Modeling and Simulation in Ecotoxicology with Applications in MATLAB and Simulink Kenneth R. Dixon 2024 Scribd Download

Download the full version of the ebook at
https://ebookfinal.com
Modeling and Simulation in Ecotoxicology with

Applications in MATLAB and Simulink Kenneth
R. Dixon
https://ebookfinal.com/download/modeling-and-
simulation-in-ecotoxicology-with-applications-in-
matlab-and-simulink-kenneth-r-dixon/
Explore and download more ebook at https://ebookfinal.com

Recommended digital products (PDF, EPUB, MOBI) that
you can download immediately if you are interested.
System Simulation Techniques with MATLAB and Simulink 1st

Edition Dingyü Xue
https://ebookfinal.com/download/system-simulation-techniques-with-
matlab-and-simulink-1st-edition-dingyu-xue/
ebookfinal.com
Modeling and Simulation of Systems Using MATLAB and

Simulink 1 Har/Cdr Edition Devendra K. Chaturvedi
https://ebookfinal.com/download/modeling-and-simulation-of-systems-
using-matlab-and-simulink-1-har-cdr-edition-devendra-k-chaturvedi/
ebookfinal.com
Computational Nanotechnology Modeling and Applications

with MATLAB R Sarhan M Musa
https://ebookfinal.com/download/computational-nanotechnology-modeling-
and-applications-with-matlab-r-sarhan-m-musa/
ebookfinal.com
Signals and Systems with MATLAB Computing and Simulink

Modeling 3rd Edition Steven T. Karris
https://ebookfinal.com/download/signals-and-systems-with-matlab-
computing-and-simulink-modeling-3rd-edition-steven-t-karris/
ebookfinal.com
Introduction to Modeling and Simulation with MATLAB and
Python 1st Edition Steven I. Gordon
https://ebookfinal.com/download/introduction-to-modeling-and-
simulation-with-matlab-and-python-1st-edition-steven-i-gordon/
ebookfinal.com
Electronically Scanned Arrays MATLAB Modeling and

Simulation 1st Edition Arik D. Brown
https://ebookfinal.com/download/electronically-scanned-arrays-matlab-
modeling-and-simulation-1st-edition-arik-d-brown/
ebookfinal.com
Numerical Simulation of Optical Wave Propagation With

Examples in MATLAB Jason D. Schmidt
https://ebookfinal.com/download/numerical-simulation-of-optical-wave-
propagation-with-examples-in-matlab-jason-d-schmidt/
ebookfinal.com
Modeling and Simulation in Polymers 1st Edition

Purushottam D. Gujrati
https://ebookfinal.com/download/modeling-and-simulation-in-
polymers-1st-edition-purushottam-d-gujrati/
ebookfinal.com
Agent based modeling and simulation with Swarm 1-st

Edition Hitoshi Iba
https://ebookfinal.com/download/agent-based-modeling-and-simulation-
with-swarm-1-st-edition-hitoshi-iba/
ebookfinal.com
Modeling and Simulation in Ecotoxicology with
Applications in MATLAB and Simulink Kenneth R.
Dixon Digital Instant Download
Author(s): Kenneth R. Dixon
ISBN(s): 9781439855171, 143985517X
Edition: Har/Cdr
File Details: PDF, 7.31 MB
Year: 2011
Language: english
Modeling and
Simulation in
Ecotoxicology
with Applications
in MATLAB®
and Simulink®
Modeling and
Simulation in
Ecotoxicology
with Applications
in MATLAB®
and Simulink®
Kenneth R. Dixon
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2012 by Taylor & Francis Group, LLC

CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20110520
International Standard Book Number-13: 978-1-4398-5518-8 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts have been made to
publish reliable data and information, but the author and publisher cannot assume responsibility for the validity of all materials
or the consequences of their use. The authors and publishers have attempted to trace the copyright holders of all material repro-
duced in this publication and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmitted, or utilized in any
form by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying, microfilming,
and recording, or in any information storage or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.com (http://www.copy-
right.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923, 978-750-8400.
CCC is a not-for-profit organization that provides licenses and registration for a variety of users. For organizations that have been
granted a photocopy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identifica-
tion and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at

http://www.crcpress.com
In Memory of Dorothy Phipps and Mildred Wanninger
Contents
Preface............................................................................................................................................ xiii
Acknowledgments............................................................................................................................ xv
About the Author ...........................................................................................................................xvii
Chapter 1 Introduction ..................................................................................................................1

1.1 Theories Underlying Predictive Models............................................................1
1.3 Reasons for Modeling and Simulation ..............................................................2
1.2.1 Alternatives and Their Consequences..................................................3
1.2.2 Relative Predictive Ability ...................................................................3
1.2.3 Instruction ............................................................................................3
1.2.4 Hypothesis and Theory Construction ..................................................3
1.2.5 Nonexistent Universes.......................................................................... 4
1.2.6 Cost.......................................................................................................4
1.2.7 Planning and Management Decision Aid.............................................4
1.2.8 System Identification ............................................................................4
1.2.9 Unanticipated Effects ...........................................................................5
1.3 What Does It Take To Be a Modeler? ...............................................................5
1.4 Why Models Fail: A Cautionary Note ..............................................................6
1.4.1 Poor Data for Parameter Estimation ....................................................6
1.4.2 Uncertainty Not Considered.................................................................6
1.4.3 Bias (Political, Social, Economic)........................................................6
1.4.4 Lack of Understanding of Real-World Systems ................................... 6
1.4.5 Misuse of Mathematics ........................................................................7
References ....................................................................................................................7
Chapter 2 Principles of Modeling and Simulation........................................................................ 9

2.1 Systems..............................................................................................................9
2.1.1 Definition..............................................................................................9
2.1.2 System Input and Output......................................................................9
2.1.3 Control Systems....................................................................................9
2.1.4 Feedback............................................................................................. 10
2.1.5 System States: Steady State versus Transient States .......................... 10
2.1.6 Discrete versus Continuous ................................................................ 10
2.1.7 Linear versus Nonlinear..................................................................... 11
2.2 Modeling........................................................................................................... 11
2.2.1 Equations............................................................................................ 11
2.2.1.1 Solution of Ordinary First-Order Differential Equations... 14
2.2.1.2 Steady-State and Transient Response ................................. 16
2.2.1.3 Difference Equation Approximation to Differential
Equation.............................................................................. 16
2.2.1.4 Numerical Solutions to Differential Equations .................. 17
2.2.2 Block Diagrams.................................................................................. 18
2.2.3 Stochastic Models............................................................................... 21
2.2.4 Individual-Based Models ................................................................... 21
2.2.5 Aggregated Models ............................................................................ 21
vii
viii Contents
2.3 Simulation........................................................................................................ 21
2.3.1 Principles of Simulation ..................................................................... 23
2.3.1.1 Principle of Communication............................................... 23
2.3.1.2 Principle of Modularity ...................................................... 23
2.3.1.3 A Modified Principle of Parsimony.................................... 23
2.3.2 Steps in Simulation............................................................................. 23
2.3.2.1 Problem Definition...............................................................24
2.3.2.2 Model Development............................................................24
2.3.2.3 Model Implementation........................................................24
2.3.2.4 Data Requirements ............................................................. 27
2.3.2.5 Model Validation ................................................................ 27
2.3.2.6 Design of Simulation Experiments.....................................28
2.3.2.7 Analyze Results of Simulation Experiments ......................28
2.3.2.8 Presentation and Implementation of Results ......................28
References ..................................................................................................................28
Chapter 3 Introduction to MATLAB® and Simulink®................................................................ 31

3.1 MATLAB ........................................................................................................ 31
3.1.1 Matrix Algebra ................................................................................... 32
3.1.2 Data Input...........................................................................................34
3.1.3 Solving Equations................................................................................ 37
3.1.4 Saving Data ........................................................................................ 39
3.1.5 Plotting Data.......................................................................................40
3.2 Simulink .......................................................................................................... 43
Exercises.....................................................................................................................44
References .................................................................................................................. 45
Chapter 4 Introduction to Stochastic Modeling .......................................................................... 47

4.1 Introduction to Probability Distributions ........................................................ 47
4.2 Example Probability Distributions .................................................................. 50
4.2.1 Continuous Distributions..................................................................... 50
4.2.1.1 Uniform............................................................................... 50
4.2.1.2 Exponential......................................................................... 51
4.2.1.3 Gamma ............................................................................... 51
4.2.1.4 Weibull................................................................................ 53
4.2.1.5 Normal ................................................................................ 53
4.2.1.6 Lognormal .......................................................................... 54
4.2.1.7 Beta ..................................................................................... 54
4.2.1.8 Triangular ........................................................................... 56
4.2.1.9 Logistic ............................................................................... 56
4.2.2 Discrete Distributions.......................................................................... 58
4.2.2.1 Bernoulli ............................................................................. 58
4.2.2.2 Binomial ............................................................................. 59
4.2.2.3 Discrete Uniform ................................................................60
4.2.2.4 Geometric ...........................................................................60
4.2.2.5 Negative Binomial .............................................................. 62
4.2.2.6 Poisson ................................................................................ 62
4.2.3 Empirical Distributions ......................................................................64
4.3 Discrete-State Markov Processes .................................................................... 65
Contents ix
4.4 Monte Carlo Simulation .................................................................................. 69

Exercises..................................................................................................................... 71
References .................................................................................................................. 72
Chapter 5 Modeling Ecotoxicology of Individuals..................................................................... 73

5.1 Toxic Effects on Individuals............................................................................ 73
5.1.1 The Dose–Response Relationship...................................................... 73
5.1.1.1 Quantal Response ............................................................... 73
5.1.1.2 Graded Response ................................................................ 78
5.1.2 Toxicokinetics .................................................................................... 78
5.1.3 Physiological Processes...................................................................... 79
5.1.3.1 Uptake................................................................................. 79
5.1.3.2 Absorption .......................................................................... 82
5.1.3.3 Distribution ......................................................................... 86
5.1.3.4 Excretion............................................................................. 87
5.1.4 Biological Processes........................................................................... 88
5.1.4.1 Reproduction....................................................................... 88
5.1.4.2 Growth ................................................................................90
5.1.4.3 Death................................................................................... 95
5.1.4.4 Movement ........................................................................... 98
5.1.4.5 Homeostasis...................................................................... 100
Exercises................................................................................................................... 105
References ................................................................................................................ 106
Chapter 6 Modeling Ecotoxicology of Populations, Communities, and Ecosystems............... 109

6.1 Effects of Toxicants on Aggregated Populations............................................ 109
6.2 Effects of Toxicants on Age-Structured Populations .................................... 113
6.3 Effects of Toxicants on Communities ........................................................... 115
6.4 Effects of Toxicants on Ecosystems .............................................................. 119
Exercises................................................................................................................... 123
References ................................................................................................................ 124
Chapter 7 Parameter Estimation ............................................................................................... 125

7.1 Linear Regression.......................................................................................... 125
7.1.1 Function: regress ......................................................................... 126
7.1.2 Function: polyfit.......................................................................... 129
7.1.3 Function: regstats....................................................................... 132
7.2 Nonlinear Regression .................................................................................... 136
7.2.1 Function: nlinfit.......................................................................... 136
7.3 Comparison between Linear and Nonlinear Regressions ............................. 144
Exercises................................................................................................................... 145
References ................................................................................................................ 145
Chapter 8 Designing Simulation Experiments ......................................................................... 147

8.1 Factorial Designs ........................................................................................... 147
8.1.1 Full Factorial Designs ...................................................................... 147
8.1.2 Fractional Factorial .......................................................................... 149
x Contents
8.2 Response Surface Designs............................................................................. 151

8.2.1 Central Composite Designs.............................................................. 152
8.2.2 Box-Behnken Designs ...................................................................... 155
Exercises................................................................................................................... 158
References ................................................................................................................ 158
Chapter 9 Analysis of Simulation Experiments........................................................................ 159

9.1 Simulation Output Analysis........................................................................... 159
9.1.1 Types of Simulations ........................................................................ 159
9.1.2 Output Analysis Methods................................................................. 159
9.2 Stability Analysis........................................................................................... 162
9.2.1 Linear Systems ................................................................................. 163
9.2.2 Nonlinear Systems ........................................................................... 165
9.2.3 Relative Stability .............................................................................. 165
9.2.4 Resilience ......................................................................................... 166
9.3 Sensitivity Analysis ....................................................................................... 166
9.4 Response Surface Methodology .................................................................... 168
Exercises................................................................................................................... 173
References ................................................................................................................ 174
Chapter 10 Model Validation...................................................................................................... 175

10.1 Validation and Reasons for Modeling and Simulation.................................. 175
10.2 Testing Hypotheses........................................................................................ 176
10.2.1 Accept the Null Hypothesis When It Is True ................................... 177
10.2.2 Reject the Null Hypothesis When It Is True .................................... 177
10.2.3 Accept the Null Hypothesis When It Is False .................................. 177
10.2.4 Reject the Null Hypothesis When It Is False ................................... 177
10.2.5 Accept the Null Hypothesis When It Is True ................................... 178
10.2.6 Reject the Null Hypothesis When It Is True .................................... 178
10.2.7 Accept the Null Hypothesis When It Is False .................................. 178
10.2.8 Reject the Null Hypothesis When It Is False ................................... 179
10.3 Statistical Techniques .................................................................................... 179
10.4 Some MATLAB Methods ............................................................................. 180
10.4.1 Paired t-test....................................................................................... 180
10.4.2 Wilcoxon Nonparametric Signed Rank Test.................................... 180
10.4.3 Linear Regression............................................................................. 183
10.4.4 Theil’s Inequality Coefficient........................................................... 184
10.4.5 Analysis of Variance ........................................................................ 186
10.4.6 Kruskal-Wallis Nonparametric ANOVA ......................................... 186
Exercises................................................................................................................... 189
References ................................................................................................................ 190
Chapter 11 A Model to Predict the Effects of Insecticides on Avian Populations ..................... 191
11.1 Problem Definition ........................................................................................ 191
11.2 Model Development........................................................................................ 191
11.3 Model Implementation .................................................................................. 191
11.3.1 Model Description............................................................................ 192
11.3.1.1 Ingestion in Food .............................................................. 192
Contents xi
11.3.1.2 Consumption of Chlorpyrifos Granules ........................... 193

11.3.1.3 Avian Loss Rates .............................................................. 194
11.3.1.4 Mortality ........................................................................... 195
11.3.2 Model Structure Validation.............................................................. 195
11.3.3 Programming the Computer Code ................................................... 196
11.4 Data Requirements ........................................................................................ 196
11.4.1 Ingestion ........................................................................................... 196
11.4.1.1 Proportion of Components in Diet.................................... 196
11.4.1.2 Granule Consumption Rate .............................................. 196
11.4.1.3 Time Spent in Treated Areas ............................................ 197
11.4.1.4 Residues in Diet Components........................................... 198
11.4.2 Avian Loss Rates.............................................................................. 199
11.4.3 Mortality............................................................................................ 199
11.5 Model Validation ...........................................................................................200
11.6 Design Simulation Experiments....................................................................200
11.7 Analyze Results of Simulation Experiments................................................. 201
11.7.1 Predicted Dose ................................................................................. 201
11.7.1.1 Ring-Necked Pheasant...................................................... 201
11.7.1.2 Northern Bobwhite ........................................................... 201
11.7.1.3 Red-Winged Blackbird ..................................................... 201
11.7.1.4 House Sparrow..................................................................203
11.7.2 Predicted Mortality .......................................................................... 203
References ................................................................................................................207
Chapter 12 Case Study: Predicting Health Risk to Bottlenose Dolphins

from Exposure to Oil Spill Toxicants........................................................................209
12.1 Problem Definition ........................................................................................209
12.2 Model Development........................................................................................209
12.3.1 Differential Equations ...................................................................... 211
12.5 Model Validation ........................................................................................... 217
12.6 Design of Simulation Experiments................................................................ 217
12.7.1 Simulation Output ............................................................................ 217
12.7.2 Sensitivity Analysis .......................................................................... 219
12.8 Presentation and Implementation of Results ................................................. 220
References ................................................................................................................ 222
Chapter 13 Case Study: Simulating the Effects of Temperature Plumes

on the Uptake of Mercury in Daphnia..................................................................... 223
13.1 Problem Definition ........................................................................................ 223
13.2 Model Development........................................................................................ 223
13.4.1 Plot Data ........................................................................................... 225
13.4.2 Plot Edited Data ............................................................................... 226
13.4.3 Estimate Model Parameters ............................................................. 226
13.4.4 Gross Uptake Model.......................................................................... 227
xii Contents
13.4.5 Estimate Parameters for Gross Uptake Model................................. 228

13.4.6 Differential Equation for Mercury Dynamics.................................. 230
13.4.7 Parameters as Functions of Temperature ......................................... 230
13.4.8 Estimate Thermal Plume Temperatures........................................... 231
13.5 Model Validation ........................................................................................... 233
13.6 Design of Simulation Experiments................................................................ 233
13.8 Presentation and Implementation of Results ................................................. 235
References ................................................................................................................ 237
Preface
This book is about the role of modeling and simulation in environmental toxicology. It covers the
steps in modeling and simulation from problem conception to validation and simulation analy-
sis. Examples of mathematical functions and simulations are presented using the MATLAB® and
Simulink® programming languages. We have proposed including this text in the MATLAB book
series. The main themes are how to develop mathematical models and run computer simulations of
the effects of toxic agents on biological and ecological processes using MATLAB software. This
book is designed as a textbook for advanced undergraduate and graduate courses in the field of envi-
ronmental toxicology. We try to present the modeling in a rigorous, yet easy-to-understand, manner.
The book can be used in courses with students who have little or no experience in modeling. We also
include MATLAB m-files or Simulink block diagrams for most examples in the text and on a CD.
Although the methodology emphasizes environmental toxicology, it can be applied rather easily to
other biological fields.
Chapter 1 is intended to introduce the student to the use of models in general and environmental
toxicology in particular. It describes how modeling and simulation play a role in the broader con-
text of ecological research. The introduction does not include equations to avoid apprehension on
the part of students with little quantitative experience. It also sets the stage for the types of models
covered in the text. Chapter 2 presents the general principles of modeling and simulation based
upon existing literature and the author’s forty years of experience. The steps in modeling and simu-
lation are described, including parameter estimation, experimental design, analysis of simulation
experiments, and validation, which are explored more fully in later chapters. Chapter 3 describes
the foundation for our modeling and simulation, which are the programming languages, MATLAB
and Simulink. These are widely used in a variety of disciplines because of their wide range of
functions and a history of software verification. We present a brief introduction to both MATLAB
and Simulink and then cover the functions used in subsequent chapters. Chapter 4 introduces sto-
chastic modeling where variability and uncertainty are acknowledged by making parameters ran-
dom variables. Parameter values can be drawn from a wide range of probability distributions as
built-in MATLAB functions. We also describe probabilistic models such as Markov chains. The
methodology of Monte Carlo simulation also is described. Chapter 5 describes toxicological pro-
cesses from the level of the individual organism. We include worked examples of process models in
either MATLAB or Simulink or both. The model descriptions include MATLAB code (m-files) or
Simulink block diagrams.
Chapter 6 describes toxicological processes at the level of populations, communities, and eco-
systems. Worked examples of population and ecosystem models in MATLAB and Simulink are
included. Chapter 7 presents parameter estimation using least squares regression methods. The
advantages and disadvantages of linear and nonlinear regression are discussed. Examples of both
techniques are presented using MATLAB. Chapter 8 presents the design of simulation experiments
similar to the experimental design applied to laboratory or field experiments. The emphasis is on
identifying significant parameters and reducing the number of simulation experiments using frac-
tional factorial designs. Examples using MATLAB are presented, including response surface meth-
ods. Chapter 9 describes several methods of postsimulation analysis, including stability analysis
and sensitivity analysis. Stability measures are presented for the transient response to a unit impulse
function. Sensitivity analysis is described as the relative change in state variables to changes in
parameter values. Examples in MATLAB are presented. Chapter 10 presents the complex and con-
troversial topic of model validation. We present a consensus view but discuss the different levels
of validation and how these are related to modeling purpose. Examples of statistical methods of
validation are included.
xiii
xiv Preface
Chapter 11 presents a case study of a model developed to assess the relative risk of mortality
following exposure to insecticides in different avian species. Many of the toxicological processes
described in Chapter 5 are included in the model. The steps in the modeling and simulation are
described for this model. Chapter 12 is a case study designed to explore the role of diving behavior
on the inhalation and distribution of naphthalene in bottlenose dolphins. The model is an example
of physiologically based toxicokinetic models. The case study in Chapter 13 looks at the dynamics
of mercury in Daphnia that are exposed to simulated thermal plumes from a hypothetical power
plant cooling system. Differences in ambient water temperature and cooling conditions are explored
in the simulations.
Kenneth R. Dixon
Texas Tech University
Lubbock, Texas
MATLAB® and Simulink® are registered trademarks of The MathWorks, Inc. For product informa-
tion, please contact:
The MathWorks, Inc.

3 Apple Hill Drive
Natick, MA 01760-2098 USA
Tel: 508-647-7000
Fax: 508-647-7011
E-mail: [email protected]
Web: www.mathworks.com
Acknowledgments
As I argue in Chapter 1, modeling and simulation require at least a modicum of math and computer
programming skills as well as a strong foundation in the basic science discipline (in this text, its
ecotoxicology). It is unusual for someone to have all those skills to the degree necessary. In my
case, I have spent time in both the quantitative and the natural science camps and can attest to the
fact that these are not easy disciplines to master. Also, I have had the good fortune to have been
associated with, and mentored by, some of the best, who have imparted some of their wisdom in
these matters.
Early in my years in graduate school, I was fortunate to take a course in systems ecology
taught by one of its founders, George M. Van Dyne. This course piqued my interest in apply-
ing systems analysis techniques to environmental issues. I was then encouraged to develop the
necessary foundations in math, statistics, and systems engineering by my advisor George W.
Cornwell at the University of Florida. In this effort, I was supported with patience and encour-
agement by Richard L. Patterson. It was Dr. Patterson who invited me to the University of
Michigan to pursue my Ph.D. in natural resource systems in the School of Natural Resources
and the Environment. It was he, and my advisor at SNRE, John Kadlec, who provided unlimited
encouragement and guidance, which gave me the confidence to continue the modeling and simu-
lation aspects of my career.
As a member of the Systems Ecology Group in the Environmental Sciences Division at Oak
Ridge National Laboratory, I was inspired to further advance my skills in applying systems analy-
sis to environmental problems. Members of this group routinely shared ideas and collaborated on
projects involving modeling and simulation. Particularly helpful in this regard were Don DeAngelis,
Hank Schugart, Robert O’Neill, J. B. Mankin, Bob Goldstein, Jerry Olsen, and M. R. Patterson.
At the University of Maryland, Joseph A. Chapman encouraged my development of a gradu-
ate course that became the precursor of the course for which this book was written. It was also
his encouragement that led to several quantitative methods in the analysis of radio-telemetry data.
Thanks go to Ron Kendall, director of The Institute of Environmental and Human Health, for pro-
viding support for the completion of this book. I also thank Lenwood Hall, John Huckabee, Rami
Naddy, and Jennifer Gottschalk Walters for the use of their data.
I have also been fortunate to have many outstanding graduate students who have not only devel-
oped their own skills in modeling and simulation, but improved mine as well. These include Eric
Albers, Randy Apodaca, Pinar Dogru, Doug Florian, William Henriques, Dan Jacobs, Min Lian,
Smita Sathe, Lori Sheeler, and Fred Snyder. Eric Albers and Lori Sheeler developed some of the
MATLAB programs in the book. I have modified most of them and any programming errors are
my own. Everyone should have a colleague to provide the expertise that one is lacking. In my case,
that person was Sam Anderson, who taught me most of what I have learned about programming in
MATLAB and Simulink. I thank all the technical support staff at The MathWorks, who too often
had to correct my programming errors or provide the solutions to programming problems.
Finally, I gratefully acknowledge the support of my family—my wife, Sheila, and Buster, Cybele,
and Beauregard, without whose loving support over the past fifteen years, the completion of this
book would not have been possible.
Kenneth R. Dixon
Texas Tech University
Lubbock, Texas
xv
About the Author
Kenneth R. Dixon is professor in the Department of
Environmental Toxicology and The Institute of Environ
mental and Human Health at Texas Tech University. He
received his B.S. degree in forestry from the University of
Florida in 1964 and his M.S. in forestry in 1968, also from
the University of Florida, specializing in statistics and sys-
tems engineering. From 1968 to 1971, Dr. Dixon worked as
a biometrician in the Institute of Statistics at North Carolina
State University. In 1974, he received his Ph.D. from the
School of Natural Resources at the University of Michigan.
After graduating, he took a postdoctoral position as an
ecologist and modeler at Oak Ridge National Laboratory.
His research primarily involved modeling the impact of
heavy metals on both aquatic and terrestrial ecosystems.
Additional activities included the environmental impact
assessment of nuclear power plants. In 1976, Ken joined
the University of Maryland Appalachian Environmental
Laboratory, where he taught courses in quantitative
methods in wildlife management. His research included furbearer population dynamics, wild-
life as monitors of environmental contamination, and spatial models of wildlife behavior.
From 1984 to 1992, Dr. Dixon headed the Wildlife Research Program in the Washington
Department of Wildlife. Research activities primarily involved the use of radio telemetry, remote
sensing, and geographic information systems to determine wildlife habitat requirements. Species
studied included the peregrine falcon, spotted owl, pygmy rabbit, harbor seal, California sea lion,
grizzly bear, gray wolf, elk, and mule deer. In 1985 he planned and implemented the department’s
geographic information system, which he then supervised until leaving the department in 1992.
Under Ken’s supervision, a statewide wildlife habitat mapping project was initiated. His home
range analysis program was written into Arc/GIS to facilitate the calculation of habitat selection.
Ken also assisted the U.S. Fish and Wildlife Service in implementing the Washington State Gap
Analysis project.
Dr. Dixon was associate professor in the Department of Environmental Toxicology and the
Institute of Wildlife and Environmental Toxicology at Clemson University from 1992 to 1997. His
research included developing and applying computer simulation models to predict the movement
and effects of toxic chemicals on wildlife populations and the environment, including the spatial
distribution of toxicants and effects at ecosystem, landscape, and regional scales using geographic
information systems. An example of research in this area was a study of topographic, soil, and
weather parameters affecting the runoff of pesticides into farm ponds in Iowa and Illinois.
His current research interests include developing and applying computer simulation models to
predict the movement of toxic chemicals in the environment and their effects on human and wildlife
populations. Dr. Dixon also studies the spatial distribution of toxicants and effects at ecosystem,
landscape, and regional scales by integrating models with geographic information systems. Current
research projects include developing terrestrial food-chain models to predict the uptake and effects
of pesticides, perchlorate, and explosives, and developing spatial models of the spread of infectious
diseases, and a real-time model of exposure and effects of atmospheric pollutants. Dr. Dixon has
taught courses in modeling, geographic information systems, ecosystems analysis, biometry, and
wildlife management.
xvii
1 Introduction
In a text on modeling and simulation, it is important to define these terms as there is a wide range
of usage for both. In fact, the two terms are often confused or used interchangeably. Some authors
have used the term simulation model, which makes it necessary to define both model and simula-
tion to distinguish simulation models from other types of models. We make a distinction between
modeling and simulation to clarify our discussion of both terms. A model can be defined as a
simplified abstract of a real-world object. For our purposes, modeling has the following quantita-
tive definition:
Definition 1.1: Modeling is the process of using abstract mathematical representations of a system
for analyzing or studying the relationships among the components of the system.
Although related to modeling, simulation has more to do with what we do with a model after it is
constructed. A formal definition of simulation appears as Definition 1.2.
Definition 1.2: Simulation is the process of using a computer to exercise a model for the purpose of
mimicking the behavior of a real system.
Having defined modeling, we could ask, “Why do modeling?” Everyone uses an implicit or “mental”
model to consider the future consequences of today’s actions (Gentner and Stevens 1983, Morgan et
al. 2002). For example, consider the impacts of the release of a toxic chemical into the environment.
One can imagine where the chemical may be transported, what plants and animals may be exposed,
and what the effects of that exposure are. This mental model can be thought of as the first step in
a more explicit model. The question then is not whether we should do modeling, but what kind of
model we will use. An explicit model, one based upon quantitative descriptions and data, provides
a more structured approach than a mental model. As put forward by the late wildlife statistician,
Douglas Chapman, “the use of mathematical language and mathematical models introduces and
forces clarity of ideas that may be otherwise lacking” (Chapman 1971, 429).
Although modeling and simulation are relatively new to environmental toxicology, there is a
long and productive history of modeling in other biological disciplines, including population biol-
ogy, ecology, and wildlife management. These integral disciplines usually are described as either
mathematical biology (emphasizing biology), or biomathematics (emphasizing mathematics). Other
biological disciplines that have a history of incorporating the use of models include biometrics,
cybernetics, and systems ecology.
1.1 THEORIES UNDERLYING PREDICTIVE MODELS

The theories behind different types of models are based on a combination of biology, toxicology,
mathematics, and statistics (Figure 1.1). Except for the time series models, which truly are naïve
models (i.e., they contain no biological information), all of the models contain some biological or
toxicological process and mathematical or statistical structure. Those models on the left side of
Figure 1.1 emphasize the biological mechanism and those on the right side emphasize the statis-
tical structure. Typical mechanistic models are time dependent and are described by difference
1
2 Modeling and Simulation in Ecotoxicology with Applications in MATLAB® and Simulink®
Disciplinary Logic and Theory of

Knowledge of Mathematics Probability
Physical
Mechanisms
Theory of
Theory of
Random
Statistics
Processes
Mechanistic Design and Statistical

Predictive Analysis of Analysis of Time
Model Experiments Series Data
Stochastic Statistical-
Probability Time-Series
Differential- Mechanistic
Mass Balance Predictive
Difference Predictive
Equations Model
Equations Model
FIGURE 1.1 Theories underlying predictive models (R. L. Patterson, personal communication).
or differential equations. Statistical models typically are static and include regression, principle
component, or Bayesian structures. Statistical-mechanistic models usually have a difference or dif-
ferential equation structure but rely on statistical analysis of experimental data for parameterization.
Stochastic differential or difference equations can take several forms. For our purposes, the models
are referred to as stochastic or probabilistic in that some (or all) of the parameters in the model are
random variables. The emphasis in this text is on predictive models using either stochastic differen-
tial or stochastic difference equations.
There are advantages and disadvantages to mechanistic models. One advantage is that the
parameters in a mechanistic model usually have real-world counterparts. These parameters then
are amenable to sensitivity analyses to determine the relative impact each parameter has on the
process. Parameters in statistical models, however, have no biological meaning attached to them.
Sensitivity analysis of these model parameters would not provide any added knowledge about
the process. Largely because mechanistic models reflect physical and chemical dynamics of the
process being modeled, they can predict future behavior of the process more accurately than sta-
tistical models can. Disadvantages of the mechanistic modeling approach are the relatively high
cost and the amount of time required to conduct the necessary experiments to estimate the model
parameters.
1.3 REASONS FOR MODELING AND SIMULATION

There are many reasons given for doing modeling and simulation. Generally, most reasons for mod-
eling and simulation involve some aspect of prediction designed to provide insight or greater under-
standing of system behavior. Particularly, for our interest, we can say that the objective of modeling
and simulation is to study the behavior of systems impacted by toxic compounds. Within that gen-
eral purpose, there are several reasons to perform simulation studies. The reasons in the following
sections provide a synthesis of most compilations (Hermann 1967, Martin 1968, Epstein 2008).
Introduction 3
1.2.1 Alternatives and Their Consequences

A model may be used to predict the consequences of perturbations on a given system. A particular
disturbance can be built into a model and the resulting effect on the system can be predicted. A
modeler is not limited, however, to the real-world conditions reflected in model development. There
are parameters that cannot be controlled very easily in the real-world system but can be controlled
in the model. Once a model has been developed, parameters in the model can be changed to pre-
dict the effects of the changes on the real-world system. For example, a model may be designed to
predict the impact of cleaning up a contaminated waste site. The cleanup would reduce the level of
contamination and wildlife exposure but also could destroy important wildlife habitat. The model
could predict the effects on habitats and animal body burdens for alternative levels of contaminant
removal. Analysis of the results could be used to determine the optimal level of cleanup. This type
of simulation experiment also can provide insight into system behavior in the real world. Simulation
experiments, combined with laboratory and field studies, should provide greater understanding of
the real-world system than either type of experiment conducted alone.
1.2.2 Relative Predictive Ability

The purpose of modeling in this case involves predicting system behavior in a relative sense as
opposed to an absolute sense. That is, the model is used to generate outcomes of effects of policy
decisions as before, but not to predict system behavior accurately and with precision. Rather, the
model is used as one tool for making predictions. Additional predictions can be made by other types
of models (see Figure 1.1), including implicit models. Taking the previous example of cleanup of a
contaminated waste site, a decision maker may have alternative models available such as statistical
models. He or she also may have considerable experience in waste site cleanup and as a result have
an intuitive understanding of the likely effects. The new model then would be used to provide addi-
tional weight of evidence for a particular management decision.
1.2.3 Instruction
Models have been used for teaching students about real-world systems. Students who have had
courses that use simulation as a teaching tool often learn more than students who have taken com-
parable courses without a modeling component (Rieber et al. 2004, Windschitl and Andre 1998).
Models that are used in instruction need to be as representative of real-world systems as possible
to assure that students do not incorrectly draw conclusions about how systems function or what
impact toxic substances will have on those systems. This will require that parameters with real-
world counterparts have realistic values and in the case of parameters that vary during a simulation,
they should be sampled from appropriate probability distributions. Simplified models that do not
completely mimic real-world systems still can be useful teaching tools if the purpose is to explore
theoretical concepts rather than to understand system behavior.
1.2.4 Hypothesis and Theory Construction

Models can be used not only to make predictions about a system that functions within a known
range of parameter values; a model also can be used to ask “what if” questions about system behav-
ior. Simulations with “hypothetical” parameter values or changing relationships among variables
in the model can lead to hypotheses about how a system functions. These hypotheses then can be
tested by conducting laboratory or field experiments. After a number of successful predictions, a
new theory of system behavior might be constructed. Sometimes a model may predict an outcome
that runs counter to conventional wisdom or is seen as questioning authority. If the model is found to
be accurate, a new hypothesis or theory could be developed. As an example, it is generally assumed
that plants will accumulate atmospheric pollutants until the plant concentration reaches equilibrium
equal to the atmospheric concentration. A mechanistic model of the uptake of atmospheric pollut-
ants, used to predict the accumulation of radioactive carbon, 14C, showed an accumulation much
higher than the atmospheric concentration (Killough et al. 1976, Dixon and Murphy 1979). The
reason for the difference was that the assumption of equilibrium failed to consider that carbon is
fixed by the plant during photosynthesis.
1.2.5 Nonexistent Universes

We have, so far, discussed the uses of models of existing systems, although we might simulate con-
ditions not observed. We also might wish to make predictions about behavior of systems that cur-
rently do not exist or may exist in the future. Models of systems that do not exist can be based upon
models of existing systems if the two systems are known not to differ significantly. This requires
the assumption that model parameters accurately represent comparable variables in the nonexistent
system. Systems that we may consider creating in the future could include construction of a wetland
for nutrient removal and applying new methods of remediating contaminated sites.
1.2.6 Cost
Simulations can reduce both the monetary and time expenditures of experiments. If we are inter-
ested in knowing the effects of a toxicant on an ecosystem, one approach would be to introduce the
contaminant into the system and then measure the effects. The resulting cost, however, could be
unacceptable, both in terms of damage to the system and the cost of conducting the experiment.
Simulation allows the experiment to be conducted a priori on the computer without the risk of
environmental damage. Simulation experiments can be run in a fraction of the time it takes to run
laboratory or field experiments. Most computer simulations take only a few seconds or minutes on
modern computers, whereas laboratory or field experiments can take weeks or months.
1.2.7 Planning and Management Decision Aid

Planners and managers need the best information available to make optimal decisions. Whether
the question being addressed concerns the impact of introducing a new pollutant into a system or
the best method of removing existing contaminants from a system, simulation can aid in providing
information on the long-term effects of alternative decisions. A particular management decision can
be built into a model and the resulting effect on the system can be predicted. A manager or decision
maker may formulate alternative policies to be simulated. They can then evaluate and compare the
consequences of their various management scenarios by analyzing the results of the model simu-
lations. For example, a scientific advisory committee was appointed by the U.S. Environmental
Protection Agency (EPA) to assess the effects of the insecticide dichlorodiphenyltrichloroethane
(DDT). The committee used a food chain model of the bioaccumulation of DDT developed by Oak
Ridge National Laboratory (O’Neill and Burke 1971) to assist in deciding whether to immediately
ban the insecticide or end its use gradually (Holcomb Research Institute 1976).
1.2.8 System Identification

In many systems, there will be a component for which the input and output can be observed, but
the internal structure or mechanisms are not understood. Such a component is referred to as a black
box. By conducting simulation experiments with different component structures, and using model
validation methods, the most likely internal structure may be identified. According to May (2004,
p. 791), “Various conjectures about underlying mechanisms can be made explicit in mathematical
terms, and the consequences can be explored and tested against the observed patterns. In this gen-
eral way, we can, in effect, explore possible worlds.”
Introduction 5
1.2.9 Unanticipated Effects

A really useful reason for modeling and simulation would be if we could predict unexpected events.
Systems, however, can be quite complex with many interconnections among subsystem compo-
nents. Therefore, it is often difficult to anticipate how a perturbation in one subsystem will affect
other subsystem components. This is particularly true of unusual or catastrophic events. These
occur so infrequently in real systems, it makes it difficult to observe and measure their impacts. And
yet, simulation experiments of these extreme events, with model parameters set at extreme values,
may be most important. First, these simulations could determine the level of perturbation that could
cause irreversible damage to the system. And second, investigating system behavior at these thresh-
olds or breakpoints can lead to greater understanding of system behavior.
The purpose for which a model is developed will greatly affect the rigor of the validation required
of the model. We discuss the implications for model validation further in Chapter 10.
1.3 WHAT DOES IT TAKE TO BE A MODELER?

A modeler in environmental toxicology must have a strong foundation in three areas: first, you must
have a strong foundation in the relevant biological toxicological processes; you also must have the
requisite skills in math and statistics; and third, you must have the necessary skills in computer
programming. Readers of this text are expected to have a good foundation in the basic toxicological
sciences and some training in math and statistics. In this text we will cover the math and statistics
techniques that are important in modeling, and give the reader a good foundation in programming,
using the MATLAB® programming language.
This text is not intended for other modelers but for toxicologists and decision makers who wish
to learn something about modeling. The reader also should be able to evaluate existing models and
critically assess simulation results. This one text therefore will not in itself make one a modeler.
What it may do is give the ecological toxicologist a basic understanding of what is involved in mod-
eling in environmental toxicology. It also will make it easier for him or her to communicate with
modelers. And, it may even inspire some toxicologists to develop a modeling knowledge base by
completing the requisite math, statistics, and computer programming courses. For an extensive dis-
cussion on training in biomathematics, see Lucas (1962). If a toxicologist does not possess adequate
knowledge in these disciplines, he or she will have to rely on others, who may not have a toxicology
background, to participate in the modeling process.
The problem of communication between toxicologist and mathematician or statistician also has
a long history and stems primarily from a different way of thinking about and developing models
(van der Vaart 1977). Toxicologists conduct experiments on a process that yield data that can then
be used to infer the structure of a model. This involves inductive reasoning. Mathematicians, on the
other hand, derive their models from first principles and logic using deductive reasoning. There are
advantages and disadvantages to each way of thinking about models. A modeler in environmental
toxicology should be cognizant of the different patterns of thought, especially when trying to com-
municate with mathematicians and statisticians.
What about a modeler who is part of a modeling effort in environmental toxicology? Obviously,
it would be better to have the modeler be knowledgeable in environmental toxicology to reduce
the possibility of errors in communication between toxicologist and modeler. Likewise, it is bet-
ter if a decision maker has some understanding of what is involved in developing models used in
decision making.
The third foundation needed is the ability to write computer programs in some language
that is amenable to modeling and simulation. A more detailed discussion of programming lan-
guages is included in Chapter 2, with an overview of the languages used in this text, MATLAB
and Simulink.
1.4 WHY MODELS FAIL: A CAUTIONARY NOTE

Most cases of model failure involve models that fail to accurately predict future events. Models designed
for other purposes do not receive much attention when they fail, mostly because the potential conse-
quences of using a faulty model are much less than when models fail at making accurate predictions. In
other cases, the failure to consider model predictions has resulted in bridges collapsing, dams and levees
bursting, and toxic waste ponds leaking (Petroski 1992, Pilkey and Pilkey-Jarvis 2007).
1.4.1 Poor Data for Parameter Estimation

One reason models fail is the use of poor data for parameter estimation. The expression “garbage-in,
garbage-out” is more than a catchy phrase. It accurately describes the results of using inappropriate
data for estimating system initial conditions or model parameters. Data can come from a number of
sources, including laboratory or field studies; these can be either historical data or data from experi-
ments designed specifically for parameter estimation. Data from laboratory studies usually have less
variability than those from field studies but they may not reflect real system variables. Field studies,
on the other hand, yield data with greater variability and are usually more costly than lab studies.
Historical data, from published literature, have unique problems. They should be from lab or field
studies that are as close to the modeled system as possible. In some cases, even these data may not
be available and parameters will have to be estimated by extrapolating from other systems. In these
cases, a conservative approach should be taken in which parameter values are estimated that can
provide simulation results that reflect a worst-case scenario.
1.4.2 Uncertainty Not Considered

One of the greatest causes of model failure is the use of constant parameter values when the param-
eter is a random variable. A real-world variable with high variability should be represented in a
model by parameters with a comparable level of variability (see Chapter 4), not just its mean value.
Models with variable parameters can produce simulations significantly different from models with
constant values for those same parameters. That is because at the tail ends of a parameter’s prob-
ability distribution, the dynamical behavior of the system can be quite different from that using just
the mean of the distribution. It is at the tails of a distribution that parameters can cause the most
extreme perturbations of a system.
1.4.3 Bias (Political, Social, Economic)

Some models are developed with a particular point of view and the accuracy of these models’ pre-
dictions is rarely tested. The models are developed, either consciously or subconsciously, to obtain
the desired results. The simulation results are then used to justify the point of view. Political bias can
come into play when state and federal regulatory agencies differ from those being regulated over
the best way to control pollution. Different models can be developed to show different outcomes of
regulatory policies. Similarly, models are used to make predictions about the outcomes of social
policies. Models can be developed to overestimate the consequences of not following a given policy,
which often can be used to justify increases in the budgets of those making policy. Perhaps the most
egregious source of bias is when a model is used to support a position that shows a product is harm-
less when, in fact, it may cause harm to human health or the environment. A product judged to be
harmless can mean significant economic benefits to the company marketing that product.
1.4.4 Lack of Understanding of Real-World Systems

The failure of predictive models to accurately predict the future results more from a lack of under-
standing of the system being modeled than from inaccuracies in computer programming. Assuming
Introduction 7
that we have an error-free computer program, it will predict only what it has been designed to do,
that is, the model can make predictions about a system based only upon the assumptions in the
model. That is why it is important for a modeler, or a model user, to have a good understanding of
the system being modeled—to assure that the appropriate mechanisms are built into the model so
that it is capable of addressing the pertinent questions posed by the modeler. In the case of modeling
wild populations or ecosystems, it is imperative that the modeler spend time in the natural system
to observe its components and their relationships.
1.4.5 Misuse of Mathematics

The population geneticist, Charles Birch, and ecologist, Henry Andrewartha, warned ecologists
to “avoid the misuse of mathematics. … Ecologists, and especially mathematicians with a slight
knowledge of biology, seem to be prone to the mistake of building a model with symbols which,
they pretend, represent certain qualities of animals” (Andrewartha and Birch 1954, 11). Fifty years
later, however, Bialek and Botstein (2004, p. 789) argued, “Understanding how to reason in the
language of mathematical symbols is essential, but one must go further to appreciate that these
symbols actually stand for variables of the natural world.” The different perspectives reflect the
increased complexity and realism of today’s models compared with those in 1954. Increased com-
plexity and computing power, as well as available simulation software, however, do not guarantee
more reliable predictions. Instead of just applying “elegant” mathematics, a modeler must have a
clear understanding of what is going on in the real world (May 2004).
REFERENCES
Andrewartha, H. G., and L. C. Birch. 1954. The Distribution and Abundance of Animals. Chicago: University
of Chicago Press.
Bialek, W., and D. Botstein. 2004. “Introductory Science and Mathematics Education for 21st-Century
Biologists.” Science 303:788–790.
Chapman, D. G. 1971. “Mathematics and Ecology.” In Statistical Ecology, Vol. 3, ed. G. P. Patil et al., 428–434.
University Park, PA: Pennsylvania State University Press.
Dixon, K. R., and B. D. Murphy. 1979. “A Discrete-Event Approach to Predicting the Effects of Atmospheric
Pollutants on Wildlife Populations Using 14C Exposure.” In Animals as Monitors of Environmental
Pollutants, 15–26. Washington, DC: National Academy of Sciences.
Epstein, J. M. 2008. “Why model?” Journal of Artificial Societies and Social Simulation 11(4):12, http://jasss.
soc.surrey.ac.uk/11/4/12.html.
Gentner, D., and A. Stevens, eds. 1983. Mental Models. Hillsdale, NJ: Lawrence Erlbaum Assoc.
Hermann, C. 1967. “Validation Problems in Games and Simulation with Special Reference to Models of
International Politics.” Behavioral Science 12:216–230.
Holcomb Research Institute. 1976. Environmental Modeling and Decision Making. New York: Praeger.
Killough, G. G., K. R. Dixon, N. T. Edwards, et al. 1976. Progress Report on Evaluation of Potential Impact
of 14C Releases from an HTGR Reprocessing Facility. ORNL/TM-5284, Oak Ridge, TN: Oak Ridge
National Laboratory.
Lucas, H. L., ed. 1962. The Cullowhee Conference on Training in Biomathematics. Raleigh, NC: Institute of
Statistics of North Carolina State College.
Martin, F. F. 1968. Computer Modeling and Simulation. New York: Wiley.
May, R. M. 2004. “Uses and Abuses of Mathematics in Biology.” Science 303:790–793.
Morgan, M. G., B. Fischoff, A. Bostrom, et al. 2002. Risk Communication: A Mental Models Approach.
Cambridge: Cambridge University Press.
O’Neill, R. V., and O. W. Burke. 1971. A Simple Systems Model for DDT and DDE Movement in the Human
Food Chain. ORNL-IBP-71-9, Oak Ridge, TN: Oak Ridge National Laboratory.
Petroski, H. 1992. To Engineer Is Human: The Role of Failure in Successful Design. New York: Vintage Books.
Pilkey, O. H., and L. Pilkey-Jarvis. 2007. Useless Arithmetic: Why Environmental Scientists Can’t Predict the
Future. New York: Columbia University Press.
Rieber, L. P., S.-C. Tzeng, and K. Tribble. 2004. “Discovery Learning, Representation, and Explanation within
a Computer-Based Simulation: Finding the Right Mix.” Learning and Instruction 14:307–323.
van der Vaart, H. R. 1977. “Some Signposts for the Education of Systems Ecologists.” In New Directions in the
Analysis of Ecological Systems, Part 1, ed. G. S. Innis, 35–41. La Jolla, CA: Simulation Council.
Windschitl, M., and T. Andre. 1998. “Using Computer Simulations to Enhance Conceptual Change: The Roles
of Constructivist Instruction and Student Epistemological Beliefs.” Journal of Research in Science
Teaching 35:145–160.
2 Principles of Modeling
and Simulation
2.1 SYSTEMS
This chapter describes principles of modeling and simulation. It is important, however, to under-
stand that these activities are grounded in the concept of systems. Therefore, we begin by describing
what a system is and why knowledge of a system is important.
2.1.1 Definition
The term system has become commonplace in many scientific disciplines. In physiology there are car-
diovascular, respiratory, and immune systems as well as the central nervous system. In ecology, there
are ecosystems. In meteorology, there are weather systems. There are expert systems and geographic
information systems. What do all of these systems have in common? Each system is a collection of
interconnected components, or subsystems, that functions as a complete entity or whole that is greater
than the sum of its separate parts. The cardiovascular system consists of the heart, lungs, blood, and
connecting veins and arteries. An ecosystem is a collection of plant and animal populations that
interact through various processes such as predation, consumption, competition, and decomposition.
A formal definition of a system will aid in later discussions of modeling and simulation.
Definition 2.1: A system is a collection of components that are interconnected in such a way as to
function as a whole.
The adverse effects of many toxicants are directly related to their ability to interfere with the nor-
mal functioning of systems. The receptor–ligand interactions of the nervous system, for example,
are affected by organophosphates. Many toxic chemicals interfere with cellular energy production
systems. Other toxic substances can impair the reproductive system by reducing spermatogenesis or
causing birth defects or abortion of the fetus.
2.1.2 System Input and Output

The components or compartments of a system usually are represented by state variables, that is,
those variables that define the state of the system. Once a system has been defined, it is possible to
identify stimuli or disturbances, called inputs, from outside the system, that operate on the system to
produce a response called the output. For example, the nematode fumigant dibromochloropropane
(DBCP) can be an input to an animal, through ingestion or inhalation, with a resulting output of
decreased spermatogenesis.
2.1.3 Control Systems

One special type of system is one in which one or more inputs to the system are controlled
or regulated by the system to produce one or more desired outputs. In engineering, this type
9
of system is called a control system. A typical example of a control system is a thermostat-

controlled heating system. The temperature is regulated or controlled by the thermostat, which
provides the input reference temperature. The input operates on the heating system, which pro-
duces the output—ambient temperature. Control systems also are found at all levels of biologi-
cal organization. At the individual organism level, endocrine systems are control systems. For
example, the hypothalamus produces a gonadotropin-releasing hormone, which is input to the
anterior pituitary gland, regulating the production of gonadotropins follicle stimulating hor-
mone (FSH) and luteinizing hormone (LH), which in turn regulate the production of gonadal
hormones. At the ecosystem level, a predator stalking or attacking its prey is another example
of a control system. The input is the observed position of the prey. This visual stimulus oper-
ates on the predator’s central nervous system and muscles to produce the output of movement
of the predator toward the prey. These examples are illustrated with block diagrams later in
this chapter.
2.1.4 Feedback
That component of the control system responsible for activating the system to produce the output is the
control action. In the previous examples, the control action in the heating system is the thermostat, in
the hypothalamus-pituitary-gonadal system it’s the androgen and estrogen receptors in both the hypo-
thalamus and pituitary, and in the predator-prey system it’s the predator’s brain. A system in which the
control action is dependent upon the output is called a closed-loop or feedback control system.
Definition 2.2: Feedback in a closed-loop system is the signal obtained by comparing the output
with the input.
Feedback can be either positive or negative. Negative feedback tends to increase stability of the
system, whereas positive feedback decreases stability. The systems described previously are all
negative feedback systems. In the heating system, the feedback signal is the output temperature,
which is compared with the reference temperature in the thermostat. In the hypothalamus-pituitary-
gonadal system, the feedback signal is the actual level of gonadal hormone, which is compared with
the normal level at the pituitary. The feedback signal in the predator-prey system is the position of
the prey, which feeds back to the predator’s brain.
2.1.5 System States: Steady State versus Transient States

In modeling and simulation of systems, we are interested in analyzing and forecasting a system’s
dynamics. The behavior of a system over time can tell us the effects of a toxicant on that system. It
can be difficult to attribute observed dynamic behavior to a disturbance from a toxicant, however,
because of inherent fluctuations or transients in the system. A system in which the initial transients
have disappeared and no new disturbances are input will be in a condition where the system compo-
nents do not change with respect to time. Such a system is said to be in a steady state. We will look
at transient behavior of a system in more detail in Section 2.2.1.2.
2.1.6 Discrete versus Continuous

Components of a system can fluctuate in a continuous fashion such as water flow in a river system.
Other components may increase or decrease at discrete intervals such as a population of fish that
spawns over a short period of the year in the river system.
In most complex systems, there will be both continuous and discrete components. In addition
to changes with respect to time, systems also can fluctuate in space. An animal population that is
growing in size is likely to expand into new habitats. Or, a population may migrate and change its
Principles of Modeling and Simulation 11
geographic location. This is an important consideration in simulating contaminant effects because

exposure can change as an animal changes locations. It follows that a system must be described as
continuous or discrete in both time and space.
2.1.7 Linear versus Nonlinear

Another way of classifying systems is by their response to inputs, whether the inputs are controlled
variables, random variables, or disturbances. Those systems that satisfy the principle of superposi-
tion are termed linear systems. This principle states that the response of a linear system resulting
from several simultaneous inputs is equal to the sum of the responses to each of the individual
inputs. Those systems that do not satisfy the principle of superposition are termed nonlinear sys-
tems. All real systems are nonlinear to some degree because for extreme input values the principle
of superposition will not hold. Some systems may be approximately linear under normal envi-
ronmental conditions. This distinction is important in ecotoxicology because our main concern is
with those situations that are not normal. Therefore, it is important to determine whether relations
between system components are nonlinear because nonlinear threshold responses may occur. Also,
some toxic chemicals may be antagonistic or synergistic. We will discuss linear systems in some
detail because an understanding of the behavior of linear systems provides a conceptual basis for
the study of nonlinear systems.
2.2 MODELING
Because the dynamics of real systems are quite complex, understanding the impacts of toxicants on
a system can be enhanced by modeling the system. In applying modeling to ecotoxicology, we are
interested in studying a “real-world” system and the effects of various toxicants on that system. The
modeling approach we take in this book is to define the system relationships in terms of quantitative
mechanisms in a model of the system. A model is a necessary abstraction of the real system. For a
model to adequately reflect the behavior of a system perturbed by a toxicant, however, requires a
mechanistic approach to modeling. The level of abstraction is determined by the objectives of the
model. A model designed to give a quick sketch of the dominant effects on a system will require less
realism than one designed to provide accurate predictions of future system behavior.
The modeling process involves three steps: (1) identification of system components and boundar-
ies, (2) identification of component interactions, and (3) characterization of those interactions using
quantitative abstractions of mechanistic processes. Once the model has been defined, it is imple-
mented on a computer. This is the process of simulation (see Section 2.3).
Models can be described in a variety of ways. In this book, we use two approaches: equations
and block diagrams.
2.2.1 Equations
Equations can define the quantitative functional relations among system components as well as rep-
resent the system dynamics, or changes in mass or energy of the system components as functions of
time. Difference equations are used to describe discrete systems whereas differential equations are
used to represent continuous systems. For some applications, where the numerical value of the state
variables remains high, difference equations can approximate continuous systems.
The behavior of models described by the two types of equations depends upon the time step used
in the model, and the difference is usually small for small time steps. This can be seen by a study of
the definition of differential equations.
Definition 2.3: A differential equation is an algebraic equality (equation) that contains either dif-
ferentials or derivatives.
y = f(t)
t
t1
FIGURE 2.1 The function y = f(t) showing the derivative of y, which is the slope at the point t1.
Definition 2.4: The derivative of a function f is another function, the differential operator d/dt,
whose value at any time t1 in the domain of f is:
df (t ) f (t + t ) − f (t )
= lim (2.1) ◾
dt 0← t t
In other words, the derivative of the function f(t) is the function at some time, t plus Δt, minus the
function at time t, divided by Δt, and then taking the limit of that function as Δt approaches zero,
provided this limit exists.
Let y = f(t). The derivative dy/dt can be interpreted as the instantaneous rate of change of y with
respect to the independent variable t. The derivative of the function f(t) also can be defined for t1 as
the slope of the tangent to the curve f(t) at t1 (Figure 2.1).
A first-order system (one with a single compartment) can be represented by the general first-order
differential equation:
dy(t )
= f ( y(t ), t ) (2.2)
dt
As pointed out previously, the derivative dy/dt is equivalent to the instantaneous rate of change
of the function f(y(t),t). The following example of a differential equation describes the decrease in
material from a single compartment, or the exponential decay process.
Example 2.1
Here we have an example of the exponential decay function (also called the exponential elimina-
tion function). The exponential decay function can be written as
y (t ) = y0e − rt (2.3)
where yo is the initial value of y at time t = 0 and r is the decay rate constant. The derivative of y
is then defined as the negative rate constant times y:
dy
= −r ⋅ y (2.4)
dt
100
80
60
y
40
20
0
0 10 20 30 40 50 60
t
(a)
–2
–4
dy/dt
–6
–8
–10
0 10 20 30 40 50 60
t
(b)
FIGURE 2.2 (a) Exponential decay function, y, and (b) its derivative, dy/dx.
Figure 2.2 illustrates the relation between a function and its derivative. Note that the derivative
(rate) takes on the value of zero when the function is parallel to the t axis (zero slope).
There are several ways of classifying differential equations. One way is based on the number of
independent variables. A partial differential equation is one with more than one independent vari-
able. A differential equation with only one independent variable is called an ordinary differential
equation. Each type can have one or more dependent variables. Examples of independent vari-
ables are the Cartesian axes in three-dimensional space and time. Partial derivatives can be used
to model the movement of toxicants in geographical space but are beyond the scope of this text.
A second type of classification of differential equations depends on the way the independent
variable time is treated. If time is expressed explicitly in the differential equation, it is a time-
variable differential equation. A time-invariant differential equation is one that has no terms that
explicitly include time as an independent variable. In other words, time is only expressed in the
derivative of the dependent variable (e.g., dy/dt where t is time).
In the last section we defined linear and nonlinear systems. There is a close correspondence
between systems and differential equations. Linear differential equations can represent linear sys-
tems and nonlinear differential equations can represent nonlinear systems. Just as linear systems
can approximate nonlinear systems under certain conditions, linear differential equations some-
times can approximate nonlinear systems. Even at the point where nonlinearities are included in a
model, a comparison with the dynamics of a linear model can be instructive. A linear differential
equation is one that has no terms that include higher powers, products, or transcendental func-
tions of the dependent variables. Any differential equation that has any such term is a nonlinear
differential equation.
In general, an nth-order linear system can be described by a linear differential equation of
order n:
d ny d n−1y dy
+ a1 + + an−1 + any = f (t ) (2.5)
dt n
dt n−1
dt
where f(t) is a linear input function, with constant coefficients. Equations of an order greater than
one are rarely used in ecological modeling (but see Clark 1971, Innis 1972). If, in Equation 2.5,
there is zero input (i.e., f (t) = 0), the equation is said to be homogeneous. In this text, we place the
emphasis on nonhomogeneous, first-order differential equations.
In general, a first-order linear system (i.e., a system with 1 compartment) can be described by
a linear differential equation of order 1:
dy (2.6)
+ ay = f (t )
dt
where, in general, a is a function.

Although the coefficients in Equation (2.6) are, in general, functions, preliminary models often
begin by assuming a is a constant. It follows therefore, that ordinary, linear, constant coefficient,
differential equations are important in the modeling of dynamic systems.
2.2.1.1 Solution of Ordinary First-Order Differential Equations

There are two ways of solving differential equations: analytical solutions and numerical solutions.
Analytical solutions are possible for relatively simple differential equations used to model the rela-
tionships between variables. Some examples of analytical solutions are described in Chapter 5.
A system of nonlinear differential equations generally is too complex to be solved analytically.
Numerical integration methods are required for these models and are discussed in greater detail
in Section 2.2.1.4. We can also use the analytical solution to linear, constant coefficient, ordinary
differential equations as an approximation to nonlinear differential equations with variable coef-
ficients. The analysis of linear, constant coefficient, differential equations also will provide us
with the foundation for the later analysis of more complex systems. In this section, we describe
the solution, y = y(t), to the nonhomogeneous, linear, constant coefficient, ordinary differential
equation (Equation [2.6]) where f(t) is a linear input function, also with constant coefficients. The
general solution y(t) to Equation 2.6 consists of two parts, the free response ya(t) and the forced
response yb(t). The free response is the solution to Equation (2.6) when there is zero input and the
solution depends only on the initial condition, y(0). The forced response yb(t) of the differential
Equation (2.6) is the solution of the differential equation when all the initial conditions are zero.
Whereas the free response has zero input, the forced response is defined by the input f(t). The sum
of these two responses comprises the total response or solution of the equation.
Example 2.2
Again, let’s look at the example of the exponential elimination function. This time we use the
letter Q for the dependent variable and include the parameter p4, which is the minimum or final
value of Q. The initial value of Q is p5. For our example of an exponential elimination or one-
compartment elimination model, we define the rate of change in the state variable Q with the
differential equation:
dQ
= − p 2(Q − p 4 ) (2.7)
dt
where
p2 = the elimination rate constant, and p4 = the minimum or steady-state value of Q.
Now, the free response, Qa(t) is:
Qa(t ) = p5e − p 2t (2.8)
where
p5 = the initial value of Q, and the forced response, Qb(t) is
Qb (t ) = p4 1− e − p2t (2.9)
The total response is the sum of the free response and the forced response:
Q(t ) = Qa(t ) + Qb(t )
= p5e − p 2t + p 4[1− e − p 2t ]
(2.10)
= p5e − p 2t + p 4 − p 4e − p 2t
= ( p5 − p 4 )e − p 2t + p 4
We can now plot each part of the solution (Figures 2.3 and 2.4).
1. Initial conditions: Q(0) = 100 and Q(0) = 50. Input, p2 = 0.05, p4 = 10.
Note that starting with two different initial conditions, the free response (a) approaches
zero for both curves, showing that the response only depends upon the initial condition.
The forced response (b) does not change with the different initial conditions but starts at
zero and approaches the value of the parameter p4. The total response is the sum of the
free response and the forced response where the total response shows the different initial
conditions but approaches the p4 value of 10 rather than zero.
2. Initial condition: Q(0) = 100. Input, p2 = 0.05, p4 = 10 and p4 = 20.
In this case, the input, p4, has no effect on the free response, which again approaches zero.
The forced responses approach the two values of p4. The total response for both input
values starts at the initial condition, Q(0) = 100, but each response approaches a different
input value of p4 (Figure 2.4).
100 10 100
Forced Response, Qb
8
Total Response, Q
80 80
Free Response, Qa
60 6 60
40 4 40
20 2 20
0 0 0
0 50 100 0 50 100 0 50 100
t t t
(a) (b) (c)
FIGURE 2.3 Solution to elimination or decay function: (a) free response, (b) forced response, and (c) total
response.
100 20 100
Forced Response, Qb
Free Response, Qa
Total Response, Q
50 10 50
0 0 0
0 50 100 0 50 100 0 50 100
t t t
(a) (b) (c)
FIGURE 2.4 Solution to elimination or decay function: (a) free response, (b) forced response, (c) total
response.
2.2.1.2 Steady-State and Transient Response

The total response also can be described by the sum of two other responses: the transient response
and the steady-state response. The transient response is that part of the total response that approaches
zero as time approaches infinity. The steady-state response is that part of the total response that
does not approach zero as time approaches infinity.
The total response in Example 2.2 was
Q(t ) = ( p 5 − p 4 )e− p 2 t + p 4 (2.11)
The transient response, then, is
QT (t ) = ( p 5 − p 4 )e− p 2 t (2.12)
which approaches zero because as time increases, the exponential function approaches zero. The
steady-state response is the constant p4, which does not approach zero:
Qss = p 4 (2.13)
The total response again is the sum of the transient response and the steady-state response.
2.2.1.3 Difference Equation Approximation to Differential Equation

A differential equation can be approximated by a difference equation, especially for small values of
Δt. The approximation can be seen by giving Δt a value of one in Equation 2.1 and then not taking
the limit:
dy(t ) y(t + t ) − y(t )

= lim
dt 0← t t
dy(t ) y(t + 1) − y(t )
≈ (2.14)
dt 1
dy(t )
≈ y(t + 1) − y(t )
dt
By substituting the discrete approximation y(t + 1) – y(t) for the derivative dy/dt, Equation (2.6) now
can be approximated by the difference equation:
dy(t )
≈ y(t + 1) − y(t ) = ay
dt (2.15)
y(t + 1) = y(t ) + ay
Example 2.3: Exponential Decay Function
To show the close approximation of a difference equation to a differential equation, we plotted the
solution to the differential Equation (2.7) with the difference equation approximation (Figure 2.5).
100
Difference equation
Differential equation
90
80
Amount Remaining, Q
70
60
50
40
30
20
0 10 20 30 40 50 60 70 80 90 100
Time, t
FIGURE 2.5 Exponential decay function plotted as the numerical solution to a differential equation and as
the difference equation approximation.
2.2.1.4 Numerical Solutions to Differential Equations

In Section 2.2.1.1, we discussed the analytical solution to ordinary differential equations. In simu-
lation, differential equations can be difficult to solve analytically. These equations may require
numerical solutions, which are usually close approximations to analytical solutions. In the chapters
that follow, we use numerical methods exclusively to solve the differential equations in a model.
There are many different numerical methods used to obtain solutions, and many variations on each
method. Conceptually, numerical methods start from an initial solution of the differential equation
to estimate the dependent variable y(tn) at time point n and then take a short step forward in time to
find the next solution y(tn+1) at time point (n + 1). We briefly describe the algorithms most commonly
used in this text.
Numerical methods can be grouped according to whether the equations are stiff or nonstiff.
A stiff equation is a differential equation whose solution can become unstable if there are drastic
changes in the rate of change (slope) of the variables or if the algorithm step size is extremely
small. For nonstiff differential equations, variations on the Runge-Kutta methods are important
iterative methods. These techniques were developed by the German mathematicians Carl David
Tolmé Runge and Martin Wilhelm Kutta. The basic algorithm divides the time interval between
predicted values of the dependent variable y(t) into smaller intervals for which the slope is esti-
mated. If there are four intervals in the time step, the method is called a fourth-order method. The
method is referred to as a one-step solver because to estimate y(tn), it needs only the solution at the
immediately preceding time point, y(tn−1). Other numerical methods, such as the Adams-Bashforth-
Moulton predictor-corrector solver, are multistep solvers in that they normally need the solutions
at several preceding time points to compute the current solution. For some equations, the nonstiff
methods might result in a magnification of errors in approximation, leading to an unstable or non-
existent solution. Methods that do not magnify approximation errors are called numerically stable.
It is important to use a stable method when solving a stiff equation. If the nonstiff methods do not
work, or are extremely slow, one should try an implicit method designed for stiff equations such as
the backward Euler method or implicit Runge-Kutta method. These methods are described further
in Chapter 3.
dy
Input Control Output dt y = ∫dy/dt
Element
∫dt
(a) (b)
FIGURE 2.6 Block diagram elements: (a) single block with an input and an output, and (b) a block with a
integral operator. The integral operates on the input derivative to create an output y. (From Joseph J. DiStefano
III et al., Feedback and Control Systems 2nd ed. Schaum’s Outline Series. New York: McGraw-Hill, 1990.
With permission from author.)
Summing Takeoff
Point Point
x + x+y x
x
y
FIGURE 2.7 Block diagram summing point and takeoff point. (Modified from Joseph J. DiStefano III, et
al., Feedback and Control Systems 2nd ed. Schaum’s Outline Series. New York: McGraw-Hill, 1990. With
permission from author.)
2.2.2 Block Diagrams

A second method of representing a model of a system is a flowchart in which blocks representing
system components are connected by lines representing the flow of information between compo-
nents. A block diagram is a particular type of flowchart that illustrates the functional relations
among system components, particularly where relationships show cause and effect. Blocks can rep-
resent state variables, other system components, and mathematical operations (Figure 2.6). As we
will show, there can be a direct transformation from a differential (or difference) equation to a block
diagram and vice versa.
Additional features of a block diagram are summing points and takeoff points (Figure 2.7). A
summing point is where two or more inputs are summed to produce a single output. A takeoff point is
where a single input is branched to provide identical inputs to two or more blocks or summing points.
A block diagram of a feedback system has a takeoff point from the output signal that feeds back to
a summing point where it is summed with the reference input (Figure 2.8). A negative feedback system
has a negative feedback signal and a positive feedback signal represents a positive feedback system.
The general structure of a block diagram is shown in Figure 2.8. This general structure includes
all of the basic elements of a block diagram, although not all elements are necessarily found in all
systems. Also, most systems will be much more complex with several parallel paths representing
many state variables and their interactions.
Controlled system. Each subsystem can be represented by a controlled system or controlled

process. Inputs to the controlled system include an internal control signal and external dis-
turbances. Output from the controlled system usually is the state variable used to describe
the subsystem.
Control elements. Control elements are the subsystem components that generate the control
signal input to the controlled system.
Disturbance
Control
Actuating Signal or
Reference (Error) Manipulated Controlled
Input Signal Feedforward Variable Output
(Control) Process
+ Elements
±
Primary Forward Path

Feedback
Signal
Feedback
Elements
Feedback Path
FIGURE 2.8 Generalized block diagram. (From Joseph J. DiStefano III, et al., Feedback and Control
Systems 2nd ed. Schaum’s Outline Series, New York: McGraw-Hill, 1990. With permission from author.)
Control signal. The control signal is produced by control elements and acts upon the con-
trolled process.
Feedback elements. The feedback elements are the subsystem components that define the
functional relationship between the controlled output and the primary feedback signal.
Controlled output. The controlled output is usually the state variable used to define the
subsystem.
Primary feedback signal. The primary feedback signal results from the action of the feedback
elements operating on the controlled output. The signal is compared with the reference
input at the summing point.
Reference input. The reference input is an external stimulus to the subsystem. The reference
input is compared with the primary feedback signal at the summing point.
Actuating signal. This signal is the input from the summing point to the control elements and
results from the summation of the reference input and the primary feedback signal.
Disturbance. Disturbance inputs are external environmental variables or toxic perturbations.
These inputs can affect the controlled system directly or indirectly through the effects on
other system components.
In Sections 2.1.3 and 2.1.4, we described three examples of negative feedback control systems,
a thermostat, the hypothalamus-pituitary-thyroid system, and a predator behavior system. The fol-
lowing examples illustrate these systems using block diagrams
Example 2.4: Thermostat
A block diagram for a thermostat is shown in Figure 2.9 (DiStefano et al. 1990).
In the thermostat model, the controlled output is the actual room temperature. The reference
input is the temperature that is set to the desired, or reference, temperature. The actual room tem-
perature then becomes a feedback signal in a negative feedback loop. The reference temperature
and the actual temperature are compared in a summing point (thermostat) to generate an actuat-
ing signal. If the set temperature is higher than the actual temperature, the actuating signal will be
positive so the control element (furnace) is actuated (turned on). The control signal, or manipulated
variable, is the heat from the furnace, which raises the room temperature. When the actual room
temperature exceeds the set temperature, the actuating signal now is negative and the furnace
turns off.
Actual
Thermostat Enclosure
Heat Enclosure Temperature
+
Furnace Environment
_ Temperature
Reference
Temperature
(Set Point)
FIGURE 2.9 Block diagram for a thermostat. (From Joseph J. DiStefano III, et al., Feedback and Control
Systems 2nd ed. Schaum’s Outline Series. New York: McGraw-Hill, 1990. With permission from author.)
+ Pituitary TSH Thyroid

Normal _ Gland Gland Blood
Thyroxine Thyroxine
Level Level
FIGURE 2.10 Block diagram for the hypothalamus-pituitary-thyroid control system. (From Joseph J.
DiStefano III, et al., Feedback and Control Systems 2nd ed. Schaum’s Outline Series. New York: McGraw-
Hill, 1990. With permission from author.)
+ Predator Predator
Brain Legs, Feet,
Prey _ and Claws Predator
Location Location
FIGURE 2.11 Block diagram for a predator behavior control system. (Modified from Joseph J. DiStefano
III, et al., Feedback and Control Systems 2nd ed. Schaum’s Outline Series, New York: McGraw-Hill, 1990.
With permission from author.)
EXAMPLE 2.5: Hypothalamus-Pituitary-Thyroid System
A simplified diagram of the hypothalamus-pituitary-thyroid system is shown in Figure 2.10

(DiStefano et al. 1990). A more complete description of this system can be found in Carr and
Norris (2006).
In this simplified system, the controlled output is the blood thyroxine level. It is transported in
the bloodstream to the brain in a negative feedback loop. There it is compared with the normal
thyroxine level (reference input). If the actuating signal is positive, the blood thyroxine level is
greater than normal. This signals the pituitary gland to reduce the secretion of the thyroid stimulat-
ing hormone (TSH). This control signal reduces the activity of the thyroid, lowering the amount of
thyroxine secreted into the bloodstream.
EXAMPLE 2.6: Predator-Prey System
Block diagrams can also be used to model animal behavior. In this example, we model the behav-
ior of a predator as it hunts its prey (Figure 2.11).
In this system, the controlled output is the location of the predator. The reference input is the
location of the prey. The relative distance between the two locations is an actuating signal pro-
cessed by the predator’s brain. The predator’s brain sends a control signal to the predator’s legs,
feet, and claws to move the predator closer to the prey. The distance is continually monitored and
the predator adjusts his position until he is able to attack the prey.
2.2.3 Stochastic Models

Deterministic modeling involves the use of parameters that take on only a single value, thus “deter-
mining” the model’s outcome. Stochastic modeling, on the other hand, involves the use of random
variables. Not only can model coefficients be functions of other variables, they can be functions of
random variables and thus can be random variables themselves.
Definition 2.5: A random variable is a variable that takes on values with some relative frequency.
In this way of classifying models, those with random (stochastic) variables are called stochastic
models and those without are called deterministic models. Random variables are used to represent
the random variation or “unexplained” variation in the state variables. We describe stochastic mod-
els more fully in Chapter 4.
2.2.4 Individual-Based Models

Models that simulate all individuals simultaneously are referred to as individual-based models.
Each individual in the simulation has a unique set of characteristics: age, size, condition, social
status, and location in the landscape. Each individual has its own history of daily foraging, repro-
duction, and eventual mortality.
A number of individual-based models have been described by Huston and DeAngelis (1988) and
DeAngelis and Gross (1992). Individual-based models have been applied to ecotoxicology by Hallam
et al. (1989) and Hallam and Lassiter (1994), among others. This approach is becoming popular for
several reasons. For one thing, it enables the modeler to include complex behavior and decision mak-
ing by individual organisms. Most importantly, it allows one to model populations in complex land-
scapes, where different individuals may be experiencing very different conditions. Individual-based
models were very uncommon up until a few years ago because they require a great amount of com-
puter power. As computers increase in power, however, these models are becoming prevalent. A
flowchart for a generic individual-based model is shown in Figure 2.12. A specific model of this type,
which is described in Chapter 11, has been developed to study avian populations exposed to agricul-
tural insecticides.
2.2.5 Aggregated Models

There are two general ways in which models of individuals can be extended to a population as a
whole. First, one can simulate not just one individual, but all individuals that make up the popula-
tion of interest. Second, one can aggregate various population members into classes, such as age
classes. The model then follows not individual organisms but variables representing the numbers of
individuals per age class. An example of an aggregated population model is described in Chapter 13.
2.3 SIMULATION
Once we have a model defined in quantitative terms, the next step in using the model in the study
of systems is simulation. We use simulation to obtain the system output as a function of time, or the
time response of the system. We do this by exercising the model on a computer.
There are many things to consider before a simulation can be performed. These include determin-
ing the type of computer and programming language, cost of running the computer, the data needed for
model parameters, and availability of expertise for the design and analysis of simulation experiments.
The types of computers available for simulation are analog, digital, and hybrids of analog and
digital computers. Digital computers are the most widely used in simulation, including applications
Random documents with unrelated
content Scribd suggests to you:
The Project Gutenberg eBook of Madame
Goubaud's pillow lace patterns, and
instructions in Honiton lace making
This ebook is for the use of anyone anywhere in the United States
and most other parts of the world at no cost and with almost no
restrictions whatsoever. You may copy it, give it away or re-use it
under the terms of the Project Gutenberg License included with this
ebook or online at www.gutenberg.org. If you are not located in the
United States, you will have to check the laws of the country where
you are located before using this eBook.
Title: Madame Goubaud's pillow lace patterns, and instructions in

Honiton lace making
Author: Madame Adolphe Goubaud
Release date: December 28, 2023 [eBook #72526]
Language: English
Original publication: London: Ward, Lock, and Tyler, 1871
Credits: Bob Taylor, Charlene Taylor and the Online Distributed

Proofreading Team at https://www.pgdp.net (This file was
produced from images generously made available by The
Internet Archive)
*** START OF THE PROJECT GUTENBERG EBOOK MADAME

GOUBAUD'S PILLOW LACE PATTERNS, AND INSTRUCTIONS IN
HONITON LACE MAKING ***
MADAME GOUBAUD’S
NEEDLEWORK BOOKS.
Price 1s., postage 1d.

FASHIONABLE AND USEFUL NEEDLEWORK.
Madame Goubaud’s Tatting Patterns.
Madame Goubaud’s Embroidery Patterns.
Madame Goubaud’s Crochet Patterns.
Madame Goubaud’s Knitting and Netting
Patterns.
Madame Goubaud’s Monograms, Initials, &c.
Madame Goubaud’s Point Lace Patterns.
Madame Goubaud’s Guipure d’Art.
LARGER SIZE
Madame Goubaud’s Guipure Work. 2s.
Madame Goubaud’s Pillow Lace. 3s. 6d.
Sold wholesale by all the Newgate-street Houses, and many
Warehousemen in London and the Provinces, and to be had of all
Berlin and Fancy Shops.
Price 6d., postage 1d. For Young Ladies at School & Home.
NEEDLEWORK INSTRUCTION BOOKS.
Madame Goubaud’s Berlin Work Instructions.
Madame Goubaud’s Embroidery Instructions.
Madame Goubaud’s Crochet Instructions.
Madame Goubaud’s TRIMMED PAPER PATTERNS of all the

NEWEST MODES of Costumes, Robes, Jackets, Mantles, Sleeves,
Underclothing, and Baby Linen are accompanied by a Flat Paper
Pattern to cut out from, and are sent to Ladies by return post on
receiving orders and remittances.—30, Henrietta-street, Covent-
garden, W.C.
Price 7s. 6d. each, postage 6d.

THE BEST BOOKS OF THEIR KIND IN THE WORLD.
BEETON’S HOUSEHOLD BOOKS.

MRS. BEETON’S HOUSEHOLD MANAGEMENT.
BEETON’S BOOK OF NEEDLEWORK.
BEETON’S BOOK OF GARDEN MANAGEMENT.
BEETON’S BOOK. OF HOME PETS.
BEETON’S LAW BOOK. (Next Month.)
Ward, Lock, & Tyler, Warwick House, Paternoster Row.

WEDDING OUTFITS
AND
BABY LINEN.
French Wove
CORSETS,
10s. 6d.
GIRLS’ HATS,
from 8s. 6d.
BERCEAUNETTE in Pretty Rose-bud Chintz, 30s.
Muslin and Lace, like Drawing, 50s. Fully fitted, £3
10s.
Outfits to India, Outfits to China,

Outfits to Japan, Outfits to Australia,
For £20, £40, £60, & £100.
An ILLUSTRATED PRICE LIST, containing much information

respecting Wedding Trousseaux, Indian and Colonial Outfits, Baby
Linen, with various Designs in Nightdresses, Morning Wrappers,
Petticoats, Corsets, Crinolines, &c., gratis and post free. Caution.—
Entered at Stationers’ Hall.
ADDLEY BOURNE,
37, PICCADILLY, W.
“Nowhere beats the heart more kindly
Than underneath the Tartan Plaid.”
MACDOUGALL & CO.,

OF INVERNESS,
Manufacturers to Her Majesty,

And their Royal Highnesses the Prince & Princess of Wales,
and the Courts of France, Russia, Prussia, and Spain,
Respectfully solicit attention to their large and most

varied assortment of real
SCOTCH & HIGHLAND MANUFACTURES.
THE HIGHLAND COSTUME,

And the COSTUMES of CALEDONIA, complete and correct, as
supplied to Royalty and Nobility. The famous Highland Cloak, an
exclusive speciality.
TWEEDS for Country and Town Wear of the most varied texture
and Patterns for Gentlemen’s wear.
Linsey Woolseys of the Newest Shades &

Patterns.
Scotch Serges, Tweeds, and Scotch Poplins for Ladies’ Dresses.
An unrivalled selection of
SHAWLS, PLAIDS, CLOAKS, JACKETS, &c.
Their famous “Flora Macdonald” Cloaks; also the new “Caledonia” &
“Lady of the Lake” Cloaks. The “Trossachs Plaid,” so favourably
received.
Tartan Stockings for Ladies and Children to
match
all the Clans, in all sizes.
SCOTCH YARNS OF EVERY DESCRIPTION FOR

KNITTING.
Prices on the same moderate scale as our Inverness House. Patterns

and
Designs forwarded, post free, by applying to Macdougall & Co.
THE ROYAL TARTAN & TWEED WAREHOUSE,

42, SACKVILLE ST.
(THREE DOORS OFF), PICCADILLY, W.
MADAME GOUBAUD’S
PILLOW LACE BOOK.

DEVONSHIRE LACE-WORKER.
MADAME GOUBAUD’S
PILLOW LACE PATTERNS,

AND
INSTRUCTIONS
IN
HONITON LACE MAKING.
NUMEROUS ILLUSTRATIONS.
LONDON:
WARD, LOCK, AND TYLER,
WARWICK HOUSE, PATERNOSTER ROW
PREFACE.
In offering this little work upon Honiton Pillow Lace Making,

Madame Goubaud tenders her sincere thanks for the reception
accorded to her volume upon Point Lace.
The great demand for Pillow Lace Instruction and Designs has
induced Madame Goubaud to begin this, the first of a series upon
Pillow Lace Making. Valenciennes, Bedford, and Maltese laces will
follow in due course.
CONTENTS.
PAGE.
Appliqué, Honiton 52
Baby Cue 21
Beading or Chain 44
Braid Work 11
Butterfly 35
Cross Cottons, To 32
Cue 21
Devonshire Turn 19
Diamond Fillings, To Join 42
Double Stitch, A 36
Fern Sprig 47
Half Stitch 18
Honiton Guipure Lace 51
Honiton Lace Stitch 20
Lace Pillows 9
Lappet, Lace 49
Lerd Works or Fillings 38
” Diamonds 40
” Square 42
Maltese Cross 17
Materials Required for Lace Working 9
Old Violet Sprig 31
One Rose Border 25
Passements 11, 13, 15
Purl Edge 28
Purl Pin 28
Purl Pin for Groundwork 30
Raised Work 36
Rose Border 23
Rose Sprig 33
Rose, Shamrock, and Thistle Sprig 43
Sewings 24
Small One Rose Sprig 37
Snatch Pin 26
Spade Border 41
Sprigs, Making up The 48
Stem Work 34
Stem Work and Leaves 17
The Tulip Sprig 29
Throwing and Tying 24
To Form a Pattern 22
To Join 38
Trefoil Pattern 45
Turkey Tail 27
Whole Stitch 16
PILLOW LACE.
The materials required for Honiton Pillow Lace work are neither
numerous nor expensive. A pillow, two dozen bobbins, some lace
thread, a needle-pin, a pair of blunt scissors, and a few pricked
patterns or passements, and lace pins are all the requisites for this
useful and beautiful work.
The pillows are made and sold at moderate prices by Mrs. Mitchell,
a Devonshire lace worker, at Long Dean-street room in the Soho
Bazaar, Oxford-street. Mrs. Mitchell supplies bobbins, thread,
passements, and all requisites, as well as gives lessons in pillow-lace
making. The pillows can be obtained with plain or handsome covers
and cloths, and with wooden, ivory, or ornamental bobbins. Learners
will find it useful to have their leading bobbins ornamented, in
distinction to the plain, until they have acquired the art of working
Honiton lace.
For braid-work Mecklenburg thread is used; No. 8 is a useful size
to begin with for Honiton lace, No. 8 and 12 or 14 for very fine work;
No. 8 is quite fine enough for beginners to manage.
The pillows are always sent out fitted up, but to change one
design for another proceed as follows:—
Unpin the cloths, remove the passement; firmly pin down the new
passement with plain, not lace, pins; replace the cloths, attach the
bobbins to a pin about three-quarters of an inch from the place
intended to begin at, divide the bobbins into three parts, and loosely
plait down to the first pinhole, insert a pin and work on as before.
Nos. 1, 2, 3.—Braid Works and Passement.

No. 1 shows a pretty braid for point lace work; this braid consists
of whole stitch and pearl.
No. 2 is the passement or pricking for a shamrock spray.
No. 3 is a simple braid in whole stitch.
No. 1.—Braid-work.
No. 2.—Shamrock Passement.
No. 3.—Braid-work.
The first lesson usually given in pillow lace making is the braid
work; the learner is at first puzzled by the number of fine bobbins,
but if it is remembered that only four are used together, and that the
plain or whole stitch is only weaving or plaiting across and back
again, the difficulty vanishes.
Braid Work.
Mecklenburg linen thread, No. 8, six pairs of bobbins. For braid
work three pairs of leaders are required. One pair of leaders remains
on each side separated by a pin; work across and back again with
the third pair thus:—Work across, twist, make a stitch, twist twice,
pin, work back, twist twice, make stitch, twist twice, pin, and repeat
as before.
The lace pins are made expressly for the lace-workers in
Devonshire, and are sold in the old-fashioned paper rows.
Nos. 4 and 5.—Passements.

No. 4 is the passement or pricked design for a spade.
No. 5 is the passement of an old-fashioned sprig formerly much in
vogue among Honiton lace workers, now seldom seen.
No. 4.—Spade Passement.

No. 5.—Old Sprig Passement.
The passements are pieces of brown parchment, having the exact

glazed pattern pricked out on them as shown on pages 11, 13, and
15.
Ladies can easily prick them from the illustration. To prick from
lace itself it is necessary to keep the lace in proper position and to
place a pin in every pinhole along the outlines, to mark any crossed
threads by a pinhole, and to show where lerd or lead-works occur by
the same means.
Great accuracy is necessary for this work, as all the beauty of the
lace depends on the correctness of the prickings.
Position in Working Pillow Lace.

Devonshire workers usually sit on low chairs with their feet on the
back rail of another chair, resting their pillow against the back of the
chair. Ladies will find that they can work most comfortably seated on
an ordinary chair with their feet on a footstool, and resting the pillow
against a table.
Nos. 6 and 7.—Passements.

These are the passements for two useful lace sprays.
No. 6.—Old Sprig Passement.

No. 7.—Turkey Tail Passement.
Whole or Plain Pillow Lace Stitch.

Four bobbins. No. 8 Mecklenburg thread. In order to explain this
stitch we must have recourse to numbers, and mark the bobbins No.
1, 2, 3, 4, No. 1 being the left-hand bobbin.
Pass No. 3 between Nos. 1 and 2; pass No. 1 over No. 3; pass No.
2 over No. 4, and pass No. 4 between Nos. 3 and 1.
Nos. 1 and 2 are the leaders, so called because they lead across
and back, weaving the whole or plain stitch across, and in the pairs
of leaders which form the edge in lace braid and lace patterns. When
one stitch is completed, the two left-hand bobbins are dropped and
two bobbins taken from the right hand to make with the two leaders
the four bobbins necessary for each stitch. Care must be taken that
the bobbins are not twisted, but that they lie each in its proper
place. In beginning work across from left to right of the pillow and
back again.
Nos. 8 and 9.—Maltese Cross and Stem Work.

No. 8 is a Maltese cross entirely executed in whole or plain pillow
lace stitch. This is a modern Honiton design.
No. 9 is an enlarged illustration of stem work and leaves.
No. 8.—Maltese Cross.

No. 9.—Stem Work and Leaves.
Half Stitch.
Mark the bobbins 1, 2, 3, 4, as for whole stitch. Pass No. 3
between Nos. 1 and 2, pass No. 1 over No. 3, pass No. 2 over No. 4,
leaving No. 4 alone, and continue with the next bobbins as in
learning whole stitch. It will be observed that half stitch is worked
like whole stitch except the last movement of passing No. 4 between
Nos. 3 and 1; to this omission the stitch owes its name. Half stitch is
used largely for the lighter parts of Honiton lace, and is shown mixed
in the rose, page 43.
Devonshire Turn or Wheel
To wind the threads on the bobbins a turn or wheel is used. No.
10 shows a Devonshire turn. The thread or cotton is always wound
from right to left, and a loop is made to prevent the bobbin from
unwinding. On gentle pressure the thread unwinds, but will hang in
place perfectly if the loop is made properly. All the bobbins should be
kept at the same length from the pillow to which they are attached.
No. 10.—Devonshire Turn.
Honiton Lace Stitch.

Honiton lace thread, No. 8 or No. 12 (No. 8 for beginners), nine
pairs of bobbins, one pair bobbins of Mecklenburg thread No. 12,
three pairs of leaders, one Mecklenburg thread bobbin next each of
the side leaders. Work the third pair of leaders across, twist three
times plain stitch (the plain stitch is worked with the leaders at each
side left and right), twist three times, insert pin, work back, twist
three times, one plain stitch, twist three times, pin.
This stitch is used for most of the Honiton patterns, and the
learner will observe that it is the same as braid work, p. 12, only the
leaders twist three times instead of twice. No notice is taken of the
Mecklenburg thread bobbins: the leaders work over and under these
as with the ordinary bobbins.
Nos. 11 and 12.—Cue Borders.

No. 11.—The cue border is worked from stem to stem, with 7 pair.
Take up a sewing with pins on either side. 8 or 9 bobbins are used,
according to if cottons are worked or not.
No. 12.—Baby cue is worked in the same manner.
No. 11.-Cue Pattern.
No. 12.—“Baby-Cue” Pattern.

To Form a Pattern.
When the whole art and mystery of whole and half stitch are
acquired so as to be executed with mechanical ease, the learner is
naturally anxious to make some use of her work, and to form a
pattern. The trefoil design, No. 8, is excellent to begin with, as it is
less intricate than many others.
In forming a pattern it will be observed that the outer edge of the
design has more pinholes than are in the inner ring of the same
design. The mode of decreasing the number of pinholes is as
follows:—* After making the stitch, working across to the out rim
and returning, take out the last pin of the inner circle and use it for
the next pinhole stitch, thus making one pinhole and pin serve for
two stitches. After this variation in the work, work backwards and
forwards in the usual way before repeating from *.
No. 13.—Rose Border.

No. 13 shows a rose border, worked with 7 pairs without cottons,
8 pair with, commencing at the stem.
No. 13.—Rose Border.
To Take Up a Sewing.
To perform this delicate operation a needle-pin is required; this
needle-pin is a fine long needle placed in a wooden handle. The real
Devonshire needle-pin is said to be a No. 8 needle stuck in a lucifer-
match.
With the needle-pin draw the thread through the edge of the lace,
pass the nearest bobbin through the loop thus formed, tie, and work
on as before.
Throwing and Tying.

These terms are used to express the mode of finishing off any
part of a sprig or design. To effect this:—Take up two leaders and tie
them in a plain knot, lay them out right and left, throw all the
bobbins in between them again, tie in a plain knot, and either cut off
the bobbins or work on as directed.
No. 14.—One Rose Border.

No. 14. one-rose border, is a succession of rose sprigs worked with
seven pairs of bobbins without cotton, is begun at the stem. For
one-rose sprig, see No. 20.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade
Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.
Let us accompany you on the journey of exploring knowledge and

personal growth!
ebookfinal.com

Modeling and Simulation in Ecotoxicology with Applications in MATLAB and Simulink Kenneth R. Dixon 2024 Scribd Download

Uploaded by

Copyright:

Available Formats

Modeling and Simulation in Ecotoxicology with Applications in MATLAB and Simulink Kenneth R. Dixon 2024 Scribd Download

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modeling and Simulation in Ecotoxicology with Applications in MATLAB and Simulink Kenneth R. Dixon 2024 Scribd Download

Uploaded by

Copyright:

Available Formats

Download the full version of the ebook at

Modeling and Simulation in Ecotoxicology with

Explore and download more ebook at https://ebookfinal.com

System Simulation Techniques with MATLAB and Simulink 1st

Modeling and Simulation of Systems Using MATLAB and

Computational Nanotechnology Modeling and Applications

Signals and Systems with MATLAB Computing and Simulink

Electronically Scanned Arrays MATLAB Modeling and

Numerical Simulation of Optical Wave Propagation With

Modeling and Simulation in Polymers 1st Edition

Agent based modeling and simulation with Swarm 1-st

© 2012 by Taylor & Francis Group, LLC

No claim to original U.S. Government works

International Standard Book Number-13: 978-1-4398-5518-8 (eBook - PDF)

and the CRC Press Web site at

Chapter 1 Introduction ..................................................................................................................1

Chapter 2 Principles of Modeling and Simulation........................................................................ 9

Chapter 3 Introduction to MATLAB® and Simulink®................................................................ 31

Chapter 4 Introduction to Stochastic Modeling .......................................................................... 47

4.4 Monte Carlo Simulation .................................................................................. 69

Chapter 5 Modeling Ecotoxicology of Individuals..................................................................... 73

Chapter 6 Modeling Ecotoxicology of Populations, Communities, and Ecosystems............... 109

Chapter 7 Parameter Estimation ............................................................................................... 125

Chapter 8 Designing Simulation Experiments ......................................................................... 147

8.2 Response Surface Designs............................................................................. 151

Chapter 9 Analysis of Simulation Experiments........................................................................ 159

Chapter 10 Model Validation...................................................................................................... 175

11.3.1.2 Consumption of Chlorpyrifos Granules ........................... 193

Chapter 12 Case Study: Predicting Health Risk to Bottlenose Dolphins

Chapter 13 Case Study: Simulating the Effects of Temperature Plumes

13.4.5 Estimate Parameters for Gross Uptake Model................................. 228

The MathWorks, Inc.

1.1 THEORIES UNDERLYING PREDICTIVE MODELS

Disciplinary Logic and Theory of

Mechanistic Design and Statistical

1.3 REASONS FOR MODELING AND SIMULATION

1.2.1 Alternatives and Their Consequences

1.2.2 Relative Predictive Ability

1.2.4 Hypothesis and Theory Construction

1.2.5 Nonexistent Universes

1.2.7 Planning and Management Decision Aid

1.2.8 System Identification

1.2.9 Unanticipated Effects

1.3 WHAT DOES IT TAKE TO BE A MODELER?

1.4 WHY MODELS FAIL: A CAUTIONARY NOTE

1.4.1 Poor Data for Parameter Estimation

1.4.2 Uncertainty Not Considered

1.4.3 Bias (Political, Social, Economic)

1.4.4 Lack of Understanding of Real-World Systems

1.4.5 Misuse of Mathematics

2.1.2 System Input and Output

2.1.3 Control Systems

of system is called a control system. A typical example of a control system is a thermostat-

2.1.5 System States: Steady State versus Transient States

2.1.6 Discrete versus Continuous

geographic location. This is an important consideration in simulating contaminant effects because

2.1.7 Linear versus Nonlinear

where, in general, a is a function.

2.2.1.1 Solution of Ordinary First-Order Differential Equations

Now, the free response, Qa(t) is:

1.1 THEORIES UNDERLYING PREDICTIVE MODELS

1.3 REASONS FOR MODELING AND SIMULATION

1.2.1 Alternatives and Their Consequences

1.2.2 Relative Predictive Ability

1.2.4 Hypothesis and Theory Construction

1.2.5 Nonexistent Universes

1.2.7 Planning and Management Decision Aid

1.2.8 System Identification

1.2.9 Unanticipated Effects

1.3 WHAT DOES IT TAKE TO BE A MODELER?

1.4 WHY MODELS FAIL: A CAUTIONARY NOTE

1.4.1 Poor Data for Parameter Estimation

1.4.2 Uncertainty Not Considered

1.4.3 Bias (Political, Social, Economic)

1.4.4 Lack of Understanding of Real-World Systems

1.4.5 Misuse of Mathematics

2.1.2 System Input and Output

2.1.3 Control Systems

2.1.5 System States: Steady State versus Transient States

2.1.6 Discrete versus Continuous

2.1.7 Linear versus Nonlinear

2.2.1.1 Solution of Ordinary First-Order Differential Equations

2.2.1.2 Steady-State and Transient Response

2.2.1.3 Difference Equation Approximation to Differential Equation

2.2.1.4 Numerical Solutions to Differential Equations

2.2.2 Block Diagrams

2.2.3 Stochastic Models

2.2.4 Individual-Based Models

2.2.5 Aggregated Models