic effects is built up in two steps: (i) the uptake and
elimination behaviour of the compound and (ii) the translation of internal concen-
trations into effects. As the most simple option, one-compartment first-order toxi-
cokinetics can be assumed. This results in an nec, a tolerance concentration, and an
elimination rate as characterizations of the various sublethal effects. The tolerance
concentration is replaced by the killing rate for lethal effects. These simple char-
acterizations are independent of exposure time and depend in a simple way on the
octanol-water partition coefficient. The possibility to remove the elimination rate
as an essential parameter makes the models with just two toxicity parameters the
simplest possible. The results of aquatic toxicity bioassays as standardized by the
OECD are suitable substrates for the models. The analysis of these results will be
discussed in detail.
Weconsiderthemethodasanapplicationofthetheoryforenergyandmassfluxes
through biological systems that we are currently developing in the Department of
Theoretical Biology VUA (see our homepage on WWW: http://www.bio.vu.nl/vak-
groepen/thb). The biological foundation is covered by the Dynamic Energy Budget
theory, which has been published separately (see Kooijman 1993, including tests
against experimental data). The Department participates in the Amsterdam Centre
for Environmental Science (ACES).
The book starts with an evaluation of the use of no-effect concentrations in leg-
islation from the perspective of the Dutch Ministry for the Environment (VROM).
Then follows a chapter with a discussion of the toxicological backgrounds of the new
method. It is intended for ecotoxicologists, with an emphasis on scientific aspects
ratherthantechnicalones. Thenextchapterismoretechnicalandisprimarilyaimed
at statisticians and scientists who have had an introduction into applied mathemat-
ics. The sections of this chapter can be read independently and discuss the different
bioassays. The last chapter describes the use of the software package DEBtox, which
handles all calculations that are required to apply the method. A diskette with the
Windows and Unix versions of DEBtox is delivered with this book.
Five persons contributed directly to parts of this book. Kees van Leeuwen and
Jack de Bruijn wrote the introduction. They work at the Dutch Ministry for the
Environment(VROM),DenHaag. MatthijsLuger, whodidagreatjobinthecoding
of DEBtox, taught us a lot about the ins and outs of the programming language C
andaboutthefinerdetailsofWindowsandUnix. HedidhisworkattheDepartment
Preface ix
of Theoretical Biology of the Vrije Universiteit Amsterdam (see final chapter). We
enjoyedthecollaborationwithArnbjørnHanstveitandNielsNyholmonalgalgrowth
very much. Arnbjørn is working at the Institute of Environmental Sciences Delft,
MW-TNO (presently reorganized as TNO-nutrition Zeist); Niels is working at the
InstituteofEnvironmentalSciences&Engineering, TechnicalUniversityofDenmark
in Lyngby.
Many other people have contributed to this work; too many to thank them in-
dividually. We are especially grateful to those who allowed us to use their data
as examples to develop our method, colleagues who discussed the technical issues
on data analysis with us, and the OECD (Nicky Grandy) and the Dutch Ministry
for the Environment (VROM, Jack de Bruijn and Cees van Leeuwen) for their very
positive attitude and their stimulating efforts. Mike Newman, Spliid Hendrik, Jos´e
Tarazona, Nico van Straalen, Wolfgang Bo¨deker, Reinhard Meister and Hugo van
den Berg made many helpful comments on the manuscript. Theo Helder (VROM)
gave valuable advice on Good Laboratory Practice and related quality standards as
applied to DEBtox.
We are also very grateful to the people who tested a pre-release of DEBtox: Tom
Aldenberg (RIVM), Jens Andersen (IMM, TU-Denmark), Jack de Bruijn (VROM),
Cees van Gestel (VU-Amsterdam), Arnbjørn Hanstveit (TNO-Nutr.), Udo Hommen
(TU-Aachen), Frans Leeuwerik (VU-Amsterdam), David Mauriello (US-EPA), Hans
Toni Ratte (TU-Aachen), Theo Traas (RIVM), Manon Vaal (RIVM), Wouter Vaes
(RITOX, Utrecht University), Jos van Wezel (VU-Amsterdam), Maurice Zeeman
(US-EPA).
Last but not least we want to thank Mari¨elle Bedaux and Truus Meijer for their
support and patience during the many evenings and weekends, when office hours
proved to be insufficient to meet the time schedule.
This work has been financially supported by VROM and Grant PAD 90-18 from
the Alternative to Animal Experiments Platform.
Bas Kooijman
Jacques Bedaux
Dept of Theoretical Biology
Vrije Universiteit
Amsterdam
x Preface