You are on page 1of 537


This page intentionally left blank


Copenhagen University,
Faculty of Pharmaceutical Sciences,
Institute A,
Section of Environmental Chemistry, Toxicology and Ecotoxicology,
University Park 2,
Copenhagen , 2100,


Elsevier B.V.
Radarweg 29, 1043 NX Amsterdam, The Netherlands

First edition 2009

Copyright  2009 Elsevier B.V. All rights reserved

The following article is US government works in the public domain and is not subject to copyright:

No part of this publication may be reproduced, stored in a retrieval system

or transmitted in any form or by any means electronic, mechanical, photocopying,
recording or otherwise without the prior written permission of the publisher.

Permissions may be sought directly from Elseviers Science & Technology Rights
Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333;
email: Alternatively you can submit your request online by
visiting the Elsevier web site at, and selecting
Obtaining permission to use Elsevier material

No responsibility is assumed by the publisher for any injury and/or damage to persons
or property as a matter of products liability, negligence or otherwise, or from any use
or operation of any methods, products, instructions or ideas contained in the material herein.
Because of rapid advances in the medical sciences, in particular, independent
verification of diagnoses and drug dosages should be made.

British Library Cataloguing in Publication Data

A catalogue record for this book is available from the British Library

Library of Congress Catalog Number: 2009929111

ISBN: 978 0 444 53466 8

For information on all Elsevier publications

visit our website at

Printed and bound in Italy

09 10 11 12 13 10 9 8 7 6 5 4 3 2 1

Contents v-vi
Contributors vii-ix
Preface xi


ECOSYSTEM SERVICES K A Brauman and G C Daily 26



BODY-SIZE PATTERNS A Basset and L Sabetta 44


EMERGENT PROPERTIES F Muller and S N Nielsen 91

SELF-ORGANIZATION D G Green, S Sadedin, and T G Leishman 98
EXERGY S E Jrgensen 128

S E Jrgensen 140


AGRICULTURE SYSTEMS O Andren and T Katterer 145


vi Contents

ALPINE FOREST W K Smith, D M Johnson, and K Reinhardt 156

CAVES F G Howarth 190
CHAPARRAL J E Keeley 195

CORAL REEFS D E Burkepile and M E Hay 201

DESERT STREAMS T K Harms, R A Sponseller, and N B Grimm 214
DESERTS C Holzapfel 222
DUNES P Moreno-Casasola 241
FLOODPLAINS B G Lockaby, W H Conner, and J Mitchell 253

FOREST PLANTATIONS D Zhang and J Stanturf 264

LAGOONS G Harris 296


RIPARIAN WETLANDS K M Wantzen and W J Junk 342

ROCKY INTERTIDAL ZONE P S Petraitis, J A D Fisher, and S Dudgeon 374
SALT MARSHES J B Zedler, C L Bonin, D J Larkin, and A Varty 384
SAVANNA L B Hutley and S A Setterfield 394

STEPPES AND PRAIRIES J M Briggs, A K Knapp, and S L Collins 405

SWAMPS C Trettin 414
TEMPERATE FOREST W S Currie and K M Bergen 417
TUNDRA R Harmsen 443

UPWELLING ECOSYSTEMS T R Anderson and M I Lucas 450

URBAN SYSTEMS T Elmqvist, C Alfsen, and J Colding 461

W H Adey L M Chu
Smithsonian Institution, Washington, DC, USA The Chinese University of Hong Kong, Hong Kong SAR,
Peoples Republic of China
C Alfsen
UNESCO, New York, NY, USA E A Colburn
Harvard University, Petersham, MA, USA
T F H Allen
University of Wisconsin, Madison, WI, USA J Colding
Royal Swedish Academy of Sciences, Stockholm,
S Allesina Sweden
University of Michigan, Ann Arbor, MI, USA
S L Collins
T R Anderson University of New Mexico, Albuquerque, NM, USA
National Oceanography Centre, Southampton, UK
W H Conner
O Andren Baruch Institute of Coastal Ecology and Forest Science,
TSBF-CIAT, Nairobi, Kenya Georgetown, SC, USA

A Basset K W Cummins
Universita` del Salento Lecce, Lecce, Italy Humboldt State University, Arcata, CA, USA

K M Bergen W S Currie
University of Michigan, Ann Arbor, MI, USA University of Michigan, Ann Arbor, MI, USA

C L Bonin G C Daily
University of Wisconsin, Madison, WI, USA Stanford University, Stanford, CA, USA

H Bossel R F Dame
University of Kassel (retd.), Zierenberg, Germany Charleston, SC, USA

K A Brauman D L DeAngelis
Stanford University, Stanford, CA, USA University of Miami, Coral Gables, FL, USA

J M Briggs S Dudgeon
Arizona State University, Tempe, AZ, USA California State University, Northridge, CA, USA

D E Burkepile T Elmqvist
Georgia Institute of Technology, Atlanta, GA, USA Stockholm University, Stockholm, Sweden

T V Callaghan B D Fath
Royal Swedish Academy of Sciences Abisko Scientific Towson University, Towson, MD, USA and
Research Station, Abisko, Sweden International Institute for Applied System Analysis,
Laxenburg, Austria
J L Casti
International Institute for Applied System Analysis, J A D Fisher
Laxenburg, Austria University of Pennsylvania, Philadelphia, PA, USA

viii List of Contributors

D G Green D J Larkin
Monash University, Clayton, VIC, Australia University of Wisconsin, Madison, WI, USA

N B Grimm T G Leishman
Arizona State University, Tempe, AZ, USA Monash University, Clayton, VIC, Australia

R Harmsen B G Lockaby
Queens University, Kingston, ON, Canada Auburn University, Auburn, AL, USA

T K Harms M I Lucas
Arizona State University, Tempe, AZ, USA National Oceanography Centre, Southampton, UK

G Harris
F Medail
University of Tasmania, Hobart, TAS, Australia
IMEP Aix-Marseille University, Aix-en-Provence,
M E Hay
Georgia Institute of Technology, Atlanta, GA, USA
J M Melack
University of California, Santa Barbara, Santa Barbara,
R A Herendeen
University of Vermont, Burlington, VT, USA

C Holzapfel J Mitchell
Rutgers University, Newark, NJ, USA Auburn University, Auburn, AL, USA

F G Howarth P Moreno-Casasola
Bishop Museum, Honolulu, HI, USA Institute of Ecology AC, Xalapa, Mexico

L B Hutley F Muller
Charles Darwin University, Darwin, NT, Australia University of Kiel, Kiel, Germany

D M Johnson S N Nielsen
USDA Forest Service, Corvallis, OR, USA Danmarks Farmaceutiske Universitet, Copenhagen,
S E Jrgensen
Copenhagen University, Copenhagen, Denmark D W Orr
Oberlin College, Oberlin, OH, USA
W J Junk
Max Planck Institute for Limnology, Plon, Germany M Pell
Swedish University of Agricultural Sciences, Uppsala,
P C Kangas Sweden
University of Maryland, College Park, MD, USA
P S Petraitis
T Katterer University of Pennsylvania, Philadelphia, PA, USA
Department of Soil Sciences, Uppsala, Sweden
K Reinhardt
P Keddy
Wake Forest University, Winston-Salem, NC, USA
Southeastern Louisiana University, Hammond, LA, USA

J E Keeley L Sabetta
University of California, Los Angeles, CA, USA Universita` del Salento Lecce, Lecce, Italy

A K Knapp S Sadedin,
Colorado State University, Fort Collins, CO, USA Monash University, Clayton, VIC, Australia

V Krivtsov A K Salomon
University of Edinburgh, Edinburgh, UK University of California, Santa Barbara, Santa Barbara,
C Korner
Botanisches Institut der Universitat Basel, Basel, U M Scharler
Switzerland University of KwaZulu-Natal, Durban, South Africa
List of Contributors ix

S A Setterfield D H Vitt
Charles Darwin University, Darwin, NT, Australia Southern Illinois University, Carbondale, IL, USA

W K Smith R B Waide
Wake Forest University, Winston-Salem, NC, USA University of New Mexico, Albuquerque, NM, USA

M Soderstrom
Montreal, QC, Canada K M Wantzen
University of Konstanz, Konstanz, Germany
R A Sponseller
Arizona State University, Tempe, AZ, USA M A Wilzbach
Humboldt State University, Arcata, CA, USA
J Stanturf
Center for Forest Disturbance Science,
Athens, GA, USA A Worman
The Royal Institute of Technology, Stockholm,
C Trettin
USDA, Forest Service, Charleston, SC, USA
J B Zedler
R R Twilley University of Wisconsin, Madison, WI, USA
Louisiana State University, Baton Rouge, LA, USA
D Zhang
R E Ulanowicz
Auburn University, Auburn, AL, USA
University of Maryland Center for Environmental Science,
Solomons, MD, USA
J-J Zhu
A Varty Institute of Applied Ecology, CAS, Shenyang, Peoples
University of Wisconsin, Madison, WI, USA Republic of China
This page intentionally left blank

S ystems ecology, also called ecosystem theory, offers today a complete theory about how ecosystems are working as
systems. The theory will inevitably be improved in the coming years, when it hopefully will be used increasingly to
explain ecological observations and to facilitate environmental management including the use of ecotechnology. The
theory is, however, sufficiently developed today to be presented as a complete theory that offers a wide spectrum of
applications. Only through a wider application of the theory or let us call what we have today propositions of a theory
it will be possible to see the shortcomings of the present theory and propose improvement of the theory.
The book consists of three parts. The part Ecosystems as Systems emphasizes the system properties of ecosystems
including the presentation of basic scientific propositions to a theory in the chapter Fundamental Laws in Ecology, while
the part Ecosystem Properties gives a more comprehensive overview of the holistic properties of ecosystems, which of
course not surprisingly are rooted in the system properties and covered by the propositions. The part Ecosystems
gives an overview of different types of ecosystems, how they function due to their characteristic ecosystem properties,
and how the scientific propositions can be applied to understand and illustrate their characteristic properties.
It is my hope that this book will be utilized intensively by ecologists and system ecologists to gain a deeper
understanding of ecosystems and their function and to initiate the development of ecology toward a more theoretical
science that can explain and predict reactions of ecosystems. By such a development, it will be possible to replace many
measurements that are often expensive to perform with sound theoretical considerations.
The book is based on the presentation of
I. systems ecology as an ecological subdiscipline and
II. a very comprehensive overview of all types of ecosystems with many illustrations of their characteristic properties

in the recently published Encyclopedia of Ecology.

Due to an excellent work by the editor of the Ecosystem Section, Donald de Angelis, and the editor of the Systems
Ecology Section, Brian Fath, in the Encyclopedia of Ecology, it has been possible to present a comprehensive and very
informative overview of all types of ecosystems and an updated ecosystem theory. I would therefore like to thank Donald
and all the authors of ecosystem entries and Brian Fath and all the authors of systems ecology entries for their
contributions to the Encyclopedia of Ecology, which made it possible to produce this broad and up to date coverage of a
very important subdiscipline in ecology.

Sven Erik Jrgensen

Copenhagen, May 2009

This page intentionally left blank
This page intentionally left blank
S E Jrgensen, Copenhagen University, Copenhagen, Denmark
2009 Elsevier B.V. All rights reserved.

According to the definition by Tansley (1935), an ecosys their characteristic ecosystem properties, and how the
tem is an integrated system composed of interacting biotic scientific propositions can be applied to understand and
and abiotic components. It is important in this definition illustrate their characteristic properties. The part
that an ecosystem is a system, which implies that it has Ecosystems as Systems emphasizes the system properties
boundaries and that we can distinguish between the sys of ecosystems and also presents basic scientific proposi
tem and its environment environment in principle tions, while the part Ecosystem Properties gives a more
means the rest of the world beyond the boundaries of comprehensive overview of the holistic properties of
the system. The components biotic as well as abiotic ecosystems, which of course not surprisingly are
are interacting, which means that they are connected rooted in the system properties.
directly or indirectly. All systems that encompass inter The chapters Ecosystem Ecology, Ecological System
acting biotic and abiotic components may be considered Thinking, and Ecosystems in the part Ecosystems as
as an ecosystem. A drop of polluted water may for Systems focus on the most fundamental system properties
instance be considered an ecosystem, because it contains that are derived from the above presented definition of
microorganisms, organic matter, and inorganic salts and ecosystems. The definition is repeated in all three chap
these components are interacting. Usually, our ecosystem ters with slight modifications. The system properties
research and management is interested in a larger area of presented in these three chapters may be summarized as
nature characterized by its function and properties, for follows:
instance a lake, a forest, or a wetland. All these three 1. Ecosystems cycle energy.
examples of ecosystems have very characteristic functions 2. Ecosystems cycle matter.
and have several unique properties that are different from 3. Life and environment are connected, which implies
other types of ecosystems. The scale that is applied for the that the environment of an ecosystem influences the
definition of an ecosystem is dependent on the function of ecosystem. This influence determines the prevailing
the ecosystem and is determined by the addressed conditions of the ecosystems, or expressed differently
problem. the external variables (also called forcing functions)
Because an ecosystem has interacting and con determine the conditions for the internal variables
nected biotic and abiotic components, it has system (also called state variables) of an ecosystem. The
properties in the sense that the components work wide spectrum of different ecosystems (the part
together to give the system emerging properties and Ecosystems gives an overview) is the result of an over
make the system more than just the sum of the com whelmingly large number of different conditions
ponents. A living organism is much more than the (combinations of external variables).
cells and the organs that make up the organism. 4. Ecosystems are whole systems and studies of ecosys
Similarly, a forest is more than just the trees it is tem dynamics therefore require holistic views.
a cooperative working unit with emerging unique
The human society is very dependent on the proper
properties characteristic of a forest.
functioning of ecosystems, because humans are using a
It is important to understand fully the function and the
wide spectrum of services offered by the ecosystems. It is
reactions of ecosystems in both ecological research and
therefore important to understand the ecosystem proper
environmental management. The two basic questions in
ties on which these services are based. The chapter
this context are
Ecosystem Services and partly the chapter Ecosystems
1. Which fundamental properties characterize present the ecosystem services, which may be classified
ecosystems? into three groups:
2. Is it possible to formulate basic scientific propositions
production services as we know them from agriculture,
that are able to explain the functions of ecosystems?
fishery, forestry, and so on;
It is attempted to answer these two core questions in the regulation services due to cycling, filtration, transloca
parts Ecosystems as Systems and Ecosystem Properties of tion, and stabilization processes;
this book, while the part Ecosystems gives an overview of cultural services such as recreation, spiritual inspiration,
different types of ecosystems, how they function due to and esthetic beauty.

4 Introduction

The chapter Fundamental Laws in Ecology gives a brief cycling the growth and development of biological compo
summary of the ecosystem properties that are rooted in nents would stop due to the lack of one or more essential
the system properties of ecosystems: elements. The chapter Cycling and Cycling Indices covers
Ecosystems are complex (many steadily varying inter cycling and cycling indices, which quantify the networks
acting components). possibilities to support the cycling processes.
Ecosystems are open. The chapters Ecological Network Analysis, Ascendancy;
Ecosystems are hierarchically organized. Ecological Network Analysis, Energy Analysis; Ecological
Ecosystems are self organizing and self regulated due Network Analysis, Environ Analysis; and Indirect Effects in
to a very large number of feedback mechanisms. Ecology present different aspects of the ecological network.
Network analysis, ENA (Ecological Network Analysis), uses
These properties are discussed in more detail in the part network theory to study the interactions between organisms
Ecosystem Properties. or populations within their environment. Ascendancy,
The chapter Fundamental Laws in Ecology proposes 10 which is covered in the chapter Ecological Network
fundamental laws of ecosystems that are consistent with the Analysis, Ascendancy, quantifies the efficiency of the net
system properties presented in the other chapters of the works on the basis of the actual flows. Development of an
part Ecosystems as Systems. The 10 propositions are able ecosystem will usually imply that the ascendancy is increas
to explain ecosystem behavior and properties. The funda ing. The chapter Ecological Network Analysis, Energy
mental tentative laws presented in this chapter are Analysis analyzes the ecological network by use of the
furthermore able to explain many ecological observations energy flows, while the chapter Indirect Effects in Ecology
and rules, which is a great advantage of having a good uses the so called environ analysis. Each object in the system
theory. By use of the theory, it is possible to conclude, has two environs, one receiving and one generating inter
without the need for observations, how an ecosystem will actions in the system. It is by analyzing these flows that it is
react to different impacts. It is therefore indeed possible to possible to deduce network properties such as network
improve research plans and develop environmental man mutualism and network synergy. Cycling the topic of the
agement plans on the basis of theoretical considerations. chapter Cycling and Cycling Indices may of course also be
The 10 propositions (tentative laws) can be shown to be considered a network property. The chapter Indirect Effects
rooted in five basic ecological system properties. in Ecology focuses on perhaps the most important network
The part Ecosystem Properties gives more information property: the presence of a strong indirect effect that in many
on the basic properties of an ecosystem. The chapter cases may even exceed the direct effect.
Autocatalysis focuses on autocatalysis, which frequently The chapter Emergent Properties deals with the topic of
increases the efficiencies and rates of ecological processes. emergent properties the ecosystem as an integrated sys
The chapter Body Size Patterns discusses the body size tem is more than the sum of the components. The emergent
pattern of ecosystems. The rate of biological processes properties are the result of all the system properties. Due to
such as growth, metabolism, mortality, generation time, the synergistic effect of the network, autocatalysis, cycling,
and respiration is dependent on the size of the organisms. self regulation and self organization, and so on, an ecosys
The spectrum of conditions in an ecosystem determines the tem acquires a number of very useful, holistic properties as a
spectrum of these fundamental ecological processes, which system properties that are often called emergent proper
would allow the best utilization of the resources in ecosys ties. Self organization itself is perhaps the most clear
tems. It implies that the conditions also determine the body example of an emergent property. The chapter Self orga
size pattern. Different ecosystems at different conditions nization looks into the emergent property of self
may therefore have a different body size pattern, which organization and how it is rooted in complex adaptive
therefore becomes a characteristic property of an ecosystem. ecosystems. This chapter discusses how the spatial patterns,
All ecosystems cycle the elements that are essential for persistence, stability, and ability to develop and evolve can
the living matter, and thereby the growth and development be explained as a result of the self organization. The differ
of ecosystems can continue, because the essential elements ences between ecosystems at an early stage and mature
are steadily recovered with a certain rate. The living mat ecosystems can also be explained by self organization.
ter needs about 22 different elements, of which the cycling Ecosystems are very complex systems. They have a
of nitrogen, carbon, phosphorus, sulfur, silica, calcium, large number of components with a large diversity, hier
sodium, and magnesium is of utmost importance. The archical organization, and nonlinear behavior. The
cycling is possible due to the ecological networks that are chapter Ecological Complexity presents various aspects
formed in all ecosystems. The network may be considered of ecological complexity, while the chapter Hierarchy
a map of the connections of abiotic and biotic components. Theory in Ecology presents the application of hierarchy
The network indicates the possibilities for interactions theory in ecology. The hierarchical organization makes it
among the components of the ecosystem. Obviously, possible to overview the complexity. It is also possible to get
cycling is very important for ecosystems, because without a better overview of the complex behavior of ecosystems by
Introduction 5

Table 1 The five basic properties that are rooted in the 10 tentative fundamental laws encompass all the system properties presented

Basic property Derived system properties

1. Ecosystems are open The forcing functions (external variables) determine the ecosystem conditions
2. Ecosystems have directionality Ecosystems show autocatalysis
Ecosystems grow and develop
Ecosystems have the propensity to maximize exergy storage and power
Ecosystems have a body size pattern
3. Ecosystems have connectivity The biotic and abiotic components of an ecosystem are connected in a network
The network gives the ecosystem mutualism and synergy
The indirect effect is significant due to the network and may even exceed the direct
Ecosystems are self-organizing and self-regulated
Ecosystems cycle energy, matter, and information
4. Ecosystems have emergent hierarchies Ecosystems are organized hierarchically
5. Ecosystems have complex dynamics Ecosystems grow and develop by increasing the biomass, the network, and the level of
Ecosystems are adaptive systems
Ecosystems grow and develop and can cope with disturbances by a propensity to
increase the exergy storage and the power
Ecosystems, particularly under natural conditions, often have a large diversity, which
gives the ecosystems a wide spectrum of different buffer capacities
Ecosystems have high buffer capacities as a result of the complex dynamics
Ecosystems recover usually rapidly and effectively after disturbances

use of goal functions and orientors that are presented in the The chapter Overview of Ecosystem Types, Their
chapter Goal Functions and Orientors. They are able to Forcing Functions, and Most Important Properties, which
quantify the development of ecosystems as a result of the is the last chapter in the part Ecosystem Properties, gives an
complex dynamics of ecosystems. One of the most useful overview of the 39 different types of ecosystems that are
orientors is exergy, which is presented in the chapter presented in the part Ecosystems. For all the 39 ecosystem
Exergy. The complex dynamics of ecosystems determine types, the most important forcing functions are indicated,
how they are able to develop and cope with disturbances. that is, the forcing functions (impacts) that may be consid
The exergy or energy that can do work of ecosystems we ered a threat to the ecosystem or the forcing functions that
cannot calculate exergy for an ecosystem due to its enor most frequently determine the ecosystem function. It is
mous complexity but we can calculate exergy for a model of possible to classify the forcing functions of the 39 ecosystems
the ecosystem will have the tendency to be as high as into four groups. The most basic properties of the four
possible under the prevailing conditions. Disturbances may ecosystem classes are presented. They are the result of the
of course cause a reduction in the ecosystem exergy, but the prevailing conditions that are determined by the forcing
organisms try to organize themselves by their network and functions. The most important properties are those that
interactions to get the best out of the situation it means in need to be maintained for the ecosystem to be able to meet
the Darwinian sense most survival, which may be expressed the threats or those that are particularly important for the
by exergy, as it covers the product of biomass and informa maintenance of the ecosystem function in spite of the impact.
tion of the ecosystem. The part Ecosystems has 40 chapters covering 39
The five fundamental properties (see chapter different types of ecosystems. Most of the Earths ecosys
Fundamental Laws in Ecology) cover all the ecosystem tems are covered by the 39 types of ecosystems. A few
properties that are presented in the parts Ecosystems as rare types of ecosystems are not included, but all ecosys
Systems and Ecosystem Properties. An overview of the tems frequently represented in nature are included. The
five basic properties and the derived additional system ecosystems that are not included will however have prop
erties close to one or more of the 39 types covered.
properties can be obtained from Table 1. Some of the
properties are derived from more than one of the five
fundamental properties, but to simplify the overview the See also: Autocatalysis; Body Size Patterns; Cycling and
derived system properties are associated with one of the Cycling Indices; Ecological Complexity; Ecological
basic properties. Particularly, the basic property that eco Network Analysis, Ascendancy; Ecological Network
systems have connectivity, which means that they form a Analysis, Energy Analysis; Ecological Network Analysis,
network, and have a complex dynamics has been used to Environ Analysis; Ecosystem Ecology; Ecosystem
derive several system properties that could also be derived Services; Ecological System Thinking; Ecosystems;
partly from one of the four other basic properties. Emergent Properties; Exergy; Fundamental Laws in
6 Ecosystem Ecology

Ecology; Goal Functions and Orientors; Hierarchy Theory Jrgensen SE (2008b) Evolutionary Essays. A Thermodynamic
Interpretation of the Evolution, 210pp.
in Ecology; Indirect Effects in Ecology; Overview of
Jrgensen SE (ed.) (2008a) Encyclopedia of Ecology, 5 vols. 4122pp,
Ecosystem Types, Their Forcing Functions, and Most Amsterdam: Elsevier.
Important Properties; Self-Organization. Jrgensen SE and Fath B (2007) A New Ecology. Systems
Perspectives. 275pp. Amsterdam: Elsevier.
Jrgensen SE, Patten BC, and Straskraba M (2000) Ecosystems
Further Reading emerging: 4. growth. Ecological Modelling 126: 249 284.
Jrgensen SE and Svirezhev YM (2004) Towards a Thermodynamic
Jrgensen SE (2004) Information theory and energy. In: Cleveland CJ (ed.) Theory for Ecological Systems. 366pp. Amsterdam: Elsevier.
Encyclopedia of Energy, vol. 3. pp. 439 449. San Diego, CA: Elsevier. Ulanowicz R, Jrgensen SE, and Fath BD (2006) Exergy, information
Jrgensen SE (2006) Eco Exergy as Sustainability. 220pp. and aggradation: An ecosystem reconciliation. Ecological Modelling
Southampton: WIT Press. 198: 520 525.

Ecosystem Ecology
B D Fath, Towson University, Towson, MD, USA and International Institute for Applied System Analysis,
Laxenburg, Austria
2008 Elsevier B.V. All rights reserved.

Introduction Ecosystem Studies

History of the Ecosystem Concept Human Influence on Ecosystems
Defining an Ecosystem Summary
Energy Flow in Ecosystems Further Reading
Biogeochemical Cycles

Introduction abstracting to energetic or material units. The advantage

of this abstraction, of course, is that energy and mass are
Ecology is a broad and diverse field of study. One of the conserved quantities, whereas number of individuals is
basic distinctions in ecology is between autecology and not. Therefore, using conserved units it is possible to
synecology, in which the former is considered the ecology construct balance equations and inputoutput models.
of individual organisms and populations, mostly concerned In fact, dimensionally, ecosystem ecology has more in
with the biological organisms themselves; and the latter, the common with organismal ecology in which the thermo
ecology of relationships among the organisms and popula regulation and physiology of a single organism is
tions, which is mostly concerned with communication of studied, which also often relies on energetic units.
material, energy, and information of the entire system of Indeed, all scales of ecological study have a role to
components. In order to study an ecosystem, one must have contribute to general scientific understanding and have
knowledge of the individual parts; thus, it is dependent on been developed to address a wide range of interesting
fieldwork and experiments grounded in autecology, but the and relevant questions regarding the natural world and
focus is much more on how these parts interact, relate to, the impact humans have on it.
and influence one another including the physical environ
mental resources on which life depends. Ecosystem
ecology, therefore, is the implementation of synecology. History of the Ecosystem Concept
In this manner, the dimensional units used in ecosystem
studies are usually the amount of energy or matter moving Systems concepts of the environment have long played a
through the system. This differs from population and com role in the development of ecology as a discipline, but
munity ecology studies in which the dimensional units are these came to a head in the early twentieth century.
typically the number of individuals (Table 1). This simple During this period, the two dominant and competing
dimensional difference has served as an unfortunate divide ecological paradigms were the organismic (e.g.,
between research conducted at the different ecological Clements) and individualistic (e.g., Gleason) views. The
scales. While ecosystem ecologists maintain that it is always organismic approach held that communities and ecosys
possible to convert species numbers into biomass or nutri tems were discernible objects that had an inherent and
ent mass, population and community ecologists often feel organized complexity resulting in a cybernetic and self
that too much unique biological detail is discarded by governing system, similar in ways to how an organism
Ecosystem Ecology 7

Table 1 Typical dimensional units of study at different describes the ecosystem thus, . . . the fundamental concep
ecological scales tion is . . . the whole system, including not only the
Ecological scale Dimensions organism complex, but also the whole complex of physical
factors forming what we call the environment of the biome
Organismal ecology dE/dt the habitat factors in the widest sense. The definition he
Population ecology dN/dt
proposed over 70 years ago sounds fresh today, since it has
Community ecology dN/dt
Ecosystem ecology dE/dt changed little if at all. The major tenets of this approach are
the explicit inclusion of abiotic processes interacting with
dE/dt change in energy over time; dN/dt change in number over time. the biota in this sense it is more along the Haeckelian lines
of ecology than the Darwinian, with an additional emphasis
regulates itself. The individualistic approach held that on the system. The latter tied the field closely to the bur
communities had observer dependent boundaries and geoning disciplines of general system theory and systems
internal development was stochastic and individual. In analysis.
this paradigm, the internal relations were synergistic, but While the conceptual underpinning of the ecosystem was
not cybernetic since the individual parts functioned inde now established, the introduction of this term was theoreti
pendently. The organismic ideas grew out of the cal, lacking guidance as to how it might be applied as a field
functional understanding of whole systems such as lakes, of study. There were around this time several whole system
and also out of the discussions involving how communities energy budgets being developed, particularly for lake eco
changed over time during succession. These ideas were systems by North American ecologists such as Forbes, Birge,
influenced by philosophers of the day such as Jan Smuts. and Juday in Wisconsin, and which were ideal test cases for
This was particularly true of German holists, such as the the ecosystem concept. Building on this work, in 1942,
limnology group at the Kaiser Wilhelm Instituts in Plon Lindemans study of Cedar Bog Lake also in Wisconsin
led by Thienemann, and others such as Leick (plant ecol was published, providing, for the first time, a clear applica
ogy) and Friedrich (zoology). Table 2 shows a summary of tion of the ecosystem concept. In addition to constructing the
some of the main ecosystem and related concepts. This food cycle of the aquatic system, he developed a metric
dialog between the holists and reductionists affected the now called the Lindeman efficiency to assess the efficiency
main currents of ecological thought during this period, and of energy movement from one trophic level to the next based
it was in part resolved by the introduction of ecosystem, on ecological feeding relations. His conceptual model of
which is both physical in nature and also systemic. Cedar Bog Lake included passive flows to detritus, but
The term ecosystem, which is ubiquitous today, both as these were not included in the trophic enumeration. Since
scientific terminology and in common vernacular, grew out then numerous additional studies have followed this same
of this climate. It was first used by Arthur Tansley in 1935 in approach and it has been applied to many habitats such as
a seminal paper in the journal Ecology, entitled The use and terrestrial, aquatic, and urban ecosystems.
abuse of vegetational concepts and terms. In fact, his reason
for coining the term ecosystem was in response, as the title Defining an Ecosystem
says, to a perceived abuse of community concepts by some
such as Clements and Cowles. While Tansley himself An ecosystem, as a unit of study, must be a bounded
brought a systems perspective, the community as organism system, yet the scale can range from a puddle, to a lake,
metaphor bothered him to the extent that he wanted to to a watershed, to a biome. Indeed, ecosystem scale is
provide a more scientific footing for the processes and inter defined more by the functioning of the system than by any
actions occurring during community development. Tansley checklist of constituent parts, and the scale of analysis
Table 2 Ecosystem and related concept

Year Term Author Concept

1887 Microcosm Forbes Broadening of the biocoenosis concept

1914 Ecoid Negri Unholistic, based on Gleasonian ideas
1928 Okologisches system Woltereck Still being used to avoid argument
1930 Holocoen Friedrich Holistic, biologistic
1935 Ecosystem Tansley Antiholistic, physicalist
1939 Biosystem Thienemann Stressing functional organization
1944 Geobioconose Sukacev Geographic, landscape ecological
1944 Bioinert body Vernadsky Biogeochemical
1948 Biochore Pallmann Landscape ecological
1950 Landschaft Troll Holistic, Gestalt viewing

Modified from Wiegleb G (2000) Lecture Notes on The History of Ecology and Nature Conservation.
8 Ecosystem Ecology

Photosynthesis Ecosystem boundary

Plant Herbivore
respiration Immigration/

production Respiration
Detritus and
Soil surface

Input transfers nutrients Output transfers


Figure 1 Conceptual diagram of a simplified ecosystem. Clear arrows, energy; dark arrows, biomass; blue arrows, water.

should be determined by the problem being addressed. methodologies of inputoutput analysis has developed
Whereas individuals perish over time and even popula into a powerful analysis tool for understanding complex
tions cannot survive indefinitely none can fix their own interactions and dependencies in ecological networks. For
energy and process their own wastes every ecosystem now though, let us concern ourselves more generally with
contains the ecological community necessary for sustain what occurs within the ecosystem boundary.
ing life: primary producers, consumers, and decomposers, Energy flow in ecosystems begins with the capture of
and the physical environment for oikos (Figure 1 shows a solar radiation by photosynthetic processes in primary
simple ecosystem model). It is this feature of ecosystems, producers (eqn [1]). Note, there are also chemoautotrophs
that they are the basic unit for sustaining life over the that capture energy in the absence of sunlight, but while
long term, which provides one of the main reasons for biologically fascinating, contribute negligible energy flux
studying them for environmental management and con to the overall global ecological energy balance
servation. The two main features of the ecosystem, energy
flow and nutrient biogeochemical cycling, comprise the Energy 6CO2 6H2 O ! C6 H12 O6 6O2 1
major areas of ecosystem ecology research. The accumulated organic matter, first as simple sugars
then combined with other elements to more complex
molecules, represents the gross primary production in the
Energy Flow in Ecosystems system, some of which is released and used for the primary
producers growth and maintenance through respiration:
The thermodynamic assessment of an ecosystem starts
C6 H12 O6 6O2 ! 6CO2 6H2 O Energy 2
with the recognition that an ecosystem is an open system,
in the sense of physics, such that it receives energy and The remainder, or net primary production, is available
matter input from outside its borders and transfers output for the rest of the ecosystem consumers including decom
back to this environment. Thus, every ecosystem must posers. Secondary production refers to the energetic
have a system boundary and must be embedded in an availability of the heterotrophic organisms, which
environment that provides low entropy energy input and accounts for the energy uptake by heterotrophs and the
can receive high entropy energy output. In addition to energy used for their maintenance. Overall ecosystem
the external resource sourcesink, there is another inter production is supported by the primary producers,
nal, within system boundary environment with which whereas ecosystem respiration includes the metabolic
each organism directly and indirectly interacts. Patten activity of all the ecosystem biota (Table 3). In this
proposed the concept of these two environments, one manner, plants provide the essential base for all ecological
external and mostly unknowable (other than the input food webs. Since it is often difficult to make direct mea
output interactions), and the second internal and measur surements of ecological production, the change in
able (i.e., external to the specific organismal component biomass measures growth, which can be used as represen
but within system boundary) as a systems approach to tative of production.
quantify indirect, yet within system interactions. This The captured energy moves through a reticulated net
approach called environ analysis relying on the work of interactions forming the complex dependency
Ecosystem Ecology 9

Table 3 Ecosystem energetics defined by net and gross production

Net primary production gross primary production respiration (autotrophs)

Net secondary production gross secondary production respiration (heterotrophs)
Net ecosystem production gross primary production ecosystem respiration (autotrophs heterotrophs)
Net production biomass (now) biomass (before)

patterns known as food webs. In a simplified food chain, material allows for the additional import of solar energy
and as first described by Lindeman, the trophic concept is until saturation is reached at about 80% of the available
used to assess the distance away from the original energy solar radiation. At this point the overall growth of the
importation, but in reality the multiple feeding pathways ecosystem begins to level off because although gross
found in ecological food webs make discrete trophic primary production is high, the overall system supports
levels a convenient yet inaccurate simplification. Elton more and more nonphotosynthetic biomass both in terms
observed that one typically finds a decreasing number of of nonphotosynthetic plant material and heterotrophs.
organisms as one proceeds up the food chain from pri When the average gross production is entirely utilized
mary producers to herbivores, carnivores, and top to support and maintain the existing structure, net pro
carnivores leading him to propose a pyramid of num duction is zero and the system has reached a steady state
bers. One can control for the individual variation in body regarding biomass growth. However, the ecosystem con
size by considering the biomass at each trophic level tinues to develop both in terms of the network
rather than the number of individuals resulting in a organization and in the information capacity. In addition
pyramid of biomass. The trophic pyramid is a thermo to being a dynamic steady state, it does not persist indefi
dynamically satisfying view of interactions since nitely because disturbances afflict the system setting it
according to the second law energy must be lost during back to earlier successional stages in which the growth
each transformation step; in addition, energy is used at and development processes begin anew, possibly with
each level for the maintenance of that level. Under this different results. In this manner, the disturbance acts
paradigm, the trophic levels apparently cap out around according to Hollings creative destruction providing the
five or six levels. Fractional trophic levels have been system the opportunity to develop along a different path
employed to account for organisms feeding at multiple way. Recent work on ecosystem growth and development
levels, but even these do not usually account for the role has focused on the orientation of thermodynamic
of detritus and decomposition, which extend the feeding indicators such as energy throughflow, energy degrada
pathways to higher numbers. However, instead of linking tion, exergy storage, and specific entropy. These orientors
provide good system level indicators of development
detritus as a source compartment in the ecosystem con
during succession or restoration of impaired ecosystems.
ceptual model, the standard paradigm is to envision two
parallel food webs one with primary producers as the base
and the other with detritus as the base without any input
from the rest of the web. If detritus were properly linked Biogeochemical Cycles
as both a source and sink in the ecosystem, then it would
be clear that higher order trophic levels are possible, if Another major focus of ecosystem ecology is understanding
not common. The higher observed trophic levels how the chemical elements necessary for life persist and
observed in some studies are not in conflict with the translocate in pools and fluxes within the ecosphere. The
laws of thermodynamics, but they show that ecosystems biosphere actively interacts with the three abiotic spheres
are more thorough at utilizing the energy within the (hydrosphere, atmosphere, and lithosphere) to provide the
system, mostly by decomposers, before it is lost as available concentration of each for life. This action has a
degraded, unavailable energy. significant impact on the relative distribution of these ele
Energy resources flowing through the ecosystem are ments. The simple sugar products of photosynthesis,
necessary to maintain all growth and development activ C6H12O6, are the base for organic matter, so carbon, hydro
ities. Organisms follow a clear life history pattern, and gen, and oxygen dominate the composition of life, and
while the timescales differ depending on the species, early while oxygen is available in the lithosphere, and hydrogen
stage energy availability is generally used for growth, in the hydrosphere, carbon is actually quite scarce in the
while later energy surplus is used for maintenance or environment, making the disproportionate amount of car
reproduction. A similar pattern is visible in ecosystem bon in biomass a hallmark of life. In fact, there are about 20
level growth and development. Net primary production is elements used regularly in living organisms, of which nine
used to build biomass and physical structure of the eco called the macronutrients are the major constituents of
system. The additional structure of photosynthetic organic matter: hydrogen, oxygen, carbon, nitrogen,
10 Ecosystem Ecology

Table 4 Percentage atomic composition of the biosphere, hydrosphere, atmosphere, and lithosphere for first 10 elements

Biosphere Hydrosphere Atmosphere Lithosphere

H 49.8 H 65.4 N 78.3 O 62.5

O 24.9 O 33.0 O 21.0 Si 21.22
C 24.9 Cl 0.33 Ar 0.93 Al 6.47
N 0.073 Na 0.28 C 0.03 H 2.92
Ca 0.046 Mg 0.03 Ne 0.002 Na 2.64
K 0.033 S 0.02 Ca 1.94
Si 0.031 Ca 0.006 Fe 1.92
Mg 0.030 K 0.006 Mg 1.84
P 0.017 C 0.002 K 1.42

calcium, potassium, silicon, magnesium, and phosphorus. New Mexico, to the Baltimore Urban Ecosystem Study.
Some of these elements are readily available in the abiotic These projects rely on a vast team of scientists to study
environment, in which case conservation through cycling the many interactions at this spatial scale. Still, the diffi
of the elements is not paramount; however, those in scarce culty lies in putting together all the pieces into an
supply, such as nitrogen and phosphorus (Table 4), are integrated whole picture of the ecosystem.
reused many times before being released from the system. Smaller scale, individual led ecological research is
These biogeochemical cycles provide the foundation to commonly conducted using microcosm and mesocosm
understand how human modification leads to eutrophica experiments. A mesocosm experiment uses designed
tion (N and P cycles) and global climate change (C cycle). equipment or enclosures in which environmental factors
Therefore, much effort has been made to study and under can be controlled and manipulated to approximate
stand these cycles, particularly the carbon, nitrogen, and natural conditions. The prevalence of this approach
phosphorus cycles, details of which are addressed else created a wealth of small scale experimentation but at
where in this encyclopedia. the expense of larger observational studies, which sparked
a fierce debate in the 1990s between the field versus
bottle approach. Indeed, the usefulness of microcosm
Ecosystem Studies experiments for ecosystem ecology was brought into
question, but the resolution has been that a multiplicity
The ecosystem perspective achieved footing in the eco of approaches is useful to address ecological questions.
logical academic community since it was central to E. P.
Odums seminal textbook Fundamentals of Ecology first pub
lished in 1953. An early implementation of this approach Human Influence on Ecosystems
at the institutional scale was attempted was in the
International Biological Program (IBP), which was run Humans have greatly altered and impacted the global
from 1964 to 1974. The program had many successes in biosphere. We recognize now the importance of main
assessing and surveying the Earths ecosystems, but faced taining functioning ecosystem services both out of our
the difficulty of compelling a top down, holistic research own necessity and for the obligation we have to the eco
paradigm on individual scientific endeavors. As a result of sphere. In 2000, the United Nations Secretary General
this conflict, the program did not deliver as much as had called for a global ecological assessment, which was
been hoped, but set the stage for the next generation of recently published as the Millennium Ecosystem
ecosystem scale research. One feature of the IBP that did Assessment (MEA) ( The report com
continue was the use of computer simulation modeling as piled by over 1350 experts from 95 countries found that
a tool to understand the complex ecological interrelations. humans have changed ecosystems more rapidly and
The journal Ecological Modelling and Systems Ecology started extensively over the last 50 years than in any comparable
in 1975 continues as an active repository for mathematical period of time in human history, resulting in a substantial
and computer based ecosystem research. and largely irreversible loss in the diversity of life on
Subsequent to the IBP, the US National Science Earth (other highlights from the report are presented
Foundation officially established the Long Term in Table 5). The MEA operated within a framework
Ecological Research Sites (LTER) in 1980 but research that identified four primary ecosystem services needed
at several of the sites dates much earlier. Currently, there by humans: supporting (nutrient cycling, primary produc
are 26 such sites ranging from the Coweeta Hydrological tion, soil formation, etc.), provisioning (food, water,
Lab in North Carolina, Hubbard Brook Ecosystem Study timber, fuel, etc.), regulating (climate, flood, disease,
in New Hampshire, Sevilleta National Wildlife Refuge in etc.), and cultural (esthetic, spiritual, educational,
Ecosystem Ecology 11

Table 5 A few of the trends identified in the Millennium Ecosystem Assessment

50% of all the synthetic nitrogen fertilizer ever used has been used 20% of the worlds coral reefs were lost and 20% degraded in
since 1985 the last several decades
60% of the increase in the atmospheric concentration of CO2 since 35% of mangrove area has been lost in the last
1750 has taken place since 1959 several decades
Approximately 60% of the ecosystem services evaluated are being Withdrawals from rivers and lakes doubled since 1960
degraded or used unsustainably

Table 6 Ecosystem Approach principles of the Convention on Biological Diversity

1 The objectives of land, water, and living resource management are a matter of societal choices
2 Management should be decentralized to the lowest appropriate level
3 Ecosystem managers should consider the effects (actual or potential) of their activities on adjacent and other ecosystems
4 Recognizing potential gains from management, there is usually a need to understand and manage the ecosystem in an economic
context. Any such ecosystem-management program should
(a) reduce those market distortions that adversely affect biological diversity;
(b) align incentives to promote biodiversity conservation and sustainable use; and
(c) internalize costs and benefits in the given ecosystem to the extent feasible
5 Conservation of ecosystem structure and functioning, in order to maintain ecosystem services, should be a priority target of the
ecosystem approach
6 Ecosystem must be managed within the limits of their functioning
7 The ecosystem approach should be undertaken at the appropriate spatial and temporal scales
8 Recognizing the varying temporal scales and lag-effects that characterize ecosystem processes, objectives for ecosystem
management should be set for the long term
9 Management must recognize the change is inevitable
10 The ecosystem approach should seek the appropriate balance between, and integration of, conservation and use of biological
11 The ecosystem approach should consider all forms of relevant information, including scientific and indigenous and local
knowledge, innovations, and practices.
12 The ecosystem approach should involve all relevant sectors of society and scientific disciplines

The 12 principles mentioned above are complementary and interlinked.

recreational, etc.). All have shown signs of stress and organisms, their environment, and human activity to pro
human pressures during the past century. One positive mote conservation, sustainability, and equity for
trend was the increase in food production (crops, live managing natural resources. The approach deals with
stock, and aquaculture), but this occurred with a the complex socioecologicaleconomic systems by pro
concomitant loss of wild fisheries and food capture, moting integrated assessment and adaptive management.
along with a substantial increase in the resource inputs The ecosystem approach of the CBD is outlined below in
required to maintain the high agricultural production. 12 principles (Table 6). Note particularly principles 58
While these observed changes to ecosystems have con that deal with ecosystem functioning, and taken in the
tributed to substantial net gain in human well being and context of the other principles assert how this ecological
economic development, they have come at an increasing functioning provides opportunities and constraints for
cost to the ecosystem health. The loss of this natural economic and social well being. Research in ecosystem
capital is typically not properly reflected in economic ecology today is directed toward improved understanding
accounts. of key issues such as ecosystem services, resilience, spatial
Since the ecosystem provides the necessary functions and functional scale, time lags, dynamics, and indirect
for life, environmental management principles being effects.
devised and implemented today use the ecosystem con
cept as foundation. In particular, there have been several
high profile international efforts such as with the Summary
Convention on Biological Diversity (CBD), a treaty
initiated in 1992 and signed by 150 government leaders Ecosystem ecology deals with the functioning at the sys
with the expressed aim to protect and promote biological tem level of the ecological community with its abiotic
diversity and sustainable development. The ecosystem environment, primarily in terms of the energy flow and
approach adopted within this convention uses scientific nutrient cycling. Research in ecosystem ecology has given
methodologies regarding ecological interactions among us a much better understanding of the processes and
12 Ecological Systems Thinking

functions necessary to sustain life. The work in natural Likens GE, Borman FH, Johnson NM, Fisher DW, and Pierce RS (1970)
Effects of forest cutting and herbicide treatment on nutrient budgets
sciences has outpaced the ability of the social institutions in the Hubbard Brook watershed ecosystem. Ecological
to adapt and implement this knowledge. However, there Monographs 20: 23 47.
is reason to be optimistic because the recent focus on the Lindeman RL (1942) The trophic dynamic aspect of ecology. Ecology
23: 399 418.
ecosystem approach in major international efforts recog Odum EP (1969) The strategy of ecosystem development. Science
nizes that humans, with their cultural diversity, are an 164: 262 270.
integral component of ecosystems. Odum HT (1957) Trophic structure and productivity of Silver Springs,
Florida. Ecological Monographs 27: 55 112.
Patten BC (1978) Systems approach to the concept of environment.
Ohio Journal of Science 78: 206 222.
See also: Ecological Network Analysis, Environ Analysis; Tansley AG (1935) The use and abuse of vegetational concepts and
Ecosystem Services; Ecosystems; Goal Functions and terms. Ecology 16: 284 307.
Weigert RG and Owen DF (1971) Trophic structure, available resources
Orientors. and population density in terrestrial versus aquatic ecosystems.
Journal of Theoretical Biology 30: 69 81.
Wiegleb G (2000) Lecture Notes on The History of Ecology and Nature
Conservation. http://board.erm.tu
Further Reading no cache1&file33&uid14 (accessed May 2007).
Chapin III FS, Matson PA, and Mooney HA (2002) Principles of
Terrestrial Ecosystem Ecology. New York: Springer.
Fath BD, Jrgensen SE, Patten BC, and Straskraba M (2004)
Ecosystem growth and development. Biosystems 77: 213 228. Relevant Website
Golley FB (1993) A History of the Ecosystem Concept in Ecology. New
Haven: Yale University Press. Millennium Ecosystem Assessment

Ecological Systems Thinking

D W Orr, Oberlin College, Oberlin, OH, USA
2008 Elsevier B.V. All rights reserved.

Introduction Summary
Applied Systems Thinking Further Reading
Environmental Education

Introduction sometimes morph into other forms and processes. In Earth

systems, small changes can have large effects somewhere
The greatest discovery of the past century had nothing to else and at some later time. Natural systems and the world
do with nuclear physics, or computer science, or genetic made by humans are intertwined in more ways than we can
engineering. Rather it was the discovery of the essential possibly imagine. The result is less like a machine than it is
connectedness of life and environment. The primary dis like a web stretching across all life forms and back through
cipline of interrelatedness is ecology beginning with the time. The effects of human actions millennia ago still ripple
work of Ernst Haeckel in the nineteenth century. The forward, intersect with other changes sometimes amplify
discovery of evolution extended the awareness of our con ing, sometimes diminishing in intensity. Some human
nections to life in time and more extensively to the story of wrought changes, such as deforestation and saline soils
life on Earth. Fields such as ecology, general systems throughout much of the Middle East, are permanent as
theory, systems dynamics, operations research, and chaos we measure time.
theory added details and theoretical depth, but with each Nothing in the preceding paragraph is particularly
advance in the precision and extent of knowledge the new or controversial. But the idea of interrelatedness has
larger story remained the same. Living systems are linked yet to take hold of us in a deep way. We still live in thrall
in food webs and ecological processes into larger systems to a world created by Descartes, Bacon, Galileo, and their
whether called the noosphere, biosphere, ecosphere, or heirs who taught us to dissect, divide, parse, and analyze
Gaia. The boundaries between life forms and between by reduction but not how to put things back together or
what we take to be living and nonliving things shift and see the world as systems and patterns. The results were
Ecological Systems Thinking 13

intellectual power without perspective so that, in time, the removal of mangroves and coastal forests that would
overspecialization became a kind of a cultural disease. otherwise have absorbed much of its energy and dam
There are many reasons why things do not change long pened the destructive effects. That, too, was known in
after their deficiencies are apparent: the inertia of habit, many circles but did not have much effect on the policies
economic inconvenience, the preservation of reputation, that prevailed along the Gulf Coast, where oil extraction,
and intellectual laziness. But the most important barrier to commerce, and gambling ruled the day.
change remains simply that science and the technology it Public attitudes toward science are often undermined by
spawned works and is a powerful presence in our daily poor education, inadequate public funding, and, sometimes,
lives. Automobiles, airplanes, the cornucopia evident in religious dogma. In the USA, evolution, once thought to be
every supermarket, miracle cures, and the wonders of an established part of science, is hotly contested as just
computers and communications are a constant reminder another theory by advocates of intelligent design. The
of the powers of a particular kind of science and a promise scientific evidence about human driven climate change is
of things to come. That much of our technology also bites indisputable, but ignored or underestimated even when
back and incurs costs that we do not see is mostly lost on alternatives are economically advantageous. The results
us. Many live in what has been called a consensus trance, are evident in the considerable data describing ecological
believing that things will go well for us, which is to say deterioration virtually everywhere and the failure to seize
that progress will continue indefinitely. Beneath such better alternatives as well. Law based on ecological knowl
ideas is the faith that nature does not set traps for unwary edge and the hope that we might calibrate our public
species, as biologist Robert Sinsheimer once put it or that business with the way the world works as a physical system
progress itself is not a self made trap. is under constant assault. Evidence about the health and
There have always been skeptics, however. Toward the ecological effects of toxins is downplayed. Public access to
end of his life, H. G. Wells could see no grounds for hope. information about the release of toxics is restricted. The
More recently, Joseph Tainter, Martin Rees, and Jared result is a significant gap between what is known about how
Diamond have expressed doubts about our longevity the world works as a physical system and the public policy
based in no small part on their views of scientific progress. in every country. The cumulative result is that we are much
Rees, for example, believes that our odds of making it to more vulnerable to ecological ruin and extreme events than
the year 2100 are no better than fifty fifty. Diamond has we might otherwise be.
cataloged the reasons why past societies have collapsed and What can be done with ecological knowledge? One
they bear more than a passing resemblance to our present answer is that ecology as a science ought to do what it
behavior. James Lovelock, coauthor of the Gaia hypothesis, has been doing, which is to say document the deterioration
believes that we are approaching a climate tipping point of ecosystems in ever finer detail. Ecology, the argument
somewhere between 400 and 500 ppm CO2 in the atmo goes, is a science and its practitioners ought to maintain
sphere after which nothing the nations of the world do will their credibility as scientists and not assume the role of
alter the outcome and the Earth will more irreversibly to a advocates and risk losing their credibility even when they
new hot state. In various ways, each of these attributes our recognize folly disguised as public policy. If that is the
vulnerability to the failure to see systems, patterns, and to future of the discipline it will, I think, flourish for a time
exercise foresight. As a result, we stumble toward a time of while the human prospect withers.
severe climate destabilization, biotic impoverishment, and There is, however, another perspective on the uses of
ecological surprises. ecology. Paul Sears in 1964 and later Paul Shepard and
The failure of ecological knowledge to penetrate very Daniel McKinley in 1969 once called the discipline the
deeply into the larger society and its decision making subversive science. They proposed ecology as an integra
systems ought to be a matter of grave concern. The tive discipline, a kind of vision across boundaries and a
early work of ecologists Howard and Eugene Odum on resistance movement an alternative to being man
the productivity of salt marshes, for example, may have fanatic. Ecology in their view offers an essential factor
slowed but certainly did not stop the juggernaut of devel . . . to all our engineering and social planning. In their
opment that has severely damaged coastal ecosystems perspective, the world needs to know what ecologists
virtually everywhere. Similarly, we know a great deal know and needs to take that knowledge seriously enough
about the services of natural systems and the impossibility to transform the ways by which we provision ourselves
of duplicating these by human means. Yet the drawdown with food, energy, materials, shelter, and livelihood.
of natural capital and the destruction of ecosystems are Ecology as a subversive science would be integrated with
still trumped by narrow short term concerns of profit and building, industry, agriculture, landscape management,
economic expansion. Sometimes the costs of ecological economics, and governance. In short, the idea of interre
folly become starkly apparent as they did following hur latedness would move from the pages of obscure scientific
ricane Katrina in the fall of 2005 in which the damage journals out to the main street, and into board rooms,
done by a class III hurricane (at landfall) was amplified by editorial offices, courtrooms, legislatures, and classrooms.
14 Ecological Systems Thinking

It would progress from being just one more interesting but have confirmed Jacksons hypothesis that the two can
obsolete idea to become the design principles for a better be stitched together, thereby eliminating a great deal of
world the default setting for everyday behavior. fossil energy and soil erosion.
Materials science is a fourth area in which ecology is
being taken seriously. Nature, as chemist Terry Collins
Applied Systems Thinking has noted, uses only a relatively few ingredients while
industrial chemistry uses virtually the entire periodic
In this regard, the news is guardedly optimistic. The art table, creating ecological havoc. The field of biomimicry
and science of high performance building is growing. The has grown in response by studying how nature works in
result is a new generation of buildings that require a fine detail. Natural systems are a carnival of color, for
fraction of the energy of conventional buildings, use mate instance, but nature does not use paints. To answer such
rials screened for environmental effects, minimize water questions, Janine Benyus, author of Biomimicry, is devel
consumption, and are landscaped to promote biological oping a database of the ways nature works to filter,
diversity, moderate microclimates, and grow foods. The reduce, recycle, color, purify, form, and join all done
best of these are highly efficient, powered substantially by without the use of toxics and fossil fuels and all of it
sunlight and feature daylight, water recycling, and interior biodegradable. The result could be a transformation of
green spaces. They are a finer calibration between our five materials and industry that dramatically reduce pollution
senses and the built environment and tend to promote and energy use.
higher user satisfaction and productivity. The costs of In these examples and elsewhere, the science of applied
building green, as it turns out, are not necessarily higher ecology has begun to seriously influence decisions and
than conventional buildings while having lower operating behavior and the evolution of architecture, engineering,
costs. The goal is to design buildings as whole systems, not materials science, agronomy, urban planning, and econom
as disjointed components. The green building movement ics. The driving force is partly economic (to reduce the
is now a worldwide movement and is transforming the costs of unnecessary energy, materials and water use) and
practice of architecture, landscape architecture, and engi partly a matter of conviction (that it is wrong to leave a
neering. It could, in time, transform the design of legacy of ruin behind us). While promising, such measures
communities and cities as well. are necessary but insufficient. Ecological thinking, in one
Business, too, is beginning to go green. The best exam way or another, must become a more central part of global
ple of a well run environmentally sensitive business is society and this is the task of education.
that of Interface, Inc., a global manufacturer of carpet
tiles and raised flooring. In the mid 1990s, company
founder and CEO, Ray Anderson, decided to transform Environmental Education
the company to eliminate waste and carbon emissions.
Interface launched a pioneering effort to develop carpet The idea of specifically environmental education entered
products that were returned to the company as a product the public discourse in the late 1960s. Among the recom
of service not otherwise discarded in a landfill. Interface mendations of the Stockholm Conference in 1972 was to
now leases carpet to its customers and takes it back to be establish an international programme in environmental
remade into new products, thereby eliminating much of education. UNESCO and UNEP subsequently under
the petrochemical sources at one end and waste at the took to prepare curricular materials, establish priorities,
other. In the past decade, the company has eliminated develop pilot projects, and organize meetings. The result
56% of its carbon emissions and is on track to becoming was a UN sponsored Conference at Tbilisi, Georgia, in
carbon neutral. The model for the company is consciously 1978 that produced a consensus statement including the
that of ecology all the way down to carpet products that words:
mimic a forest floor. Interface is not alone. Other compa
nies like Wal Mart and DuPont are beginning to Environmental education . . . should constitute a compre
transform themselves as well. Some day, perhaps, all hensive lifelong education . . . it should prepare the
business will be powered by sunlight with materials cycles individual for life through an understanding of the
that mimic the circular flow of nutrients in ecosystems. major problems of the contemporary world, and the pro
In agriculture, Wes Jackson, co founder of the Land vision of skills and attributes needed to play a productive
Institute, is pioneering the development of natural role towards improving life and protecting the environ
systems agriculture. The goal is to model agriculture ment with due regard given to ethical values. By adopting
on ecological systems such as forests and prairies. If a holistic approach, rooted in a broad interdisciplinary
successful, the end product will be agricultural polycul base, it recreates an overall perspective which acknowl
tures of high yield perennials, long thought to be a edges the fact that natural environment and manmade
biological impossibility. The early results, however, environment are profoundly interdependent . . ..
Ecological Systems Thinking 15

The Tbilisi Conference produced 41 recommendations questions are important differences about what it means
spanning the needs for environmental education between to be human, what part of that definition ought to remain
developed and less developed countries. In the subsequent inviolable, and about the manipulation of natural systems
decades, initiatives, including those spawned by Agenda 21 through technological means such as genetic engineering.
and discussions about the Earth Charter, have advanced Is the problem, in other words, one in education or one of
the discussion of environmental education into a major part education?
of the dialog about the role of education relative to the What can be said with certainty is that public school
human prospect. There is no serious discussion about the ing and higher education have been underachievers in the
transition to sustainability launched by the Brundtland task of inculcating essential knowledge about the envir
Report in 1987 that does not include changing the goals onment. Public opinion surveys show high levels of
and methods of education. From Tbilisi (1978), Talloires support for environmental quality but little ecological
(1990), and subsequent international gatherings, a strong knowledge. In the words of one typical survey, people
consensus about the importance of environment in higher have acquired a substantial familiarity with environmen
education is clearly apparent. tal issues, but [have] a long way to go in developing a
Despite considerable progress, both conceptually and working environmental/energy knowledge. Much of
practically, there are serious differences about the goals what people know about the environment is derived
and methods of environmental education that reflect and, from television in bits and pieces and not through direct
in some ways, amplify larger disagreements about educa experience with nature or through cultural transmission.
tion. At the lowest level, there is a general consensus that One particularly encouraging aspect is the develop
the young ought to know something about how nature ment of environmental education in institutions of
works as a physical system the rudiments of biology and higher education. Stemming from innovations in the
planetary science. There is considerably less agreement 1980s, a vibrant campus ecology movement has emerged
about how this should be incorporated into the standard in Europe, Australasia, and the USA, along with a wide
curriculum or at what level. Most elementary schools discussion of sustainability of educational institutions.
include curricular components such as Project Learning Beginning with the studies of college food, energy use,
Tree or Wet and Wild that introduce children to what and pollution, the movement has grown in subsequent
was once called natural history along with some field decades to a worldwide scale. Hundreds of colleges and
experience and practical outdoor skills. But the later universities globally have organized efforts to system
inclusion of values or discussion about the causes of atically reduce energy use, water consumption, and
environmental ills has often been controversial, especially material flows. Campus sustainability and climate stabi
when it has led to questions about conventional economic lity have come to the center of institutional planning,
or political wisdom. purchasing, and construction. Beginning in the late
In important respects, all education is environmental 1990s with the advent of means to promote and measure
education, that is, by what is included or excluded stu environmental performance of buildings, the construc
dents are taught that they are part of or apart from tion of academic facilities is undergoing a rapid
ecological systems. The standard, discipline centric cur revolution. Green or high performance building stan
riculum may have contributed to a mindset that helped to dards are increasingly regarded as necessary to reduce
create environmental problems by separating subjects energy and maintenance costs as well as laboratories for
into boxes and conceptually by separating people from research and education. Many of the problems of sus
nature. As a result, graduates are often ignorant of ecolo tainability ecological design, applications of solar
gical relationships or why they are worthy of energy, water purification, food production, ecological
consideration. Not surprisingly, the first response to pro restoration, and landscape management can be studied
posals for environmental education attempted to in buildings and adjacent landscapes at a scale that is
accommodate environmental issues and ecology into for both significant yet manageable. Given recent develop
mal education as a kind of add on. More radical critics ments on many campuses, it is not inconceivable that
proposed that formal education ought to be reformed educational institutions at all levels will one day become
along ecological lines, raising another and no less con models of ecological design mirroring the larger solu
tentious issues. From either perspective, environmental tions necessary to the transition to sustainability.
mismanagement and the larger discussion of sustainability
raise questions about the meaning of human mastery over
nature, or more accurately as C. S. Lewis once put it: what Summary
does it mean for some men to control other men through
the mastery of some parts of nature? What is the core In the decades since the Stockholm Conference in
knowledge of the environment that ought to be standard 1972, environmental education has emerged as a sig
in an educational curriculum? At the heart of such nificant component of education virtually everywhere
16 Ecosystems

in the world. It has, for the most part, flourished at all Lovelock J (2006) The Revenge of Gaia. London: Penguin Books.
Lovelock J The Gaia Hypothesis. New York: Oxford University Press.
levels of education. There are magazines and journals Lovins A (2005) Winning the Oil Endgame. Snowmass, CO: Rocky
such as Sustainability in Higher Education, professional Mountain Institute.
associations, and regular conferences. It is not difficult Oakeshott M (1989) The Voice of Liberal Learning. New Haven, CT: Yale
University Press.
to imagine all of this as the start of something like an Orr D (1992) Ecological Literacy. Albany, NY: Suny Press.
ecological enlightenment emerging in the decades or Orr D (1994) Earth in Mind. Washington, DC: Island Press.
centuries ahead. But no such thing is certain. If edu Orr D (2006) Design on the Edge. Cambridge, MA: MIT Press.
OSullivan E (2005) Millennium Ecosystem Assessment Report,
cation is to be midwife to a deeper, broader, and vols. 1 5. Washington, DC: Island Press.
sustainable transformation, it will have to surmount Rees M (2003) Our Final Hour. New York: Basic Books.
Sears P (1964) Ecology A subversive subject. BioScience
serious challenges.
14(7): 11 13.
Shepard P and McKinley D (eds.) (1969) The Subversive Science.
Boston: Houghton Mifflin.
Sinsheimer R (1978) The Presumptions of Science. Daedalus
Further Reading 107: 23 36.
Sobel D (1996) Beyond Ecophobia. Great Barrington, MA: The Orion
Barlett P and Chase G (eds.) (2004) Sustainability on Campus. Society.
Cambridge, MA: MIT Press. Steffen W, Sanderson A, Jager J, et al. (2004) Global Change and the
Benyus J (1998) Biomimicry. New York: William Morrow. Earth System. Berlin: Springer.
Bowers C (1993) Education, Cultural Myths, and the Ecological Crisis. Tenner E (1996) Why Things Bite Back: Technology and the Revenge of
Albany, NY: SUNY Press. Unintended Consequences. New York: Knopf.
Bowers C (1995) Educating for an Ecologically Sustainable Culture. Union of Concerned Scientists (1992) World Scientists Warning to
Albany, NY: SUNY Press. Humankind. Boston: Union of Concerned Scientists.
Corcoran P and Wals A (eds.) (2004) Higher Education and the Challenge US Department of Health, Education, and Welfare (1978) Toward an
of Sustainability. Dordrecht, The Netherlands: Kluwer Academic. Action Plan: A Report on the Tbilisi Conference on Environmental
Coyle K (2005) Environmental Literacy in America. Washington, DC: The Education. Washington, DC: US Government Printing Office.
National Environmental Education & Training Foundation. Vernadsky V (1998) The Biosphere. New York: Springer.
Creighton S (1998) Greening the Ivory Tower. Cambridge, MA: MIT Washburn J (2005) University INC: The Corporate Corruption of Higher
Press. Education. New York: Basic Books.
de Chardin T (1965) The Phenomenon of Man. New York: Harper Wright R (2005) A Short History of Progress. New York: Carroll &
Torchbooks. Graf.
Fischetti M (2001) Drowning New Orleans. Scientific American Wright T (2004) Evolution of sustainability declarations in higher
(October, 2001): 76 85. education. In: Corcoran PB and Wals AEJ (eds.) Higher Education
Kuhn T (1963) The Structure of Scientific Revolutions. Chicago: and the Challenge of Sustainability, pp. 7 19. Dordrecht, The
University of Chicago Press. Netherlands: Kluwer Academic.

A K Salomon, University of California, Santa Barbara, Santa Barbara, CA, USA
2008 Elsevier B.V. All rights reserved.

What Is an Ecosystem? Ecosystem Perspectives in Conservation Science

Studying Ecosystem Dynamics Further Reading
Ecosystem Function and Biodiversity

What Is an Ecosystem? Typically, boundaries between ecosystems are diffuse. An

ecotone is a transition zone between two distinct ecosys
Coined by A. G. Tansley in 1935, the term ecosystem tems (i.e., the tundraboreal forest ecotone).
refers to an integrated system composed of a biotic com
munity, its abiotic environment, and their dynamic
interactions. A diversity of ecosystems exist through the History
world, from tropical mangroves to temperate alpine lakes, Over 70 years ago, Sir Arthur Tansley (Figure 2) pre
each with a unique set of components and dynamics sented the notion that ecologists needed to consider the
(Figure 1). Ecosystems can be classified according to whole system, including both organisms and physical
their components and physical context yet their classifi factors, and that these components could not be separated
cation is highly dependent on the spatial scale of scrutiny. or viewed in isolation. By suggesting that ecosystems are
Ecosystems 17

(a) (b) (c)

(d) (e) (f)

Figure 1 (a) Kelp forest, (b) subarctic alpine tundra, (c) tropical coastal sand dune, (d) tropical mangrove, (e) alpine lake, and
(f) temperate coastal rain forest. Photos by Anne Salomon, Tim Storr, and Tim Langlois.

system of biotic and abiotic components. He considered

how the lake food web and processes driving nutrient
flux affected the rate of succession of the whole lake
ecosystem, a significant departure from traditional inter
pretations of succession.
By the late 1950s and early 1960s, system wide
energy fluxes were quantified in various ecosystems by
E. P. Odum and J. M. Teal. In the late 1960s, Likens,
Bormann, and others took an ecosystem approach to
studying biogeochemical cycles by manipulating whole
watersheds in the Hubbard Brook Experimental Forest to
determine whether logging, burning, or pesticide and
herbicide use had an appreciable effect on nutrient loss
from the ecosystem. This research set an important
precedent in demonstrating the value of conducting
experiments at the scale of an entire ecosystem (see the
section entitled Whole ecosystem experiments), a sig
nificant advancement which continues to inform
ecosystem studies today.

Figure 2 Sir Arthur G. Tansley coined the term ecosystem in

1935. From New Phytologist 55: 145, 1956. Ecosystem Components and Properties
Ecosystems can be thought of as energy transformers
dynamic, interacting systems, Tansleys ecosystem con and nutrient processors composed of organisms within a
cept transformed modern ecology. It led directly to food web that require continual input of energy to balance
considerations of energy flux through ecosystems and that lost during metabolism, growth, and reproduction.
the pathbreaking, now classic work of R. L. Lindeman in These organisms are either primary producers (auto
1942, one of the first formal investigations into the func trophs), which derive their energy by using sunlight to
tioning of an ecosystem, in this case a senescent lake, convert inorganic carbon into organic carbon, or second
Cedar Creek Bog, in Minnesota. Inspired by the work of ary producers (heterotophs), which use organic carbon as
C. Elton, Lindeman focused on the trophic (i.e., feeding) their energy source. Organisms that perform similar types
relationships within the lake, grouping together organ of ecosystem functions can be broadly categorized by
isms of the lake according to their position in the their functional group. For example, herbivores are
food web. To study the cycling of nutrients and the heterotophs that eat autotrophs, carnivores are hetero
efficiency of energy transfer among trophic levels over trophs that eat other heterotrophs, while detritivores
time, Lindeman considered the lake as an integrated are heterotrophs that eat nonliving organic material
18 Ecosystems

(detritus) derived from either autotrophs or heterotrophs Solar energy is transformed into chemical energy by
(Figure 3). Herbivores, carnivores, and dertitivores are primary producers via photosynthesis, the process of con
collectively known as consumers. verting inorganic carbon (CO2) from the air into organic
Classifying organisms according to their feeding rela carbon (C6H12O2) in the form of carbohydrates. Gross
tionships is the basis of defining an organisms trophic primary production is the energy or carbon fixed via
level; the first trophic level includes autotrophs; the photosynthesis over a specific period of time, while net
second trophic level includes herbivores and so on. primary production is the energy or carbon fixed in
Ecosystem components that make up a trophic level are photosynthesis, minus energy or carbon which is lost via
quantified in terms of biomass (the weight or standing respiration, per unit time. Production by secondary pro
crop of organisms), while ecosystem dynamics, the flow of ducers is simply the amount of energy or material formed
energy and materials among system components, are per unit term.
quantified in terms of rates. A careful distinction needs to be made between
Typically, ecologists quantifying ecosystem dynamics production rates and static estimates of standing crop
use carbon as their currency to describe material flow and biomass, particularly because the two need not be
energy to quantify energy flux. Material flow and energy related. For example, two populations at equilibrium,
flow differ in one important property, namely their ability in which input equals output, might have the same
to be recycled. Chemical materials within an ecosystem standing stock biomass but drastically different pro
are recycled through an ecosystems component. In con duction rates because turnover rates can vary
trast, energy moves through an ecosystem only once and (Figure 4). For example, on surf swept shores from
is not recycled (Figure 3). Most energy is transformed to Alaska to California, two species of macroalgal pri
heat and ultimately lost from the system. Consequently, mary producers grow in the low rocky intertidal zone
the continual input of new solar energy is what keeps an of temperate coastal ecosystems (Figure 5). The rib
ecosystem operational. bon kelp, Alaria marginata, is an annual alga with high
growth rates, whereas sea cabbage, Hedophyllum sessile,
Heat Heat is a perennial alga with comparatively lower growth
rates. Although they differ greatly in their production
4 rates, in mid July, during the peak of the growing
Detritivores Detritus season, these two species can have almost equivalent
stand crop biomasses.

3 Carnivores
Trophic level

Heat (a) (b)

Herbivores High Low

2 production production

Standing Standing
Heat crop crop
biomass biomass
1 Primary producers (high turnover) (low turnover)

High input Low input

Figure 4 Standing crop biomass is not always correlated to
production rates. Here, two hypothetical species with
populations at equilibrium, where input equals output, have an
equivalent standing crop biomass but differ in their turnover
rates. Population (a) has high input, high production, and high
Figure 3 Energy flows and material cycles in an ecosystem. turnover rates, whereas population (b) has low input, low
Materials move through the trophic levels and eventually cycle production, and low turnover rates. In reality, populations are
back to the primary producers via the decomposition of detritus rarely at equilibrium so standing crop biomass fluctuates
by microorganisms. Energy, originating as solar energy, is depending on input rates and the amount of production
transferred through the trophic levels via chemical energy and is consumed by higher trophic levels. Adapted from Krebs C (2001)
lost via the radiation of heat at each step. Adapted from Ecology: The Experimental Analysis of Distribution and
DeAngelis DL (1992) Dynamics of Nutrient Cycling and Food Abundance, 5th edn. San Francisco: Addison-Wesley
Webs. New York, NY: Chapman and Hall. Educational Publishers, Inc.
Ecosystems 19

(a) (b)


Figure 5 (a) In the low intertidal zone of temperate coastal ecosystems, (b) the ribbon kelp, Alaria marginata, is an annual alga with
high growth rates, whereas (c) the sea cabbage kelp, Hedophyllum sessile, is a perennial alga with lower growth rates. During the peak
of the growing season, these two species can have a similar stand crop biomass but differ greatly in their production rates because one
is an annual and the other is a perennial. Photo by Anne Salomon and Mandy Lindeberg.

Ecosystem Efficiency photoplankton allows researchers to estimate the propor

tion of this production that is taken by the fishery.
The efficiency of energy transfer within an ecosystem can
It has been estimated that 8% of the worlds aquatic
be estimated as its trophic transfer efficiency, the fraction
primary production is required to sustain global fisheries.
of production passing from one trophic level to the next.
Considering continental shelf and upwelling areas speci
The energy not transferred is lost in respiration or to
fically, these ecosystems provide one fourth to one third
detritus. Knowing the trophic transfer efficiency of an
of the primary production required for fisheries. This
ecosystem can allow researchers to estimate the primary
high fraction leaves little margin for error in maintaining
production required to sustain a particular trophic level. resilient ecosystems and sustainable fisheries.
For example, in aquatic ecosystems, trophic transfer
efficiency can vary anywhere between 2% and 24%, and
average 10%. Assuming a trophic efficiency of 10%,
researchers can estimate how much phytoplankton produc Large-Scale Shifts in Ecosystems
tion is required to support a particular fishery. Consider the A growing body of empirical evidence suggests that
open ocean fishery for tuna, bonitos, and billfish. These are ecosystems may shift abruptly among alternative
all top predators, operating at the fourth trophic level. states. In fact, large scale shifts in ecosystems have
According to world catch statistics recorded by the Food been observed in lakes, coral reefs, woodlands, des
and Agriculture Organization, in 1990, 2 975 000 t of serts, and oceans. For example, a distinct shift
these predators were caught, equivalent to 0.1 g of carbon occurred in the Pacific Ocean ecosystem around
per m2 of open ocean per year. To support this yield of 1977 and 1989. Abrupt changes in the time series of
tuna, bonitos, and billfish, researchers can calculate the fish catches, zooplankton abundance, oyster condition,
production rates of the trophic levels below, assuming a and other marine ecosystem properties signified con
trophic efficiency of 10% and equilibrium conditions. spicuous shifts from one relatively stable condition to
Essentially, to produce of 0.1 gC m 2 yr 1 of harvested another (Figure 6). Also termed regime shifts, the
predators (tuna, bonitos, and billfish) requires implications of these abrupt transitions for fisheries
1 gC m 2 yr 1 of pelagic fish to have been consumed by and oceanic CO2 uptake are profound, yet the
the top predators, 10 gC m 2 yr 1 of zooplankton to be mechanisms driving these shifts remain poorly under
consumed by the pelagic fishes, and 100 gC m 2 yr 1 of stood. It appears that changes in oceanic circulation
phytoplankton. Note that these values represent the pro driven by weather patterns can be evoked as the
duction that is transferred up trophic levels. They do not dominant causes of this state shift. However, compe
represent the standing stock of biomass at each trophic tition and predation are becoming increasingly
level. Knowing the net primary production of the recognized as important drivers of change altering
20 Ecosystems

1977 Regime shift


Ecosystem state


1965 1970 1975 1980 1985 1990

1989 Regime shift

Ecosystem state



1975 1980 1985 1990 1995 2000
Figure 6 Distinct shifts in ecosystem states, also referred to as regime shifts, occurred in the Pacific Ocean ecosystem around 1977
and 1989. The ecosystem state index shown here was calculated based on the average of climatic and biological time series. From
Scheffer M, Carpenter S, Foley JA, Folke C, and Walker B (2001) Catastrophic shifts in ecosystems. Nature 413: 591596.

oceanic community dynamics. In fact, fisheries are isotopes provide powerful tools for estimating material
well known to affect entire food webs and the trophic flux and trophic positions.
organization of ecosystems. Therefore, one could The elements C, N, S, H, and O all have more than
imagine that the sensitivity of a single keystone spe one isotope. For example, carbon has several isotopes, two
cies to subtle environmental change could cause of which are 13C and 12C. In nature, only 1% of carbon is
major shifts in community composition. Given this 13
C. Isotopic composition is typically expressed in 
interplay between and within the biotic and abiotic values, which are parts per thousand differences from a
components of an ecosystem, resolving the causes of standard. For carbon,
regime shifts in oceanic ecosystems will likely require  13  
an understanding of the interactions between the C=12 Csample
13 C 13 C=12 C
1  103
effects of fisheries and the effects of physical climate standard

change. Consequently,  values express the ratio of heavy to light

isotope in a sample. Increases in these values denote
increases in the amount of the heavy isotope component.
Studying Ecosystem Dynamics The standard reference material for carbon is PeeDee
limestone, while the standard for nitrogen is nitrogen
Stable Isotopes gas in the atmosphere. Natural variation in stable isotopic
Important insights into ecosystem dynamics can be composition can be detected with great precision with a
revealed through the use of naturally occurring stable mass spectrometer.
isotopes. These alternate forms of elements can reveal Stable isotopes record two kinds of information. Process
both the source of material flowing through an ecosystem information is revealed by physical and chemical reactions
and its consumers trophic position. This is because dif which alter stable isotope ratios, while source information is
ferent sources of organic matter can have unique isotopic revealed by the isotopic signatures of source materials.
signatures which are altered in a consistent manner as When organisms take up carbon and nitrogen, chemical
materials are transferred throughout an ecosystem, from reactions occur which discriminate among isotopes, thereby
trophic level to trophic level. Consequently, stable altering the ratio of heavy to light isotope. This is known as
Ecosystems 21

fractionation. Although carbon fractionates very little Whole Ecosystem Experiments

(0.4, 1 SD 1), the mean trophic fractionation of
Large scale, whole ecosystem experiments have contrib
15N is 3.4 (1 SD 1), meaning that 15N increases
uted considerably to our understanding of ecosystem
on average by 3.4 with every trophic transfer. Because the
dynamics. With its beginnings in wholesale watershed
15N of a consumer is typically enriched by 3.4 relative to
experiments in the 1960s, ecosystems are now being stud
its diet, nitrogen isotopes can be used to estimate trophic
ied experimentally and analyzed as system of interacting
position. Stable isotopes can provide a continuous measure
species processing nutrients and energy within the con
of trophic position that integrates the assimilation of energy
text of changing abiotic conditions. This is particularly
or material flow through all the different trophic pathways
relevant these days given the effects of anthropogenic
leading to an organism. In contrast, 13C can be used to
climate forcing and pollution in both terrestrial and ocea
evaluate the ultimate sources of carbon for an organism
nic ecosystems.
when the isotopic signatures of the sources are different.
A classic series of whole lake nutrient addition experi
Stable isotopes can track the fate of different sources of
ments conducted in northwestern Ontario by David
carbon through an ecosystem, because a consumers iso
Schindler and his research group illustrated the role of
topic signature reflects those of the key primary producers
phosphorus in temperate lake eutrophication. To separate
it consumes. For example, in both lake and coastal marine
the effects of phosphorus and nitrate, the researchers split a
ecosystems, 13C is useful for differentiating between two
lake with a curtain and fertilized one side with carbon and
major sources of available energy, benthic (nearshore)
nitrogen and the other with phosphorus, carbon, and nitro
production from attached macroalgae, and pelagic (open
gen. Within 2 months, a highly visible algal bloom had
water) production from phytoplankton. This is because
developed in the basin in which phosphorus had been
macroalgae and macroalgal detritus (specifically kelp of
added providing experimental evidence that phosphorus
the order Laminariales) is typically more enriched in 13C
is the limiting nutrient for phytoplankton production in
(less negative 13C) relative to phytoplankton due to
freshwater lakes. Certainly, algae may show signs of nitro
boundary layer effects. Researchers have exploited this
gen or carbon limitation when phosphorus is added to a
difference to answer many important ecosystem level
lake; however, other processes often compensate for these
questions. Below are two examples.
deficiencies. For instance, CO2 is rarely limiting because
During the late 1970s and early 1980s, in the western
physical factors such as water turbulence and gas exchange
Aleutian Islands of Alaska, where sea otters had recovered
regulate its availiblity. Further, nitrogen can be fixed by
from overexploitation and suppressed their herbivorous
blue green algae. These species, which are favored when
urchin prey, productive kelp beds dominated. There,
nitrogen is in short supply, increases the availability of
transplanted filter feeders, barnacles and mussels, grew
nitrogen to algae, and the lake eventually returns to a
up to 5 times faster compared to islands devoid of kelp
state of phosphorus limitation. The practical significance
where sea otters were scarce and urchin densities high.
of these results is that lake europhication can be prevented
Stable isotope analysis revealed that the fast growing
with management policies that control phosphorus input
filter feeders were enriched in carbon suggesting that
into lake and rivers.
macroalgae was the carbon source responsible for this
magnification of secondary production.
In four Wisconsin lakes, experimental manipulations
Using Management Policies as Ecosystem
of fish communities and nutrient loading rates were con
ducted to test the interactive effects of food web structure
and nutrient availability on lake productivity and carbon It has become increasingly common to use management
exchange with the atmosphere. The presence of top pre policies as experiments and test their effects on ecosystem
dators determined whether the experimentally enriched dynamics. An excellent example of this approach is the
lakes operated as net sinks or net sources of atmospheric use of marine reserves to investigate the ecosystem level
carbon. Specifically, the removal of piscivorous fishes consequences of fishing. Essentially, well enforced mar
caused an increase in planktivorous fishes, a decrease in ine reserves constitute large scale human exclusion
large bodied zooplankton grazers, and enhanced primary experiments and provide controls by which to test the
production, thereby increasing influx rates of atmospheric ecosystem effects of reducing consumer biomass via fish
carbon into the lake. Atmospheric carbon was traced to ing at an ecologically relevant scale. Dramatic shifts in
upper trophic levels with 13C. Here, naturally occurring nearshore community structure have been documented in
stable isotopes and experimental manipulations con well established and well protected marine reserves in
ducted at the scale of whole ecosystems illustrated that both Chile and New Zealand. In northeastern New
top predators fundamentally alter biogeochemical pro Zealands two oldest marine reserves, the Leigh Marine
cesses that control a lakes ecosystem dynamics and Reserve and Tawharanui Marine Park, previously fished
interactions with the atmosphere. predators, snapper (Pagrus auratus) and rock lobster
22 Ecosystems

(Jasus edwardsii), have increased in abundance by 14 and to experiment with whole ecosystems. Under such cir
3.8 fold, respectively, compared to adjacent fished waters. cumstances, researchers have used alternative techniques
Increased predation leading to reduced survivorship and to explore ecosystem dynamics. Models in ecology have a
cryptic behavior of their herbivorous prey, the sea urchin venerable tradition for both teaching and understanding
(Evechinus chloroticus), has allowed the macroalga (Ecklonia complex processes. Ecosystem models are now being used
radiata) to increase significantly within the reserves, a to gain insight into the ecosystem level consequences of
trend that has been developing in the Leigh reserve for management policies, from fisheries to carbon emissions.
the past 25 years (Figure 7). Although this provides For more information on ecosystem models and using
evidence that fishing can indirectly reduce ecosystem management policies as experiments, see the section
productivity, the trophic dynamics described above are entitled Socialecological systems, Humans as key eco
context dependent and vary as a function of depth, wave system components.
exposure, and oceanographic circulation (Figure 8). For
example, both in the presence and absence of fishing,
urchin densities decline to nearly 0 individuals per m2 Ecosystem Function and Biodiversity
below depths greater than 10 m due to unfavorable con
ditions for recruitment, despite the presence or absence of Accelerating rates of species extinction have prompted
snapper and lobster, while at depths above 3 m, wave researchers to formally investigate the role of biodiversity
surge can preclude urchin grazing both inside and outside in providing, maintaining, and even promoting ecosys
the reserves. Furthermore, where oceanic conditions tem function. Typically, studies experimentally modify
hinder urchin recruitment, the effects of fishing on species diversity and examine how this influences
macroalgae become less clear cut. These physical con the fluxes of energy and matter that are fundamental to
straints highlight the importance of abiotic context on all ecological processes. In many cases, studies are
biotic interactions. Ultimately, one can gain a lot of infor designed to document the effects of species richness on
mation by using management policies as experiments. the efficiency by which communities produce biomass,
Although policy experiments have played an impor although the effects of species diversity on other ecosys
tant role in elucidating ecosystem dynamics, in many tem functions such as decomposition rates, nutrient
cases, it is politically intractable or logistically impossible retention, and CO2 uptake rates have also been examined.

(a) (b)

Fished Snapper Lobster

3 Snapper Lobster

Trophic level

Sea Sea
urchin + urchin


Figure 7 (a) In nearshore fished ecosystems in northeastern New Zealand, snapper and lobster densities have been reduced due to
fishing pressure resulting in high sea urchin densities, urchin barrens, and reduced kelp production. (b) In marine reserves, where
previously fished snapper and lobster have recovered, sea urchins that have not been consumed by these predators behave cryptically,
hiding in crevices. Consequently, kelp forests of Ecklonia radiata dominate. Photos by Nick Shears, Hernando Acosta, and Timothy
Ecosystems 23

(a) East Auckland (b) Spring


(c) Summer

Chlorophyll a (mg m3)

0 0.1 0.2 0.5 10 3.0

Figure 8 The effects of fishing on nearshore ecosystems are influenced locally by wave exposure and regionally by oceanographic
circulation. (a) In northeastern New Zealand, ocean circulation patterns influence nutrient delivery and thus (b) spring and (c) summer
pelagic primary production. Satellite images: SeaWiFs Project, Ocean Color Web.

Several seminal studies report a positive relationship communities, almost all of the ecosystem function meas
between biodiversity and ecosystem function. Yet, the ured can be provided by relatively few species,
generality of the results, and the mechanisms driving suggesting that many species are in fact redundant.
them, have provoked considerable debate and several Few empirical studies support type A relationships,
counterexamples exist. rather, empirical evidence points to the prevalence of
At the crux of the debate lies a question with deep type B relationships. In fact, a recent meta analysis of
historical roots: do some species exert stronger control 111 such studies conducted in multiple ecosystems on
over ecosystem processes than others? Imagine two dis numerous trophic groups found that the average effect
tinct positive relationships between biodiversity and of decreasing species richness is to decrease the biomass of
ecosystem function (Figure 9). In type A communities, the focal trophic group, leading to less complete depletion
every single species contributes to the ecosystem function of resources used by that group. Further, the most
measured, even the rare species. By contrast, in type B species rich polycultures performed no differently than
the single most productive species used in the experi
High ment. Consequently, these average effects of species
diversity on ecosystem production are best explained by
Type B
the loss of the most productive species from a diverse
community. These results could be considered consistent
Ecosystem function

with what has become known as the sampling effect.

Critics argue that a positive relationship between
species diversity and ecosystem function is a sampling
artifact rather than a result of experimentally manipu
Type A lated biodiversity per se. Such a sampling effect can arise
because communities comprising more species have a
greater chance of being dominated by the most produc
tive taxa. Yet, controversy surrounding the sampling
effect itself exists given the duality in its possible inter
Low High pretation: is this a real biological mechanism that operates
Biodiversity in nature or is it an experimental artifact of using random
Figure 9 Type A communities: every single species con- draws of species to assemble experimental communities?
tributes equally to ecosystem functioning. Type B communities: To add to the ecosystem functionbiodiversity debate is
ecosystem function is provided by only a few species. the critical issue that many of these studies focus on a
24 Ecosystems

single trophic level and neglect or dismiss multiple 3. supporting services formation and preservation of
trophic level interactions, such as herbivory and other soils, protection from ultraviolet rays, pollination
disturbances well known to alter ecosystem processes, of natural vegetation and agricultural crops, cycling
calling into question the generality of these results. of nutrients, seed dispersal, maintenance of biodiver
Despite the controversy, these studies generally rein sity, primary production; and
force the notion that certain species exert much stronger 4. cultural services spiritual, esthetic, recreational.
control over ecological processes than others. However,
Although critical to human existence, ecosystem services
identifying which species these are in advance of extinc
are often taken for granted or at best, greatly undervalued.
tion remains a challenge. Nonetheless, identifying the
This is ironic given that many ecosystem services are very
mechanisms driving ecosystem functioning is an impor
difficult and expensive to duplicate, if they can be dupli
tant conservation priority given that human well being
relies on a multitude of these functions. cated at all. Normally, ecosystem services are considered
free despite their obvious economic value. For example,
over 100 000 species of animals provide free pollination
Ecosystem Perspectives in Conservation
services, including bats, bees, flies, moths, beetles, birds,
and butterflies (Figure 10). Based on the estimate that
Ecosystem Services one third of human food comes from plants pollinated by
wild pollinators, pollination has been valued at US$46
Humans have always relied on nature for environmental
billion per year in the US alone. Globally, the worlds
assets like clean water and soil formation. Today, these
ecosystem services have been valued at US$33 trillion a
assets are receiving global attention as ecosystem ser
year, nearly twice as much as the gross national product of
vices, the conditions and processes by which natural
all of the worlds countries.
ecosystems sustain and fulfill human life. Natural ecosys
The idea of paying for ecosystem services has been
tems perform a diversity of ecosystem services on which
gaining momentum. Yet, because ecosystem services are
human civilization depends:
typically not sold in markets, they usually lack a market
1. regulating services purification of air and water, value. Given the value of natural capital, nonmarket
detoxification and decomposition of wastes, moderation valuation approaches are being developed by economists
of weather extremes, climate regulation, erosion control, and ecologists to account for ecosystem services in
flood control, mitigation of drought and floods, regula decision making processes. The notion being that eco
tion of disease carrying organisms and agricultural pests; nomic valuation gives decision makers a common
2. provisioning services provision of food, fuel, fiber, currency to assess the relative importance of ecosystem
and freshwater; processes and other forms of capital.

Figure 10 Pollination services, provided by bees, bats, butterflies, and birds to name a few, have been valued at US$46 billion per
year in the US alone. Consider the global value of this important ecosystem service. Photos by Steve Gaines, Heather Tallis.
Ecosystems 25

Yet, assigning value to ecosystem services is tricky and completion of this ambitious ecological study, there is now
some analysts object to nonmarket valuation, because it is a growing movement to make the value of ecosystem ser
a strictly anthropogenic measure and does not account for vices an integral part of current policy initiatives.
nonhuman values and needs. Yet, in democratic countries,
environmental policy outcomes are determined by the
SocialEcological Systems, Humans as Key
desires of the majority of citizens, and voting on a pre
Ecosystem Components
ferred policy alternative is ultimately an anthropogenic
activity. A second objection to nonmarket valuation is a Humans are a major force in global change and drive eco
disagreement with pricing the natural world and dissatis system dynamics, from local environments to the entire
faction with the capitalistic premise that everything is biosphere. At the same time, human societies and global
thought of in terms of commodities and money. The economies rely on ecosystem services. As such, human and
point of valuation, however, is to frame choices and natural systems can no longer be treated independently
clarify the tradeoffs between alternative outcomes (i.e., because natural and social systems are strongly linked.
draining a wetland may increase the supply of develop Accumulating evidence suggests that effective environmen
able land for housing but does so at the cost of decreased tal management and conservation strategies must take an
habitat and potential water quality degradation). Finally, integrated approach, one that considers the interactions and
a third objection to nonmarket valuation stems from the feedbacks between and within social, economic, and
uncertainty in identifying and quantifying all ecosystem ecological systems. As a result, the concept of coupled
services. Advocates argue that economic valuation need socialecological systems has become an emerging focus
not cover all values and that progress is made by captur in environmental and social science and ecosystem manage
ing values that are presently overlooked. ment. Socialecological systems are considered as evolving,
Despite the uncertainties, valuing ecosystem services integrated systems that typically behave in nonlinear ways.
can sometimes pay off. When New York City compared The concept of resilience the capacity to buffer change
the coast of an artificial water filtration plant valued at has been increasingly used as an approach for understanding
US$68 billion, plus an annual operating cost of US$300 the dynamics of socialecological systems. Two useful tools
million, the city chose to restore the natural capital of the for building resilience in socialecological systems are struc
Catskill Mountains for this watersheds inherent water tured scenario modeling and active adaptive management.
filtration services and for a fraction of the cost (US$660 Models of linked socialecological systems have
million). Ultimately, the valuation of ecosystem services, been developed to inform management conflicts over
even if flawed, may get ecosystem processes on the deci water quality, fisheries, and rangelands. These models repre
sion making table and lead to more sustainable policies in sent ecosystems coupled to socioeconomic drivers and are
light of ever expanding human populations. explored with stakeholders to probe the management deci
Ecosystem services are threatened by growth in the scale sion making processes. Alternative scenarios force
of human enterprise (population size, per capita consump participants to be absolutely explicit about their assumptions
tion rates) and a mismatch between short term needs and and biases, thereby improving communication between sta
long term societal well being. With a global population keholders and exposing the ecological consequences of
soon to number 9 billion people, ecosystem services are various management policies.
becoming so degraded, some regions in the world risk Adaptive management is an approach where manage
ecological collapse. Many human activities alter, disrupt, ment policies themselves are deliberately used as
impair, or reengineer ecosystem services such as overfish experimental treatments. As information is gained, poli
ing, deforestation, introduction of invasive species, cies are modified accordingly. This approach helps isolate
destruction of wetlands, erosion of soils, runoff of pesticides, anthropogenic effects from sources of natural variation
fertilizers, and animal wastes, pollution of land, water, and and, most importantly, considers the consequences of a
air resources. The consequences of degrading ecosystem human perturbation on the whole ecosystem. In contrast,
services on human well being were examined in the basic research on various parts of an ecosystem leads to
Millennium Ecosystem Assessment (MA) 2005, which con the challenge of assembling all the data into a practical
cluded that well over half of the worlds ecosystems services framework. Yet, biotic and abiotic ecosystem components
are being degraded or used unsustainably. The MA devel are not additive, they interact. Due to these interactions,
oped global ecological scenarios as a process to inform the dynamics of an ecosystem cannot be extrapolated
future policy options. These scenarios were based on a from the simple addition of an ecosystems components.
suite of models that were designed to forecast future change. Adaptive management examines the response of the sys
The MA based its scenario analyses on ecosystem services. tem as a whole rather than a sum of its parts. Furthermore,
Specifically, scenarios were developed to anticipate this approach involves adaptive learning and adaptive
responses of ecosystem services to alternative futures driven institutions that acknowledge uncertainties and can
by different sets of policy decisions. Following the respond to nonlinearities. In sum, structured scenario
26 Ecosystem Services

modeling and policy experimentation are tools that can long term sustainability. These practices were driven by
be used to examine the resilience of socialecological inadequate information on ecosystem dynamics, ignorance
systems to alternative management policies and conserva of the space and timescales on which ecosystem processes
tion strategies. operate, and a prevailing public perception that immediate
economic and social value outweighed the risk of alterna
tive management. Seeking to overcome these obstacles,
Ecosystem-Based Management
ecosystem based management relies on research at all levels
Recognizing the need to sustain the integrity and resili of ecological organization, explicitly recognizes the
ence of socialecological systems has led to calls for dynamic character of ecosystems, acknowledge that
ecosystem based management, a management approach ecological processes operate over a wide range of temporal
that considers all ecosystem components, including and spatial scales and are context dependent, and
humans and the physical environment. With the overall presupposes that our current knowledge of ecosystem func
goal of sustaining ecosystem structure and function, this tion is provisional and subject to change. Ultimately,
management approach: ecosystem based management recognizes the importance
of human needs while addressing the reality that the capa
focuses on key ecosystem processes and their responses
to perturbations;
city of our world to meet those needs in perpetuity has
limits and depends on the functioning of resilient
integrates ecological, social, and economic goals and
recognizes humans as key components of the ecosystem;

defines management based on ecological boundaries

rather than political ones; See also: Ecosystem Ecology.

addresses the complexity of natural processes and social

systems by identifying and confronting uncertainty;
Further Reading
uses adaptive management where policies are used as
experiments and are modified as information is gained; Cardinale BJ, Srivastava DS, Duffy JE, et al. (2006) Effects of biodiversity
engages multiple stakeholders in a collaborative pro
cess to identify problems, understand the mechanisms
on the functioning of trophic groups and ecosystems. Nature
443: 989 992.
Daily GC (ed.) (1997) Natures Services: Societal Dependence on
driving them, and create and test solutions; and Natural Ecosystems. Washington, DC: Island Press.
considers the interactions among ecosystems (terres
trial, freshwater, and marine).
DeAngelis DL (1992) Dynamics of Nutrient Cycling and Food Webs.
New York, NY: Chapman and Hall.
Krebs C (2001) Ecology: The Experimental Analysis of Distribution and
Ecosystem based management is driven by explicit goals, Abundance, 5th edn. San Francisco: Addison Wesley Educational
Publishers, Inc.
executed by policies and protocols, and made adaptable Millennium Ecosystem Assessment (2005) Ecosystems and Human
by using policies as experiments, monitoring their out Well Being: Synthesis. Washington, DC: Island Press.
comes and altering them as knowledge is gained. Pauly D and Christensen V (1995) Primary production required to
sustain global fisheries. Nature 374: 255 257.
Traditionally, management practices have focused on Scheffer M, Carpenter S, Foley JA, Folke C, and Walker B (2001)
maximizing short term yield and economic gain over Catastrophic shifts in ecosystems. Nature 413: 591 596.

Ecosystem Services
K A Brauman and G C Daily, Stanford University, Stanford, CA, USA
2008 Elsevier B.V. All rights reserved.

Introduction Capturing the Value of Ecosystem Services

Defining Ecosystem Services Conclusions
Examples of Ecosystem Services Further Reading

Although many societies have developed the technologi
The worlds ecosystems yield a flow of essential services cal capacity to engineer replacements for some services,
that sustain and fulfill human life, from seafood and tim such as water purification and flood control, no society
ber production to soil renewal and personal inspiration. can fully replace the range and scale of benefits that
Ecosystem Services 27

ecosystems supply. Thus, ecosystems are capital assets, Table 1 A classification of ecosystem services. Examples of
worthy of at least the level of attention and investment ecosystem services and how they can be categorized
given to other forms of capital. Yet, relative to physical,
Provisioning services: Production of. . .
financial, human, and social capital, ecosystem capital is Food
poorly understood, scarcely monitored, and, in many Seafood, agricultural crops, livestock, spices
cases, undergoing rapid degradation and depletion. Pharmaceuticals
Recognition of ecosystem services dates back at least Medicinal products, precursors to synthetic pharmaceuticals
Durable materials
to Plato. This recognition of human dependence on eco Natural fiber, timber
systems, in the past and today, is often triggered by their Energy
disruption and loss. Direct enjoyment of services, such as Biomass fuels, low-sediment water for hydropower
the extraction of timber, fish, and freshwater, can reduce Industrial products
Waxes, oils, fragrances, dyes, latex, rubber
the quantity and quality produced. The provision of eco
Genetic resources
system services can also be affected indirectly and Intermediate goods that enhance the production of other
inadvertently. Deforestation, for instance, has exposed goods
the critical role of forests in the hydrological cycle Regulating services: Generation of. . .
mitigating flooding and reducing erosion. Release of Cycling and filtration processes
toxic substances has uncovered the nature and value of Detoxification and decomposition of wastes
Generation and renewal of soil fertility
physical and chemical processes, governed in part by Purification of air and water
microorganisms that disperse and break down hazardous Translocation processes
materials. Thinning of the stratospheric ozone layer has Dispersal of seeds to sustain tree and other plant cover
sharpened awareness of the value of its service in screen Pollination of crops and other plants
Stabilizing processes
ing out harmful ultraviolet radiation.
Coastal and river channel stability
Control of the majority of potential pest species
Carbon sequestration
Partial stabilization of climate
Defining Ecosystem Services Protection from disasters:
regulation of hydrological cycle (mitigation of floods and
Simply put, ecosystem services are the conditions moderation of weather extremes (such as of temperature
and processes through which ecosystems, and the and wind)
biodiversity that makes them up, sustain and fulfill Cultural services: Provision of. . .
human life. Ecosystem services are tightly interrelated, Esthetic beauty, serenity
making their classification somewhat arbitrary. Recreational opportunities
Cultural, intellectual, and spiritual inspiration
The Millennium Ecosystem Assessment (MA) the for
Supporting services: Preservation of. . .
mal international effort to elevate awareness and Processes underlying services in the classes above
understanding of societal dependence on ecosystems Options
has suggested four categories. maintenance of the ecological components and systems
First, provisioning services provide goods such as needed for future supply of the goods and services above
and others awaiting discovery
food, freshwater, timber, and fiber for direct human use;
these are a familiar part of the economy. Second, and
much less widely appreciated, regulating services main
Tradeoffs in Managing the Flows of Ecosystem
tain a world in which it is biophysically possible for
people to live, providing such benefits as water purifica
tion, pollination of crops, flood control, and climate Biophysical constraints on human activities, such as lim
stabilization. Third, cultural services make the world a ited supplies of energy, land, and water, typically manifest
place in which people want to live; they include recrea themselves as tradeoffs between different uses. Thus,
tion as well as esthetic, intellectual, and spiritual managing ecosystem services involves difficult ethical
inspiration. Fourth, supporting services create the back and political decisions about which services to develop
drop for the conditions and processes on which society and how to do so. At local scales, allocation of limited
depends more directly. All of these services are provided resources to alternative activities typically involves a
by complex chemical, physical, and biological cycles, zero sum game, illustrated by the widespread redirection
powered by the sun, and operate at scales ranging from of water from agriculture to urban and industrial pur
smaller than the period at the end of this sentence to as poses. At global scales, different groups of people compete
large as the entire biosphere (Table 1). for use of Earths open access resources and waste sinks,
28 Ecosystem Services

such as the atmospheres capacity to absorb CO2 and services, from local to global, and explore the tradeoffs
other greenhouse gases without inducing climate change. inherent in their management.
Making informed decisions about how to use ecosys
tem goods and services hinges on understanding these
tradeoffs: knowing the joint products the suite and
Pollination Services Provided by Bees
level of services that ecosystems can provide. For exam
ple, an ecosystem managed exclusively for agriculture Pollination, the movement of genetic material in the
may yield a greater return on agricultural products than form of pollen grains, is a key step in the development of
one managed for multiple services, but understanding most food crops. Even crops that do not rely on insect
that diversified management may produce greater pollination wind pollinated or self pollinated crops
overall returns could influence management decisions are sometimes more productive when visited by an insect
(Figure 1). pollinator. Bees are a particularly important group of insect
Provision of biodiversity is one supporting service that pollinators, responsible for pollinating 6070% of the
has historically been discounted when managing for other worlds total flowering plant species, including nearly 900
ecosystem services. Biodiversity, however, can provide food crops worldwide, such as apples, avocados, cucum
irreplaceable benefits. Genetic diversity, for example, bers, and squash. These crops comprise 1530% of the
allows for both the survival and evolution of the ecosys worlds food production, and bees are credited with $4.2
tems we depend on for myriad benefits. Recent research billion in annual crop productivity in California alone.
indicates that diverse systems are more resilient, and Bees are especially important pollen vectors in part
therefore provide ecosystem services more reliably in because physical adaptations, such as hairs designed to
the long term, than monocultures. While under optimal pick up pollen, and behavioral adaptations, such as fidelity
conditions managing for a single species may provide to a single species of plant on each pollen gathering trip,
superior timber supplies or nutrient sequestration, given ensure good pollen transport and cross pollination.
natural and human caused variability in temperature, In the US, most major agricultural enterprises that rely
rainfall, and other environmental factors, managing for a on bee pollination import managed bees, almost always
diverse system will more consistently provide services in the European honeybee Apis mellifera. The available stock
an uncertain world. of managed honeybees has declined dramatically, how
ever, dropping by over 50% in the last 50 years, while
demand for pollination services has increased in many
areas. This decline in managed bee populations has many
Examples of Ecosystem Services causes, including increased pesticide use, disease in the
hives, and downsizing of stocks that have hybridized with
Ecosystem services can be explored by focusing either on Africanized bees, introducing traits that make managed
a single service that may be provided by various ecosys bees more aggressive and thus a liability to the farmer.
tems or by looking at a single ecosystem that may provide The contribution of native, wild bees to agricultural
a variety of services. Here we illustrate both approaches, pollination was ignored, and assumed to be negligible,
considering first pollination services provided by bees until the early 2000s. Since then, research has shown
then the suite of services provided by wetlands and for that native bees serve an important role in pollination,
ests. We highlight the differences in scale of delivery of picking up slack when managed bee pollination is





Water purification Climate

stabilization $$


Figure 1 Joint products of ecosystems. Many ecosystems are currently managed to exploit only one service. Managing for multiple
services can increase ecosystem benefits.
Ecosystem Services 29

insufficient and enhancing crop production in general. the possibility that managed bee populations could fail
Farms with generous native bee habitat nearby may be because of disease, hybridization, or other causes.
able to fully or partially replace pollination by managed
bees. In some cases, native bees are more efficient polli
Services Provided by Wetlands
nators than European honeybees. The variety of wild
bees, with distinct physical and behavioral traits, allows Areas inundated by fresh, brackish, and salt water are all
them, as a group, to pollinate a wide variety of flowering considered wetlands; among many wetland types are fens
plants. Tomatoes, for example, have pollen that is acces and bogs, tidal marshes, riparian zones, and lakeshores.
sible only by vibrating the flower, which bumble bees and Wetlands, which cover less than 9% of the Earths surface,
some other native bees can, while honeybees cannot. can be extremely productive and many are disproportio
Though tomatoes are self pollinating and do not require nately large providers of ecosystem services. Three of the
an insect vector, native bees promote cross pollination, key services that wetlands provide are flood mitigation,
which, for example, significantly increases the fruit set water purification, and biodiversity support.
and size of Sungold cherry tomatoes. In the upper part of a watershed, many wetlands store
The contributions of native bees to crop production water that flows overland toward rivers and streams.
are usually undocumented and underestimated, and they They can release this water into the main channel slowly,
are always unpaid, at least directly. Though hives of reducing and delaying flood peaks. Downstream, wet
managed honeybees must be rented or maintained, wild lands can absorb and reduce peak flood levels, providing
bees pollinate at no cost to the farmer. Populations of area into which flood waters can spread, dissipating flood
native bees are under great threat, however, by land energy by slowing water movement, and removing flood
management practices that promote the use of pesticides water through transpiration and infiltration.
and the loss, fragmentation, and degradation of habitat. The same physical characteristics of wetlands that slow
Protecting native bees without protecting the ecosystems and absorb overland flow related to flooding can also
provide a mechanism for storing and detoxifying urban
in which they live is impossible. Native habitat, unlike
and agricultural wastewater before it discharges directly
agricultural monocropping, provides the year round sup
into a main channel. Wetlands filter out various nutrients,
ply of blooming plants that wild bees require for
other pollutants, and sediment: they support anaerobic
sustenance. Native habitat also provides nesting areas;
bacteria that denitrify waste; the plants take up and store
most wild bees are solitary, laying a single egg in a nest
nutrients; and by slowing and redirecting water flow, wet
cavity dug into the ground or into dead wood, not forming
lands enhance sedimentation the accreting sediments
social hives. In order to reap the benefits of native polli
can effectively bury pollutants. While many wetlands
nators, food resources and nesting habitat must be
can purify water very economically, their effectiveness
available within a short distance of crops, possibly as
depends on many factors, including rate of inflow, amount
hedgerows, in ditches, or around water ponds. A study of sediment and organics in the wastewater, residence time
of wild bee pollination of coffee in Costa Rica showed that of wastewater in the wetland, and total surface area.
farms closer to tropical forest remnants were visited by A wide variety of animals rely on wetlands for survival.
many more species of wild bees than those further away. Plant species that deliver flood abatement and water
Had the far sites been adequately pollinated, coffee yield purification can also support biodiversity, providing
would have been increased by nearly 20% and misshapen varied food and shelter. A riparian wetland, for example,
coffee beans reduced by 27%. A lower bound estimate of might provide food plants and underground burrows for
the pollination services from these patches is US $62 000 muskrats; seeds, food plants, and nest building materials
per year (in the early 2000s). for ducks; and food and shelter for fish and invertebrates.
The diversity of the native bee population is one of its Wetlands provide a variety of other services as well.
strengths. Many species of bees participate in pollination, Major products associated with wetlands are peat, timber,
and the abundance of different species varies year by year. and mulch. Regulating services in addition to flood
This diversity allows the native pollinator community to mitigation and water purification include waste detoxifi
be both resistant, maintaining functionality in the face of cation, carbon storage, and control of pests and diseases.
environmental upheaval, and resilient, able to reestablish Wetlands provide many cultural services as well,
itself in the wake of a destructive event. When the popula particularly recreation services such as bird watching,
tion of Apis declined dramatically in the second year of the boating, and hunting. Wetlands also provide key support
Costa Rica study, sites close to forest fragments showed ing services, such as soil formation and buffering
minimal loss of pollination while pollinator visits dropped freshwater aquifers from saltwater intrusion.
by nearly 50% further away. Thus, as well as enhancing Worldwide, wetlands are estimated to provide many
pollination services in conjunction with managed bees, billions of dollars in services each year. They are recog
native bee populations provide important insurance against nized by the international treaty, the Ramsar Convention
30 Ecosystem Services

on Wetlands, and regulated by domestic law in many destroyed, as happens when timber is harvested, the car
countries. Nonetheless, they have historically undergone bon they are sequestering is released to the atmosphere.
widespread losses in favor of other land uses; worldwide, Although most organic compounds do return to the atmo
50% of wetlands are estimated to have been lost since 1900. sphere as CO2 when living organisms die and decompose,
While the services provided by wetlands are widely in a functioning forest ecosystem some is buried and
recognized, simultaneously maximizing multiple services sequestered. About 25% of the human caused increase
may not be possible. In some cases this is related to location: in CO2 concentration in the atmosphere during the
upland watersheds may be very important for flood control past 20 years resulted from land use change, primarily
but may be too far upstream to have an impact on water deforestation.
purification. In other cases one service may thrive to the Forests in a watershed, on the hillslopes that drain into a
detriment of another: a wetland that is absorbing a heavy river, influence the water quality in that river. In part this is
nutrient load may be overtaken by a single, aggressive plant because higher intensity uses, such as agriculture input
species and thus fail to be an effective reservoir for biodi pollutants like nutrients and pesticides into a system while
versity. Finally, it can be costly to measure function and forests do not. Forests themselves also reduce sediment and
hence difficult to judge how effectively a wetland is per nutrient runoff. Clearing trees can have an impact as soon as
forming a given service or how to manage for that particular the next rainy season on sediment and nutrient loads in
service. streams, as demonstrated in the classic Hubbard Brook
experiment. In some cases, water users have invested in
forests to keep their water supplies clean. New York City
Services Provided by Forests
recently invested US$ 250 million to acquire and protect
Forests provide a wide array of services, such as timber land in the Catskills watershed that supplies water to the
production, climate stabilization, provision of water quan city. By working with landowners to reduce pesticide and
tity and quality, and cultural benefits, such as recreation. fertilizer application and to plant buffer strips along water
Some management options increase the supply of several ways, New York City reduced potential contamination of
services, but often one service is enhanced to the detri its drinking water. In conjunction with related conservation
ment of others. investments amounting to US$ 1.5 billion, the city
Forests are often managed for provisioning services, thereby obviated the need to build a filtration plant pro
particularly for timber. But even within the category of jected to cost between US$ 6 and US$ 8 billion.
provisioning services, management options differ. If a Forests can also play an important role regulating the
forest is considered exclusively a supplier of timber, timing and quantity of runoff. The economic value
managers will encourage the growth of only certain of forests in the watershed of the Yangtze River above
kinds of trees, possibly nonnative fast growing trees, and Three Gorges Dam, in western Hubei Province, Central
will cultivate them so that they grow in a uniform way, China, was quantified in a study published in 2000. Here,
typically straight and tall. When the trees are deemed the Gexhouba Hydroelectric Power Plant, the largest
mature, they will be cut down, often all at once. By hydro facility in China, producing 15.7 billion kW
contrast, if a forest is regarded as a supplier of diverse annually, requires a narrow range of flows on the Yangtze
benefits, it may be managed to nurture a wide array of in order to run at full power. If the water level is too high,
valued species that would not be available in the mono then water must be released through the sluice gates,
crop forest described above. causing the water level below the dam to rise, reducing
Forests also have both short term and medium term the amount of power that can be produced; at very high
impacts on climate. Temperature regulation happens in flows, turbines are drowned and cannot work at all. If the
forests when the canopy shades the ground and when water is too low, then generators cannot run at full power.
dark colored foliage absorbs heat. Forests can in certain The goal of the hydroelectric facilitys managers is for
circumstances also influence precipitation in cloud for the river to have flow depths that vary as little as possible,
ests, for example, trees and epiphytes intercept and as this has been shown to be much more important for
condense water directly from the air, and that water power generation than the total flow. Upstream forests
runs down trunks to plants and soil below. On a longer damp fluctuations in stream flow by reducing runoff in
timescale, forests play a role in carbon cycling and seques wet periods through canopy interception, leaf litter
tration; when forest plants, bacteria, and algae respire, absorption, and soil and groundwater storage; increased
they take CO2 out of the atmosphere. Plants, soils, and infiltration provides base flow in dry periods through
the animals that eat them in forests, grasslands, and other groundwater discharge. Though water flow regulation is
terrestrial ecosystems store 2000 billion tons of carbon a function of vegetation, soil type, and slope, which occur
worldwide, about half the amount of carbon stored in the in a heterogeneous mix through the watershed, forests
ocean and nearly three times that stored in the atmo and even shrubs with all types of soils and slopes consis
sphere. However, if these ecosystems are burned or tently provided better water regulation than grasses,
Ecosystem Services 31

orchards, and crop agricultural fields. This study esti Capturing the Value of Ecosystem
mated the value of electricity produced by the hydro Services
facility due to water regulation by the forest at over
US$ 600 000 per year (in the early 2000s), or about 2.2 Despite their obvious importance to human well being,
times the income derived from forest product services in people tend to think of ecosystems as being economically
this area. Because trees lose water to the atmosphere productive in narrow terms, often assigning value only to
through transpiration, however, the total water available the production of conventional commodities or to real
downstream was decreased by the forest. estate development. Provision of ecosystem services is only
Different management regimes will yield different rarely considered in costbenefit analyses, preparation
suites of services. Some services can never be coproduced; of environmental impact statements, or other assessments
other services will almost always be produced in tandem, of alternative paths of development. There is no shortage of
though often to differing degrees. For the hypothetical markets for ecosystem goods (such as clean water and water
forest illustrated in Figure 2, cattle and timber cannot be melons), but the services underpinning these goods (such as
produced on the same parcel of land conversion to water purification and bee pollination) often have no mone
pasture optimizes livestock but reduces timber output dra tary value. This is in part because ecosystem services are
matically. Under timber maximization, once trees are generally public goods, free to any user, and therefore
harvested they are not available for climate or hydrologic difficult to value. Because people mostly do not pay for
regulation, though before harvest those services will be them, it can be difficult to discern what the supply, demand,
produced, as well as some habitat and hiking trails. and willingness to pay for services actually are. As a result,
Carbon sequestration, hydropower, recreation, and preser there are no direct price mechanisms to signal the scarcity or
vation of biodiversity tend to be coproduced, but there are degradation of these public goods before they fail.
tradeoffs in their optimal supply. Maximizing biodiversity, While for some goods and services price reflects value
for example, produces all four to their fullest extent but or importance, when ecosystem services are assigned
allows for no timber supply. Bringing selective logging monetary value they tend to be priced much lower than
back into the management regime reduces supply of the their importance suggests. This is true in part because
other services somewhat; maximizing timber yield reduces when supply is much larger than demand, prices are low,
them much more dramatically. no matter how necessary the good. The pricing of dia
Tradeoffs between services are also tradeoffs between monds and water is illustrative. Lost in the desert, a
consumers, such as local recreationalists, regional users of traveler would happily trade all the diamonds in the
hydropower, and global beneficiaries of carbon sequestra world for a single cup of water; back in the marketplace,
tion and biodiversity conservation. These tradeoffs our traveler would find that diamonds are many, many
underscore the importance of valuation, making explicit times more costly than water. Water is inexpensive or free
who benefits from ecosystem services and who pays for because, like many ecosystem goods and services, it tends
them. Conceiving of ecosystem functions as services and to be far more abundant than the volume demanded by
assigning a monetary value to them provides a tool for people; when ecosystems are functioning well, even more
decision makers to weigh different management options. is available.

Convert to pasture Maximize timber yield Maximize biodiversity and Diverse portfolio, selective
recreation logging
Management regime
Livestock production Timber production Carbon sequestration
Hydropower Recreation Preservation of biodiversity
Figure 2 Tradeoffs associated with alternative management objectives for a hypothetical forest ecosystem.
32 Ecosystem Services

Ecosystem services are also often undervalued because Conclusions

prices are based on current supplies and demands, so the
amount we are willing to pay for continued nutrient Because ecosystem services explicitly invoke human
retention in a wetland may be low today even if we can beneficiaries, basic scientific understanding of the eco
predict that nutrient laden runoff from increased agricul system processes producing goods and services is
ture will threaten a downstream fishery tomorrow. meaningful only in the context of economic valuation
Further, prices are based on marginal utility for exam and institutional structures. There is still much to
ple, the amount someone would be willing to pay for the learn on many fronts. Important questions include:
carbon stored in one more tree in a forest. If that forest is Which ecosystems supply which services? What levels
clear cut, we lose all of the carbon storage and, since the and types of ecosystem protection are required to
loss of each tree changes the value of the next, we cannot sustain service supply? Can we develop robust meth
account for the whole loss using the price of the first tree. odologies for the valuation of ecosystems? Even if
Precise valuation of ecosystem services is often not clear answers are absent to all of these questions,
required to provide appropriate economic incentives for numerous and diverse efforts are now underway
protecting the ecosystems that supply them. Incentives worldwide to protect vital ecosystem services, often
need only make it more economically appealing to a using innovative economic incentives.
landowner to maintain hedgerows as habitat for native Explicitly identifying and valuing the goods and
pollinators than to cultivate every last square meter of a services provided by ecosystems has two obvious benefits.
field, for instance, or make it pay to preserve a wetland First, understanding the role of ecosystem services
rather than filling it to build houses. A farm, as illustrated powerfully justifies habitat preservation and biodiversity
below, might generate enough income from nonagricul conservation as vital, though often overlooked, policy
tural commodities to alter its land management regime objectives. While a wetland surely provides existence
(Table 2). Incentives to protect and maintain ecosystems and option values to some people, the benefits provided
can be provided by the government, privately through by the wetlands nutrient retention and flood mitigation
markets, or through hybrid institutions such as cap and services are both universal and undeniable. Tastes may
trade systems supported by government policy. differ over beauty, but they are in firm accord over the
A variety of tools for valuing ecosystem services and high costs of polluted water and flooded homes. Second, if
creating incentives for their conservation are currently given the opportunity, natural systems can in many cases
being developed, including capital markets such as the quite literally pay their own way. Market mechanisms and
institutions that can capture and maximize service values
Chicago Climate Exchange, wetland mitigation banks,
can effectively promote environmental protection at the
and outright payments, often involving privatepublic
local, regional, national, and international levels. In some
partnerships, for services, as is occurring in Australia,
cases, however, protection of ecosystem services will not
Costa Rica, and Mexico. These market based approaches
justify conservation of natural habitats. In other cases, the
provide a much better indication of value than early,
services will be largely irrelevant to environmental pro
more theoretical attempts to quantify the value of eco
tection efforts. While a focus on ecosystem services
system services. While valuation is not necessarily a
provides great potential to promote environmental pro
solution or end in itself, it is a powerful way of organizing
tection, its practical implications remain largely
information and an important tool in the much larger unexamined.
process of decision making.

See also: Riparian Wetlands.

Table 2 A hypothetical farm business in 15 years

Further Reading
Commodity Share of farm business (%)
Brauman KA, Daily GC, Duarte TK, and Mooney HA (2007) The nature
Wheat 40 and value of ecosystem services: An overview highlighting services.
Wool 15 Annual Review of Environmental and Resources 32: 67 98.
Chichilnisky G and Heal G (1998) Economic returns from the biosphere
Water filtration 15
Commentary. Nature 391: 629 630.
Timber 10 Committee to Review the New York City Watershed Management
Carbon sequestration 7.5 Strategy (2000) Watershed Management for Potable Water Supply:
Salinity mitigation 7.5 Assessing the New York City strategy. Washington, DC: National
Biodiversity 5 Academy Press.
Daily GC (ed.) (1997) Natures Services: Societal Dependence on
In this model, traditional agricultural commodities account for 55% of Natural Ecosystems. Washington, DC: Island Press.
revenues, as opposed to 100% today. Nonagricultural income is Daily GC and Ellison K (2002) The New Economy of Nature: The Quest
supplied by a mature market for ecosystem goods and services. to Make Conservation Profitable. Washington, DC: Island Press.
Fundamental Laws in Ecology 33

Daily GC, Soderqvist T, Aniyar S, et al. (2000) The value of nature and National Academy of Sciences of the United States of America
the nature of value. Science 289: 395 396. 99: 16812 16816.
Findlay SEG, Kiviat E, Nieder WC, and Blain BA (2002) Functional Millennium Ecosystem Assessment (2005) Ecosystems and Human
assessment of a reference wetland set as a tool for science, Well being: Current State and Trends: Findings of the Condition and
management and restoration. Aquatic Sciences 64: 107 117. Trends Working Group. Washington, DC: Island Press.
Guo Z (2000) An assessment of ecosystem services: Water flow Postel SL and Thompson BH (2005) Watershed protection: Capturing
regulation and hydroelectric power production. Ecological the benefits of natures water supply services. Natural Resources
Applications 10: 925 936. Forum 29: 98 108.
Heal G (2000) Nature and the Marketplace: Capturing the Value of Ricketts TH, Daily GC, Ehrlich PR, and Michener C (2004) Economic
Ecosystem Services, Washington, DC: Island Press. value of tropical forest to coffee production. Proceedings of the
Heal G, Daily GC, and Salzman J (2001) Protecting natural capital National Academy of Sciences of the United States of America
through ecosystem service districts. Stanford Environmental Law 101: 12579 12582.
Journal 20: 333 364. Zedler JB and Kercher S (2005) Wetland resources: Status, trends,
Kremen C, Williams NM, and Thorp RW (2002) Crop pollination from ecosystem services, and restorability. Annual Review of Environment
native bees at risk from agricultural intensification. Proceedings of the and Resources 30: 39 74.

Fundamental Laws in Ecology

S E Jrgensen, Copenhagen University, Copenhagen, Denmark
2008 Elsevier B.V. All rights reserved.

The Need for Fundamental Laws Other Ecosystem Theories

Systems Ecology in the Jet Stream of Scientific Summary
Development Further Reading
Systems Ecology: Ten Tentative Fundamental Laws
An Attempt to Formulate an Ecosystem Theory

The Need for Fundamental Laws there is an increasing understanding for the need of
knowledge syntheses to a more holistic image. Today
Humans have always strived toward finding a structure or this search for a holistic understanding of complex sys
a pattern in their observations to develop a theory. tems is considered one of the greatest scientific challenges
Science does not make sense without theory. Without of the twenty first century by many scientists.
theory, our observations become only a beautiful collec Several important contributions to systems ecology
tion of impressions without explanation or application to have attempted to capture the features and characteristics
solve problems of human interest. The alternative to of ecosystems, their processes, and their dynamics. The
scientific theory is to observe everything which is not different theories and approaches look inconsistent at
possible. A well developed theory can be used to make first glance, but when examined more closely, their com
predictions. plementarity becomes evident . This commonality and
Our scientific knowledge has to be coherent in order to consensus regarding ecosystem dynamics was asserted by
apply the underlying theory and explain our observations. Jrgensen in the first edition of Integration of Ecosystem
Ecology has only partially been able to condense the Theories: A Pattern (1992), and later editions (2nd edn.
systematic collection of observations and knowledge 1997 and 3rd edn. 2002) have only enhanced the percep
about ecosystems into testable laws and principles. tion that the theories form a pattern and that they are
During the last few decades systems ecologists have highly consistent. It is clear from recent meetings and
developed hypotheses that together with basic laws from discussions that today we have a general ecosystem the
biochemistry and thermodynamics are proposed as a ory which is rooted in a consensus of the pattern of
first attempt to formulate fundamental laws in ecology. ecosystem dynamics. The ecosystem theory presented
The inherent complexity of ecosystems means that it here combines the work of several scientists, and provides
is necessary to break from the long reductionistic scien a foundation for further progress in systems ecology,
tific tradition to a new holistic ecological approach. ecosystem theory, and ecology. Furthermore, it may be
Reductionistic science has had a continuous chain of feasible to use a few fundamental laws to derive other
successes since Descartes and Newton. Lately, however, laws to explain most observations. We do not know yet to
34 Fundamental Laws in Ecology

what extent this is possible in ecology, but at least we all with different individual properties and interaction
propose a promising direction for a useful, comprehensive potentials. It would be impossible to observe all compo
ecosystem theory. Only by the application of the theory nents and even more impossible to observe all the possible
can we assess how and where the theory needs interactions among these 10151020 different components.
improvements. Such complexity leads to a nondeterministic picture in
ecology. In accordance with quantum complementarity,
light can only be described by an interpretation as both
Systems Ecology in the Jet Stream waves and particles (photons). An ecosystem is much
of Scientific Development more complex than light. Therefore, a full (holistic)
description of an ecosystem will also, not surprisingly,
Seven general scientific theories have changed our per require two or more complementary descriptions.
ception of nature radically during the last 100 years: Various descriptions suggest ecosystems as dissipative,
general and special relativity, quantum theory, quantum self organizing systems that follow a dynamic to increase
complementarity, Godels theorem, chaos theory, and energy, emergy, ascendency (see Ecological Network
theory for far from thermodynamic equilibrium systems. Analysis, Ascendency), or eco exergy which are not in
With these seven theories, we understand today that conflict, because they cover different aspects of the eco
nature is much more complex than we thought 100 system. All descriptions help to understand ecosystem
years ago, but we also have tools to understand this com dynamics, but some may be more applicable for addres
plexity better, which has entailed that we have a general sing specific ecosystem questions.
ecosystem theory today. Godels theorem that there are no complete theories
The speed of light is the absolute upper limit for any they are all based on some assumptions is of course also
transmission of matter, energy, and information according valid for ecological theories. We shall not expect a com
to the special relativity theory. This has given a comple plete theory based on no assumptions and which can be
tely new meaning to the concept of locality. It has also in used in all contexts.
systems ecology brought another meaning of network: Newtonian Physics is based on the reversibility of all
links among components that share a locality and of the processes. Prigogines new interpretation of the second
hierarchical organization: networks of smaller and smaller law of thermodynamics has shown that time has an
localities that are linked together on the next level of the arrow. All processes are irreversible and evolution is
hierarchy. Relativity theory also gives us a clear under rooted in this irreversibility. Einsteins special relativity
standing of the lack of absolute measures, which was the theory, which provides the speed of light as an upper
governing scientific perception before the twentieth cen speed making it impossible to change the light signals
tury. When we use ecological indicators to assess which give information about a previous event, also
ecosystem health, we can only apply them relatively to supports the principle of irreversibility. We cannot
other (similar) ecosystems; and, when we use thermo change the past but only the future. With the enormous
dynamic calculations of ecosystems we know that we complexity of ecosystems it also implies that the
cannot get the absolute value but only an index or relative same conditions will never be repeated . Ecosystems
value because ecosystems are too complex to allow us to are always confronted in space and time with new chal
include all the components in our calculations. Quantum lenges, which explains the enormous diversity that
theory and later chaos theory upended the deterministic characterizes the biosphere. Clearly, systems ecology
world picture: we cannot determine the future in all has not developed in a vacuum, but has been largely
detail, even if we know all details of the present condi influenced by the general scientific development during
tions. The world is ontically open. In the nuclear world, the last 100 years. A summary of a general ecosystem
uncertainty is due to our inevitable impact on nuclear theory is presented here. The current proposed theory
particles, while in ecology the uncertainty is due to the consists of ten laws.
enormous complexity. Ecosystems are middle number
systems. The number of components in such systems is
many orders of magnitude smaller than the number of Systems Ecology: Ten Tentative
atoms in a room but too many to be countable. Further Fundamental Laws An Attempt
complicating the situation is that while the atoms are to Formulate an Ecosystem Theory
represented by a few different types all ecosystem com
ponents are different even among organisms of the same A tentative ecosystem theory consisting of eight basic
species. A room may contain 1028 components but they laws has previously been presented, but it seems to be
are represented by only 10 or 20 different types of mole an advantage to split one of the laws into three due to
cules with exactly the same properties. An ecosystem some recent results, which are presented below with a few
contains in the order of 10151020 different components comments.
Fundamental Laws in Ecology 35

1. All ecosystems are open systems embedded in an environ components: the ecosystem is more than the sum of
ment from which they receive energy (matter) input and discharge the parts.
energy (matter) output. From a thermodynamic viewpoint, 7. All ecosystem processes are irreversible (this is probably the
this principle is a prerequisite for ecological processes. most useful way to express the second law of thermodynamics in
If ecosystems could be isolated, then they would be at ther ecology). Living organisms need energy to maintain, grow,
modynamic equilibrium without life and without gradients. and develop. This energy is lost as heat to the environ
This law is rooted in Prigogines use of thermodynamics far ment, and cannot be recovered again as usable energy for
from thermodynamic equilibrium. The openness explains, the organism. Evolution can only be understood in the
according to Prigogine, why the system can be maintained light of the irreversibility principle rooted in the second
far from thermodynamic equilibrium without violating the law of thermodynamics. Evolution is a step wise develop
second law of thermodynamics. ment based on previously achieved solutions to survive in
2. Ecosystems have many levels of organization and operate a changing and dynamic world. Due to the structural and
hierarchically. This principle is used again and again when genetic encapsulation of these solutions, evolution has
ecosystems are described: atoms, molecules, cells, organs, produced more and more complex solutions. Eco exergy
organisms, populations, communities, ecosystems, and the expressed by Kullbachs measure of information (see
ecosphere. The law is based on the differences in scale of Exergy) is one way to measure this development.
local interactions. The distance between components 8. Biological processes use captured energy (input) to move
becomes essential because it takes time for events and further from thermodynamic equilibrium and maintain a state of
signals to propagate. Ecological complexity makes it low entropy and high exergy relative to its surrounding and to
necessary to distinguish between different levels with thermodynamic equilibrium. This is just another way of
different local interactions. expressing that ecosystems can grow. It has been shown
3. Thermodynamically, carbon based life has a viability that eco exergy of an ecosystem corresponds to the
domain determined between about 250 and 350 K. It is within amount of energy that is needed to break down the
this temperature range that there is a good balance system.
between the opposing ordering and disordering 9. After the initial capture of energy across a boundary,
processes: decomposition of organic matter and building ecosystem growth and development is possible by (1) an increase
of biochemically important compounds. At lower tem of the physical structure (biomass), (2) an increase of the network
peratures the process rates are too slow and at higher (more cycling), or (3) an increase of information embodied in the
temperatures the enzymes catalyzing the biochemical system. All three growth and development forms imply
formation processes decompose too rapidly. At 0 K that the system is moving away from thermodynamic
there is no disorder, but no order (structure) can be equilibrium and all three are associated with an increase
created. At increasing temperatures, the order (struc of (1) the eco exergy stored in the ecosystem, (2) the
ture) creating processes increase, but the cost of energy flow in the system (power), and (3) the ascen
maintaining the structure in the face of disordering dency. When cycling increases, the eco exergy storage
processes also increases. capacity, the energy use efficiency, and spacetime dif
4. Mass, including biomass, and energy are conserved. This ferentiation all increase. When the information increases,
principle is used again and again in ecology and particu the feedback control becomes more effective, the animal
larly in ecological modeling. gets bigger, which implies that the specific respiration
5. The carbon based life on Earth has a characteristic basic decreases, and there is a tendency to replace r strategist
biochemistry which all organisms share. It implies that many with K strategists. Notice that the first growth form cor
similar biochemical compounds can be found in all living responds to the eco exergy of organic matter, 18.7 kJ g 1,
organisms. They have largely the same elementary com while the increase of the network plus the increase of the
position, which can be represented using around 25 information correspond to the eco exergy calculated as
elements. This principle allows one to identify stoichio ( 1)c (see Exergy). Notice also that the three growth and
metric relations in ecology. development forms are in accordance with EP Odums
6. No ecological entity exists in isolation but is connected to trends of ecosystem development (Table 1). A typical
others. The theoretical minimum unit for any ecosystem growth and development sequence is present as follows
is two populations, one that fixes energy and another (Figure 1): increased biomass (form 1) has a positive
that decomposes and cycles waste, but in reality viable feedback allowing even more additional solar energy
ecosystems are complex networks of interacting popula capture, until a limit of around 75% of the available
tions. This reinforces the openness principle at the scale solar energy is reached. Thereafter the ecosystem con
of the individual component. The network interactions tinues to grow and develop by increasing network
provide the environmental niche in which each compo interactions (form 2) and improving energy efficiencies
nent acts. This network has a synergistic effect on the (form 3).
36 Fundamental Laws in Ecology

Table 1 Differences between initial stage and mature stage according to Odum (1959 and 1969) are indicated
with reference to the three growth forms

Growth form Properties Early stages Late or mature stage

1 (biomass) Production/respiration >> 1 << 1 Close to 1

Production/biomass High Low
Respiration/biomass High Low
Yield (relative) High Low
Total biomass Small Large
Inorganic nutrients Extra biotic Intra biotic
2 (network) Patterns Poorly organized Well organized
Niche specialization Broad Narrow
Life cycles Simple Complex
Mineral cycles Open Closed
Nutrient exchange rate Rapid Slow
Life span Short Long
Ecological network Simple Complex
Stability Poor Good
Ecological buffer capacity Low High
3 (information) Size of organisms Small Large
Diversity, ecological Low High
Diversity, biological Low High
Internal symbiosis Undeveloped Developed
Stability (resistance to external perturbations) Poor Good
Ecological buffer capacity Low High
Feedback control Poor Good
Growth form Rapid growth Feedback controlled growth
Types r-strategists K-strategists

Reproduced by permission of Elsevier.

100% incoming solar radiation

Growth form 2 and 3: the network,

Exergy captured by the ecosystem

energy cycling and information

are increasing

Growth form 1: the biomass is increasing

and is able to capture more solar radiation

Exergy stored in the ecosystem

Figure 1 The development of an ecosystem is illustrated by plotting exergy captured from the inflowing solar radiation toward the
exergy stored in the ecosystem. Growth form 1 is dominant in the first phase of the development from an early-stage ecosystem to a
mature ecosystem. By increasing the biomass the percentage of solar radiation captured increases up to about 80% corresponding to
what is physically possible. Growth forms 2 and 3 are dominant in the intermediate phase and when the ecosystem is in a mature stage.
Thereby more exergy is stored without increasing the exergy needed for maintenance. The system becomes in other words more
effective in the use of the solar radiation according to Prigogines minimum-entropy principle. The exergy stored is increased for all three
growth forms. Reproduced by permission of Elsevier.
Fundamental Laws in Ecology 37

10. An ecosystem receiving solar radiation will attempt to possible to make exact predictions on their develop
maximize eco exergy storage or maximize power such that if ment due to their enormous complexity.
more than one possibility is offered, then in the long run the one II. Ecosystems have directionality.
which moves the system furthest from thermodynamic equilibrium III. Ecosystems have connectivity.
will be selected. The eco exergy storage and energy flow IV. Ecosystems have emergent hierarchies.
increase during all three growth and development forms V. Ecosystems have a complex dynamics (growth and
see above. When an ecosystem evolves it can apply all disturbances).
three forms in a continuous Darwinian selection process.
Steps toward a wider application of a theoretical explana
The nested spacetime differentiation in organisms opti
tion of ecological observations will reinforce the
mizes thermodynamic efficiency as expressed in the tenth
fundamentals of ecology. The experience from physics
law, because it allows the organism to simultaneously
shows that a theory advances through wider use, because
exploit equilibrium and nonequilibrium energy transfer
every application will either support the theory or
with minimum dissipation.
improve it by demonstrating where it fails. In conclusion,
A special issue of Ecological Modelling (vol. 158) was a tentative ecosystem theory is proposed which has broad
devoted to use the proposed ecosystem theory to explain explanatory power today, but will improve with more
ecological observations that were unexplained in the eco experience providing an even stronger theoretical basis
logical literature. for ecology.

Other Ecosystem Theories See also: Exergy; Hierarchy Theory in Ecology.

The ten fundamental laws presented above have been

formulated in a slightly different manner in the scientific Further Reading
literature and other systems ecologists may emphasize
Elsasser WM (1975) The Chief Abstraction of Biology. Amsterdam:
other aspects. For instance, H. T. Odum could emphasize North Holland.
maximum power more than eco exergy storage; but, as Fath B, Jrgensen SE, Patten BC, and Strakraba M (2004) Ecosystem,
pointed out, they are two perspectives of the same basic growth and development. BioSystems 77: 213 228.
Fath BD, Patten BC, and Choi JS (2001) Complementarity of ecological
dynamic. Such perspectives show that a complex system goal functions. Journal of Theoretical Biology 208(4): 493 506.
should be described by several different viewpoints Ho MW and Ulanowicz R (2005) Sustainable systems as organisms?
according to the complementarity theory. However, this BioSystems 82: 39 51.
Jrgensen SE (1990) Ecosystem theory, ecological buffer capacity,
attempt to provide fundamental laws does not mean that uncertainty and complexity. Ecological Modelling 52: 125 133.
there are no other candidates in the literature. For exam Jrgensen SE (1995) The growth rate of zooplankton at the edge of
ple, the allometric principles are fundamental principles chaos: Ecological models. Journal of Theoretical Biology
175: 13 21.
in ecology. Emergent properties are also sometimes con Jrgensen SE (2002) Integration of Ecosystem Theories: A Pattern, 3rd
sidered sufficiently general to be considered as a edn. Dordrecht, The Netherlands: Kluwer Academic Publishing
fundamental principle. Other ecologists still withhold Company (1st edn. 1992, 2nd edn. 1997).
Jrgensen SE and Fath B (2004) Application of thermodynamic
that fundamental laws exist, preferring to focus on principles in ecology. Ecological Complexity 1: 267 280.
descriptions of fundamental properties and processes. Jrgensen SE, Fath BD, Bastianoni S, et al. (2007) Systems Ecology:
Therefore, the discussion about which laws should be A New Perspective, 275pp. Amsterdam: Elsevier.
Jrgensen SE, Patten BC, and Strakraba M (2000) Ecosystems
considered the fundamental laws in ecology and systems emerging: 4. growth. Ecological Modelling 126: 249 284.
ecology is still open. Jrgensen SE and Svirezhev YM (2004) Towards a Thermodynamic
Theory for Ecological Systems, 366pp. Oxford: Elsevier.
Margalef RA (1968) Perspectives in Ecological Theory. Chicago, IL:
Chicago University Press.
Summary Margalef RA (1995) Information theory and complex ecology.
In: Patten BC and Jrgensen SE (eds.) Complex Ecology, pp. 40 50.
Princeton, NJ: Prentice Hall.
Advancement in our understanding of ecosystem theories Margalef RA (1997) Our Biosphere,178pp. Nordbunte, Oldendorf,
has led to a tentative consensus of the principle laws in Germany: Ecology Institute.
ecology as outlined above. A priority now is to gain wider Margalef RA (2001) Exosomatic structures and captive energies relevant
in succession and evolution. In: Jrgensen SE (ed.) Thermodynamics
application of the theory and to promote, in general, and Ecological Modelling, pp. 117 132. Boco Raton, FL: CRC
ecology as a theoretical science. As such the following Press.
synthesis was recently put forth: Morowitz HJ (1968) Energy Flow in Biology. Biological Organisation
as a Problem in Thermal Physics,179pp. New York: Academic
I. Ecosystems are physically and ontically open, mean Press (see also the review by H.T. Odum, Science 164: 683 684).
Odum EP (1959) Fundamentals of Ecology, 2nd edn. Philadelphia, PA:
ing that they can exchange mass, energy, and W.B. Saunders.
information with the surroundings and that it is not Odum HT (1983) System Ecology, 510pp. New York: Wiley Interscience.
38 Fundamental Laws in Ecology

Odum HT (1996) Environmental Accounting Emergy and Decision Schrdinger E (1994) What is Life? Cambridge: Cambridge University
Making, 370pp. New York: Wiley. Press.
Odum HT (1998) Self organization, transformity, and information. Svirezhev YM (2001) Thermodynamics and theory of stability.
Science 242: 1132 1139. In: Jrgensen SE (ed.) Thermodynamics and Ecological Modelling,
Patten BC (1991) Network ecology: Indirect determination of the life pp. 117 132. Boco Raton, FL: CRC Press.
environment relationship in ecosystems. In: Higashi M and Ulanowicz RE (1986) Growth and Development. Ecosystems
Burns TP (eds.) Theoretical Studies of Ecosystems: The Network Phenomenology, 204pp. New York: Springer.
Perspective, pp. 288 351. Cambridge: Cambridge University Ulanowicz RE (1997) Ecology, The Ascendent Perspective. New York:
Press. Columbia University Press.
This page intentionally left blank
R E Ulanowicz, University of Maryland Center for Environmental Science, Solomons, MD, USA
2008 Elsevier B.V. All rights reserved.

Introduction Centripetality and Agency

Autocatalysis in Ecology Further Reading

Introduction the bladder. This feeding upon microheterotrophs

helps the Utricularia to grow and increase its surface
In chemistry the term catalysis means the speeding up of a area (process A). In nature the surface of Utricularia
chemical reaction. It follows that autocatalysis then means plants is always host to a film of diatomaceous algal
the catalysis of a chemical reaction by one of the pro growth known as periphyton, so that more surface
ducts of the reaction. For example, oxalic acid oxidizes area encourages the growth of more periphyton (pro
purple permanganate. When a few crystals of MnSO4 cess B). More periphyton in its turn means more food
are added to a mixture of the chemicals, the conversion to support the growth of any number of species of
to Mn(II) is sped up. If no MnSO4 is added, then the small microheterotrophs (process C). The autocataly
reaction will gradually speed up of itself, because Mn(II) tic cycle is closed when it is noted that a greater
is gradually being created by the reaction, and this pro density of microheterotrophs provides more resources
duct autocatalyzes the reaction itself. Autocatalysis in for the Utricularia to grow (process A again) by cap
chemistry is usually considered to occur among relatively turing and absorbing more abundant zooplankton
simple, fixed, and inflexible reactants. As such it is com (Figure 2b).
monly regarded as a subclass of general mechanisms. Unlike in chemistry, the actors in ecology are more
complex, malleable entities with capabilities to undergo
small, incremental alterations. Such malleability sub
Autocatalysis in Ecology stantially enhances the repertoires of autocatalysis and
enables it to exhibit some very nonmechanical beha
In systems ecology, autocatalysis is regarded as a general viors. This is especially the case when autocatalysis
ized form of mutualism, that is, an association between involves processes that can change in stochastic and
organisms of two different species in which each member nonpredictable ways. An important characteristic of
benefits. In systems ecology focus remains more on pro causal cycles (e.g., autocatalysis) is that when random
cesses and less on objects. Hence, an autocatalytic events impinge upon them, they usually yield nonran
configuration of two or more ecological processes is one dom results. This is the consequence of the first and
in which the processes can be arrayed in a closed cycle, foremost attribute of autocatalysis its generation of
wherein each process in the cycle facilitates the next. selection pressure.
Without loss of generality, one may focus on a serial, To see how autocatalysis generates selection, one
circular conjunction of three processes A, B, and C begins by considering a small spontaneous change in
(Figure 1). Thus, any increase in the rate of process A is process B. If that change either makes B more sensi
likely to induce a corresponding increase in process B, tive to A or a more effective catalyst of C, then the
which in turn elicits an increase in process C, and whence transition will receive enhanced stimulus from A. In
back to A. the Utricularia example, diatoms that have a higher P/
A didactic example of autocatalysis in ecology is B ratio and are more palatable to microheterotrophs
the community that builds around the aquatic macro would be favored as members of the periphyton com
phyte, Utricularia (commonly called Bladderwort). All munity. Conversely, if the change in B makes it either
members of the genus Utricularia are carnivorous less sensitive to the effects of A or a weaker catalyst of
plants. Scattered along its feather like stems and C, then that perturbation will likely receive dimin
leaves are small bladders, called utricles (Figure 2a). ished support from A. Hence, the response of this
Each utricle has a few hair like triggers at its term causal circuit is decidedly not symmetric, and out of
inal end, which, when touched by a feeding this asymmetry emerges a direction. This direction is
microheterotroph, opens the end of the bladder, and not imparted or cued by any externality; its action
the animal is sucked into the utricle by a negative resides wholly within the system. As one might expect
osmotic pressure that the plant had maintained inside from a causal circuit, the resulting directionality is in

42 Autocatalysis

with teleology. There is no externally determined or

+ preexisting goal toward which the system strives.
+ Direction arises purely out of the immediate response
+ by the internal system to a novel, random event
impacting one of the autocatalytic members.
Figure 1 Schematic of a hypothetical three-component
autocatalytic cycle.
Centripetality and Agency

(a) A second important and related directionality emerges

out of autocatalysis that of centripetality. To see this
one notes in particular that any change in B is likely to
involve a change in the amounts of material and energy
that are required to sustain process B. As a corollary to
selection pressure one immediately recognizes the ten
dency to reward and support any changes that serve to
bring ever more resources into B. Because this condi
tion pertains to any and all members of the causal
circuit, any autocatalytic cycle becomes the epicenter
of a centripetal flow of resources toward which as many
resources as possible will converge (Figure 3). That is,
an autocatalytic loop defines itself as the focus of
centripetal flows. A didactic example of such centripe
tality is coral reef communities, which by their
considerable synergistic activities draw a richness of
(b) nutrients out of a desert like and relatively inactive
surrounding sea.
The centripetality generated by autocatalysis is a
Zooplankton much neglected and essential attribute of the life pro
cess. For example, evolutionary narratives are replete
with explicit or implicit references to such actions as
striving or struggling, but the origin of such directional
Utricle behaviors almost always remains unmentioned. Such
actions are simply postulated. Centripetality, however,
appears to be at the very roots of such behaviors. To see
this, one only needs to recognize that it is centripetality
that gives rise to the much vaunted competition, which is
yto the crux of evolutionary theory. For centripetality guar
riph layer
Pe tony antees that, whenever two or more autocatalytic loops
Leaflet iph

Figure 2 (a) Sketch of a typical leaf of Utricularia floridana,

with detail of the interior of a utricle containing a captured Centripetality
invertebrate. (b) Schematic of the autocatalytic loop in the
Utricularia system. Macrophyte provides necessary surface upon
which periphyton (speckled area) can grow. Zooplankton
consumes periphyton, and is itself trapped in bladder and
absorbed in turn by the Utricularia.

part tautologous, that is, autocatalytic systems respond

to random events over time in such a way as to
increase their degree of autocatalysis. It should be
emphasized that this directionality, by virtue of its
internal and transient nature, should not be conflated Figure 3 Centripetal action as engendered by autocatalysis.
Autocatalysis 43

(a) (b) (c)



Figure 4 (a) Original configuration. (b) Competition between component B and a new component D, which is either more sensitive to
catalysis by A or a better catalyst of C. (c) B is replaced by D, and the loop section ABC by that of ADC.

exist in the same system and draw from the same pool of functions. On the one hand is the stochastic, entropic
finite resources, competition among the foci necessarily tendency to fall apart, which at the same time generates
ensues. In particular, whenever two loops share pathway new diversities of form and behavior. Arrayed against the
segments in common, the result of this competition is inevitable centrifugal drift toward disorder is the autoca
likely to be the exclusion or radical diminution of one of talytic selection and centripetal pull toward greater
the nonoverlapping sections. For example, should a new activity and tighter organization. Opposing thrusts though
element D happen to appear and to connect with A and they are, the continued development of life would be
C in parallel to their connections with B (Figure 4), then impossible without the actions of both.
if D is more sensitive to A and/or a better catalyst of Finally, the focal position that autocatalytic configura
C, the ensuing dynamics should favor D over B to the tions of processes occupy in the phenomenon of life is
extent that B will either fade into the background or aptly illustrated by considering what differs between a
disappear altogether. That is, the selection pressure living organism (say a deer) and the same entity immedi
and centripetality generated by complex autocatalysis ately upon death. The mass of the deer remains the same,
(a macroscopic ensemble) is capable of shaping and repla as does its overall form, chemical constitution, embodied
cing its own elements. Perhaps the instances that spring energy, and genomic configuration. What the live deer
most quickly to mind here involve the evolution of obli had that the dead deer no longer possesses is simply its
gately mutualistic pollinators, such as yuccas and yucca configuration of autocatalytic processes.
moths, which coevolve with the yucca so as to displace
other pollinators.
One notes in passing that the same tendency to replace See also: Ecological Network Analysis, Ascendency;
B with D could as readily replace a defective or destroyed Ecological Network Analysis, Energy Analysis.
B with another similar component B9, that is, autocatalysis
lies behind the ability of living systems to repair themselves.
It becomes obvious that the autocatalytic system is no
longer acting merely at the behest of externalities, but it is Further Reading
actively drawing ever more resources unto itself. In fact, Eigen M and Schuster P (1979) The Hypercycle: A Principle of Natural
the tendency of centripetality to transform as much as Self Organization. Berlin: Springer.
possible into itself lies at the very crux of evolutionary Kauffman SA (1995) At Home in the Universe: The Search for Laws of
Self Organization and Complexity. New York: Oxford University
drive; for absent such striving, there would be no compe Press.
tition at the next level up. Ulanowicz RE (1995) Utricularias secret: The advantages of positive
Furthermore, one perceives autocatalytic action as the feedback in oligotrophic environments. Ecological Modelling
79: 49 57.
agency behind one of a pair of agonistic tendencies that Ulanowicz RE (1997) Ecology, the Ascendent Perspective. New York:
together account for the patterns of life forms and Columbia University Press.
44 Body-Size Patterns

Body-Size Patterns
A Basset and L Sabetta, Universita` del Salento Lecce, Lecce, Italy
2008 Elsevier B.V. All rights reserved.

Background Community Level Patterns

The Problem of Measuring Body Size Decoding Mechanisms
Population and Species-Level Patterns Further Reading

Background trophic links and competitive ranking among co occur

ring species differing in body size. However, the simple,
Life in the biosphere shows an impressive variety of energy related, taxon free explanation emphasizes the
individual shapes and body sizes. From the smallest ecological relevance of body size patterns, which are
microorganism (approx. 10 13 g) to the largest mammal conceptually independent of species specific resource
(>108 g), living things cover more than 21 orders of mag requirements and species composition.
nitude of body size. The largest living organisms are Body size patterns include population or species level
actually plants (giant sequoia, Sequoiadendron giganteum patterns, such as the body size range pattern, and com
(Lindl.) Buchholz), but since most of their bodies are munity, landscape, or continental level patterns. The
actually dead bark tissues, their living biomass is lower latter include variations with individual body size in
than that of the largest mammals. Given this impressive number and biomass of individuals, number of species,
variability of sizes, consistent body size patterns, so com population densities, and energy used by populations.
mon at every scale of observation as to be considered The body size ratio between co occurring species pairs,
universal, can be detected. known as the Hutchinson ratio, also shows deterministic
The first body size patterns to be emphasized were and consistent patterns of variation within the commu
that there are many small and few large individuals and nity level body size patterns.
species in the biosphere. The range of body sizes from
the smallest to the largest individuals may vary sub
stantially, when moving from marine to brackish water The Problem of Measuring Body Size
to freshwater and terrestrial ecosystems, as well as from
tropical to polar ecosystems or from lowlands to high Measurements of body size include linear dimensions
lands, but the pattern of many small and few large (e.g., body length, body width, and the length or width
individuals still holds. This simple and universal obser of some morphological attribute of individuals), body
vation was reported by Charles Elton in the first half of surface area and biovolume, and weight (e.g., wet
the twentieth century in his pivotal book Animal Ecology. weight, dry weight, and body mass as ash free dry
This pattern can be explained by means of simple, weight). The energy content of biomass, measured in
taxon free, energy related arguments: since small indi energy units, can also be used as a measurement of body
viduals require less energy per unit of time than large size.
individuals for their maintenance and activity, a fixed Body size patterns are derived from individual bio
productivity will support, at equilibrium, a higher den mass data, although the original data may have been
sity of small than large individuals. This explanation is obtained as the body length, morphological attribute
actually an oversimplification of the real world; there length or size, body wet weight, body volume, or cell
are at least two other components that need to be taken volume of unicellular individuals. Some conversion is
into account in order to decode the body size abun required, because individual biomass cannot always be
dance patterns into a deterministic mechanism of measured, although body size in general is an easily
community organization: a phylogenetic and evolution measurable characteristic of individuals. Indeed, in
ary component, determining the actual diversity of many cases it is necessary to avoid the destructive analysis
species and body sizes at continental and global scales; required to measure individual biomass, and in other
and an interaction component, selecting the body sizes cases individuals are simply too small.
and the species best suited to withstand the locally Indirect measurements of individual biomass, where
occurring abiotic conditions and structural habitat lengths or biovolumes are converted to weights, may
architecture (abiotic niche filtering), and determining make the body size patterns weaker or harder to detect,
Body-Size Patterns 45

depending on the dimensions measured, on the preci These mechanistic explanations of the relationship
sion of the allometric relationship used with respect to between range size and body size are not mutually exclu
the specific set of data, on the precision of biovolume sive and may be reinforcing.
detection, and on the adequateness of the conversion
equations. As regards the weight per length allometry,
the comparability of the seasonal period, climatic con Community Level Patterns
ditions, sex ratio and the reproductive status of
individuals, and resource availability all have to be Body SizeAbundance Distributions
taken into account as major sources of variation. As Body sizeabundance distributions describe the variation
regards biovolume, the complexity of individual or cell of some measurements of individual abundance with
shape, taxon specific weight per unit of biovolume, and individual body mass. The measurements of abundance
the type of weight unit used (C, biomass) all have to be used are number and biomass of individuals of each
taken into account in order to minimize the bias intro population within a guild or a community, number or
duced by using indirect measurements and conversion biomass of individuals in successful populations within a
factors. species range, at the regional, continental, and global
scale, number or biomass of individuals within a com
munity and number or biomass of individuals in base 2
Population and Species-Level Patterns logarithmic body size classes. Whatever criteria for
Range-Size Patterns grouping individuals are selected, within populations,
communities or size classes, at the guild, community,
The range of a species is its natural area of geographic landscape, continental or global scale, as number or
distribution. Considering the overall range of species and biomass, a negative relationship between individual
body sizes occurring in the biosphere, there does not seem density and body size is generally observed. However,
to be any simple and deterministic relationship between the shape and coefficient of these relationships, the
body size and species range size: very large species, such mechanisms involved and the ecological significance
as some cetaceans, and very small species, such as many vary according to the criteria selected, and each single
microorganisms, can have very wide natural ranges. body size pattern provides different information, contri
However, within much more restricted taxonomic buting to a better understanding of the role of individual
groups, small bodied species tend to have smaller mini body size in structuring and organizing ecological
mum geographic ranges than large bodied species. The communities.
interspecific relationships of body size to geographic The selection of either species populations or body
range size commonly exhibit an approximately triangular size classes as a grouping criterion creates two main
form, where species of all body sizes may have large categories of sizeabundance distributions: taxonomi
geographic ranges while the minimum range size of a cally based and nontaxonomically based. The latter are
species tends to increase with body size. commonly referred to as size spectra. Studies of terres
The relationship between body size and home range trial ecosystems have preferentially used body size
size (i.e., the minimum space needed by an individual to abundance distributions based on the taxonomic grouping
successfully complete its life cycle) can help to account of individuals into populations and communities, whereas
for patterns of natural range size. Since home range size studies of aquatic ecosystems have preferentially used
(H ) scales with individual body size (BS) according to an body sizeabundance distributions as taxon free pat
allometric equation (H aBSb), in which the slope (b) is terns, grouping individuals into logarithmic body size
significantly larger than 1, large bodied species may classes independently of their taxonomy.
require a larger total geographic range than small species
in order to maintain minimum viable population sizes in
all local areas. This results in the triangular relationship Taxonomically based sizeabundance
between body size and range size, because there is not distributions
necessarily an upper limit on the range size of small On average, population densities (PDs) scale with indivi
bodied species. dual body size (BS) according to the allometric equation
The dependence of a species fundamental niche
PD a1 BSb1
space and dispersal ability on body size may also help
to explain range size patterns, since species of large body where b1 is typically lower than 0 and a1 is the specific
size are potentially able to maintain homeostasis in a density. a1 expresses the combined action of factors such
wider range of conditions and to successfully colonize a as average energy transfer efficiency, average energy
larger proportion of their potential range than small availability, and temperature driven shifts in the meta
bodied species. bolic rates of the populations in question.
46 Body-Size Patterns

Broadly speaking, taxonomically based sizeabundance These include the resource density perceived by indivi
distribution derives from the notion that since the energy duals and the individuals exploitation efficiency, both of
requirement of individuals (Met) increases with individual which decrease with increasing body size. Intercepts of
body size according to a well known allometric equation global sizeabundance distributions express the average
energy use efficiency of the group of populations consid
Met a2 BSb2 ered. Compilations of global size abundance distributions
for ectothermic and endothermic species show different
where b2 has been consistently found to be close to 0.75, intercept (a1) values, the former having less negative
the number of individuals of each population supported intercepts than the latter due to the cost of being homoeo
by the available resources must decrease with average thermic. Similarly compilations of size abundance
individual body size. Assuming that resource availability distributions of herbivores have higher a1 values than
is homogeneous across species and body sizes, the slope of those obtained for carnivores, reflecting the overall effi
the body sizeabundance distribution (b1) is expected to ciency of energy transfer in food webs.
be 0.75. When sizeabundance distributions are compiled at
The processes underlying body sizeabundance dis the local level, where the body size and abundance of
tributions, and hence their information content and
each species (N ) is measured at the same location, body
ecological meaning, depend on whether they account
size generally explains only a small part of the variation in
for density values and average body sizes of species on a
population abundance, and the regression slope is much
regional, continental, or global scale (hereafter, global
higher than the expected 0.75. The observed deviations
scale sizeabundance distributions), for density and aver
from the expected slope in local sizeabundance distribu
age individual body size of co occurring populations
tions are suggestive of size biases in resource acquisition
within guilds or communities (hereafter, local scale
that could be driven by size asymmetry in competition.
sizeabundance distributions), or for average population
An alternative hypothesis to explain the deviation of local
densities and individual body sizes of entire guilds or
sizeabundance distributions from global ones is that the
communities along ecological, climatic, or biogeographic
former typically examines a smaller range of sizes than
gradients (hereafter, cross community sizeabundance
the latter. Observing a smaller portion of the overall
Global scale sizeabundance distributions are among relationship accentuates the noise in the local sample.
the most extensively studied. They cover regional, con This could explain why local sizeabundance distribu
tinental, and global scales, and the broadest range of tions in aquatic environments, covering a larger spectrum
taxonomic variation, with a bias toward birds and mam of body sizes than terrestrial ones, are also generally
mals, for which more extensive databases of population stronger. In fact, at the local scale, triangular shaped
densities and body sizes are available at every spatial sizeabundance distributions are much more commonly
scale. Data used to compile global scale sizeabundance observed than simple allometric relationships. Triangular
distributions typically describe densities of successful distributions have three major attributes: an upper
populations within the species geographic range, which bound, a lower bound, and a dispersion of points in the
may be close to the maximum carrying capacity. Most sizeabundance space (Figure 1a). The upper bound of
commonly, populations included in the global scale size the triangular shaped sizeabundance distributions is
abundance distributions do not coexist, and affect each determined by the body size scaling of the dominant
other through vertical or horizontal interactions. For species population densities. The upper bound has
large compilations of population densities, population been used as a proxy of the complete local size
density generally scales very closely with body size, abundance distribution, under the assumption that the
with a slope near the value of 0.75. The close agreement ecological role of rare and occasional species, being
between the slope observed for global scale sizeabun weak, is unclear. The procedure may be useful for applied
dance distributions and that expected on the grounds of purposes but since most species are rare the assumption is
simple energetic arguments confirms that at the continen not generally acceptable. The body size dependency of
tal and global scales, availability of resources or energy is the minimum viable population may explain an expected
not correlated with species body size. The homogeneity lower bound, which is difficult to measure because of the
of resource or energy availability across species body sizes problems with correctly quantifying the rarity of popula
is an interesting, but far from straightforward, aspect of tions. The density of points between these two bounds is
global sizeabundance distributions. It implies that the determined mainly by regional processes and horizontal
advantage for large species arising from their wider niches and vertical partitioning rules. The ecological informa
(and thus greater availability of resources) with respect to tion carried by the intercepts of the sizeabundance
small species, is counter balanced by the presence of distributions is of lower value at local than at global
other body size related factors which compensate. scale because whenever slopes are different, as often
Body-Size Patterns 47

(a) BS and Ntot. In general cross community sizeabundance

Species abundance (log(no. m2))
distributions tend to be well described by allometric
equations, whose slopes tend to be similar to the inverse
of the scaling exponent of metabolic rates with individual
body size. A similarity between observed and expected
slopes has been also detected in guilds and communities
which are not regulated by self thinning rules, such as
2 bird and phytoplankton guilds. However, since much
4 fewer data are available for cross community size
6 abundance distributions than for global and local size
0 5 10 15 20 abundance distributions, the underlying mechanisms
Body size (log(g)) remain to be determined.
Number of species

Nontaxonomically based sizeabundance

30 In large aquatic ecosystems, early studies of body size
20 abundance distributions focused on energy transfer (i.e.,
10 how information on productivity and energy transfer may
0 be gained from body size data, which can be collected
1 3 5 7 9 11 13 15 17 19 21 relatively easily). In accordance with this objective, they
Body size (log(g)) dealt with particles rather than with species, dividing
(c) particles suspended in the water column into logarith
1000.00 185 taxa 46 taxa
mic base 2 size classes, irrespective of species and
19 taxa
Body size (log(mg))

100.00 including nonliving organic particles. Thus the ni parti

10.00 cles in the ith body size class of average mass BSi may
1.00 represent more than one species, and every species can
occur in more than one class. Nontaxonomic size
abundance distributions (hereafter referred to as size
0.01 spectra) have been quantified for many different guilds
0.00 and communities, including plankton, benthos and fish
0 50 100 150 200 250
guilds, woodland and forest plant guilds, as well as mar
Species rank
ine, freshwater, and terrestrial ecosystems; however, a
Figure 1 Body-size patterns of macroinvertebrate guilds of large proportion of the ecological literature addressing
transitional water ecosystems in the Mediterranean and Black
Sea Eco-regions. Both local sizeabundance distributions (a) and
size spectra deal with the pelagic marine environment.
body sizespecies distributions (b) are triangular shaped. The According to the classification reported for taxonomi
upper bounds of the triangular distributions are reported. The cally based body sizeabundance distributions, almost all
graph (c) emphasizes that species of transitional water size spectra are local, being determined at the guild or
macroinvertebrates are clumped around the mode of the body community scale. Size spectra can be compiled with two
sizespecies distribution, with 74% of the species being grouped
in 2 out of the 5 order of magnitudes occurring between the size
different types of data, that is, biomass and number of
of the smallest and the largest species. individuals. Both biomass size spectra and number size
spectra can cover different body size ranges, describing
either entire communities or single guilds.
occurs when comparing local sizeabundance distribu Regarding biomass size spectra, the amount of biomass
tions, comparisons between intercepts are not possible. has been shown both empirically and theoretically to be
Classifying all the individuals in a population into constant when plankton individuals are organized into
guilds or communities and averaging their mass, we may logarithmic size classes. As a result of this equal partition
then describe every guild and community with two sim ing of biomass, the slope of a straight line fitted to
ple parameters: mean organism size (BS) and total plankton biomass size spectra is expected to be 0; this
community abundance (Ntot). The scaling of total com relationship is known as the linear biomass hypothesis,
munity abundance with mean organism size leads to which has strong experimental support in aquatic pelagic
cross community sizeabundance distributions. Cross environments, particularly when a large spectrum of sizes
community sizeabundance distributions were first stud and trophic levels are considered. Often the data is sub
ied in self thinning plant and sessile communities, where, jected to a normalization procedure, which consists of
as organisms grow, there is space for fewer and fewer dividing the biomass in each size class by the width of
individuals, determining a negative relationship between the size class. In normalized biomass size spectra, biomass
48 Body-Size Patterns

in each size class decreases isometrically with the average Since b2 has been found to be consistently close to 0.75,
class size, the slope being close to 1. The linear biomass the scaling of energy use rates with individual body size
hypothesis implies that in pelagic systems, the number of depends on b1, which is generally expected to be negative,
individuals within logarithmically increasing size classes since, at every spatial scale of ecological organization,
declines linearly with average body size. The slope of the many small and few large individuals occur.
allometric equation tends to be close to 1; when number Assuming that resource availability is homogeneous
size spectra are normalized, the expected slope is equal across species and that species do not limit each others
to 2. Nevertheless, within pelagic size spectra, a series of resource availability and have optimized the efficiencies
dome like distributions are typically detected, corre of resource exploitation and use, then population densi
sponding mainly to different functional guilds within ties are expected to scale with individual body size with a
which there is a poor fit with linear statistical regressions. slope (b1) of 0.75, and the amount of energy each species
Dome like distributions and gaps in number and uses per unit of area is expected to be independent of
biomass size spectra occur not only between but also body size:
within functional groups, such as phytoplankton and zoo
E a2  a1 BS0:75 0:75 a3 BS0
plankton, even when they are not attributable to
incomplete censuses of species or to systematic under The independence of energy use per unit area from
estimation of intraspecific size variation. Dome like body size is known as the energetic equivalence rule
patterns of biomass distribution have been observed (EER). Whenever b1 is consistently lower, more negative,
both in freshwater and marine ecosystems, as well as in than 0.75, small species dominate energy use.
macro zoobenthos and fish. Therefore, by restricting the Conversely, if b1 is consistently larger, less negative,
range of body size considered and addressing specific than 0.75, large species make a disproportionately large
functional groups, size spectra tend to have a shape simi use of the available energy per unit of area.
lar to the triangular shape of local sizeabundance Global sizeabundance distributions seem to agree
distribution. Most commonly, the maximum number with the EEF. At the global scale, the energy use of
and biomass of individuals, either partitioned into species the most successful populations within the species range
or irrespective of species, occur at some small but inter seems to be actually independent of the body size of
mediate body size, rather than at the smallest size. individuals within populations. On the other hand, local
Two kinds of scaling in the relationship between body sizeabundance distributions, which commonly show
size and abundance within size spectra may be recog scaling exponents higher, less negative, than 0.75 show
nized. A unique and primary slope reflecting the size that within local guilds and communities large species
dependency of metabolism (metabolic scaling), and a normally dominate energy use. Dominance of small spe
collection of secondary slopes which represent the scaling cies has also been detected at the local scale usually in
of numerical or biomass abundance with body size within relation to some degree of stress. Therefore, the shape and
groups of organisms having similar production efficien slope of local sizeabundance distribution, and conse
cies (ecological scaling). Size dependent coexistence quently the body size scaling of energy use, can have
relationships are likely to be representative of the second practical applications in ecology.
ary slopes, leading to a dominance of large cells/species,
and slopes that are less negative than predicted by the
linear biomass hypothesis. Ecological scaling can also Body-SizeSpecies Distributions
produce dome like patterns in size spectra within the Understanding biodiversity is a major goal of ecology.
size range of each functional group. Since many small and few large species occurs in the
biosphere, at every scale, from the community to the
continental and global level, describing and understand
ing the scaling of biodiversity patterns with individual
Body SizeEnergy Use Distributions
body size is also a key topic.
The body size dependence of both metabolic rates and Basically, whenever organisms perceive a two
population densities makes it possible to evaluate popula dimensional (2D) habitat, they sample habitats on a grid
tions rates of energy use and how they scale with proportional to the reciprocal of the square of their linear
individual body size. Indeed, the rate at which energy dimension (L). Therefore, the likelihood of niches being
flows through a population (E ) can be evaluated as the opened up to species specializing in particular resources
product of individual metabolism (Met) and population and habitat patches is proportional to L 2 or to BS 0.67.
density (PD), as follows: Consequently, the number of species (S) is expected to
decrease with individual body size according to L 2.
E Met  PD a2 BSb2  a1 BSb1 a2  a1 BSb2 b1 Whenever organisms perceive a 3D habitat, the species
Body-Size Patterns 49

number is expected to be proportional to L 3 or BS 1. consistent with the Hutchinson ratio have been observed
Considering that the linear dimension of individuals for many groups of animals, including, birds, desert
represents the ruler (L) they use to sample the habitat rodents, and lizards. The Hutchinson ratio corresponds
and that habitats are rarely completely homogeneous at to a limiting similarity threshold; therefore, the average
every scale of perception, individuals perceive the habitat size ratio between species is expected to vary with
to be fractal. The perceived 2D habitat scale is L 2D, resource limitation. Actually, the average size ratio
where D is the fractal dimension of the habitat as well as between species pairs decreases with increasing richness
of the resources. The fractal dimension is a habitat prop and with decreasing guild trophic level.
erty that in many field studies has been found to be close That co occurring species within guilds tend to have
to 1.5. This would mean that in 2D habitats the number of different body size, with an average size ratio close to the
species (S) is expected to be between L D and L 2D, that is, expected 22.26, is a very general observation in ecology.
between L 1.5 and L 3.0, where L 2D is analogous to the However, the ecological relevance of the Hutchinson
upper bound of sizeabundance distributions. In 3D ratio has been questioned, mainly due to two key critical
habitats the number of species (S) is expected to be observations, apparently in contrast with the interpreta
proportional to L 3D, that is, to L 4.5. Assuming D  1.5, tion that size ratios between species correspond to a low
a tenfold decrease in individual size determines a three enough niche overlap between species pairs to allow
fold increase in the perceived length of each habitat edge, interspecific coexistence: (1) many nonliving things in
a tenfold increase in apparent habitat surface and a max nature as well as many objects built by humans, from
imum tenfold increase in S. nails to musical instruments, are scaled in size according
Available data on both the full range of taxa and to the Hutchinson ratio; and (2) size spacing between
particular groups of species consistently show that the species pairs does not always seem to be related to niche
speciessize distributions are humped, with the mode in spacing. The latter is the most critical issue. However, a
some small but intermediate size class (Figure 1b). The functional link between the body size ratios of coexisting
underestimation of the number of existing small species species and competitive coexistence conditions may also
may be an explanation of humped distributions covering be derived independently of any niche spacing. Body
the whole scale of size from the smallest to the largest size mediated coexistence between species differing in
species. Underestimation of small species is less likely to size may result from simple energetic constraints on indi
explain humped distributions observed within restricted vidual space use regardless of any a priori resource
taxonomic groups, such as invertebrates, birds, and mam partitioning: that is, size ratios between species may cri
mals. Within restricted taxonomic groups, it seems likely tically affect species coexistence even if niche spacing is
that an optimal body size exists, where species and indi not detected.
viduals perform optimally and tend to be clumped
(Figure 1c). An optimal body size of between 100 g and Decoding Mechanisms
1 kg has been proposed for mammals and an optimal body
size of 33 g has been proposed for birds. Two hypotheses At their most simple, body size patterns depend on
have been proposed to explain the size dependency of phylogenetic and evolutionary constraints, on energetic
species performance at every scale: the energy conversion constraints, and on interactions with the habitat structure
hypothesis, addressing optimal size according to the size and with co occurring species.
dependency of the efficiency of energy conversion into As regards phylogenetic and evolutionary constraints,
offspring; and the energy control hypothesis, addressing body size patterns are in some way dependent on existing
optimal body size according to the species performance biodiversity and its evolutionary basis. Each taxon per
in monopolising resources. forms best under a fixed set of conditions and has
bioengineering constraints on its performance. For exam
ple, insects cannot be too large and birds cannot be too
Body-Size Ratios
small; therefore, although both of them can take advan
Coexisting species of potential competitors commonly tage of a 3D space, the complete spectrum from insects to
differ in body size. Using consumer body size as a proxy birds has dome like distributions which incorporate the
of resource size, this difference may explain competitive bioengineering constraints of the two groups of species.
coexistence; in his famous paper Homage to Santa As regards energy constraints, metabolic theory gives a
Rosalia: or why are there so many kind of animals general explanation of body size patterns in terms of
Hutchinson proposed that in order to coexist species energy and temperature constraints on metabolism and
must be spaced in size with a ratio between their linear the intrinsic properties of energy partitioning. Metabolic
dimensions of at least 1.28 (2.02.26 in biomass), which is theory sets the theoretical expectations of body size pat
commonly referred to as the Hutchinson ratio. Patterns terns, under the assumption that they are basically driven
of body size spacing between coexisting species pairs by simple energy constraints
50 Cycling and Cycling Indices

As regards interactions, clearly populations interact Gaston K (2003) The Structure and Dynamics of Geographic Ranges.
Oxford: Oxford University Press.
with their environment and with co occurring species. Holling CS (1992) Cross scale morphology, geometry and dynamics of
Textural habitat architecture and body size mediated ecosystems. Ecological Monographs 62: 447 502.
coexistence hypotheses have been proposed to explain Hutchinson GE (1959) Homage to Santa Rosalia, or why are
there so many kinds of animals? American Naturalist 93: 145 159.
the abiotic and biotic components of interaction regarding Lawton JH (1990) Species richness and population abundance of
its influence on the observed body size patterns. animal assemblages. Patterns in body size: Abundance space.
Philosophical Transactions of the Royal Society of London, Series B
330: 283 291.
May RM (1986) The search for patterns in the balance of nature:
Advances and retreats. Ecology 67: 1115 1126.
Further Reading Peters RH (1983) The Ecological Implications of Body Size. Cambridge,
UK: Cambridge University Press.
Brown JH, Gillooly JF, Allen AP, Savage VM, and West GB (2004) Sheldon RW, Prakas A, and Sutcliffe WH, Jr. (1972) The size distribution
Towards a metabolic theory of ecology. Ecology 85: 1771 1789. of particles in the ocean. Limnology and Oceanography 17: 327 340.
Brown JH and West GB (2000) Scaling in Biology. Oxford: Oxford White EP, Ernest SKM, Kerkhoff AJ, and Enquist BJ (2007) Relationship
University Press. between body size and abundance in ecology. Trends in Ecology
Elton C (1927) Animal Ecology. London: Sidgwick and Jackson. and Evolution 22: 323 330.

Cycling and Cycling Indices

S Allesina, University of Michigan, Ann Arbor, MI, USA
2008 Elsevier B.V. All rights reserved.

Introduction Limitations of FCI

Definition of Cycle Number of Cycles in Food Webs
Cycles in Food Webs Finding Cycles in Ecological Networks
Structure of Cycles in Ecological Networks: Strongly Removing Cycles in Ecological Networks
Connected Components Ecological Applications of Cycle Analysis
Quantifying Cycled Fraction: Finns Cycling Index Further Reading

Introduction Definition of Cycle

Given the finite amount of chemical compounds in the A very common way of describing ecosystems is by means
biosphere, it is inevitable that the same material will be of graphs. Graphs are constituted by nodes (representing
utilized repeatedly by different organisms. This phenom species or functional groups of species) connected by
enon is addressed as recycling or simply cycling of energy arrows (or edges, arcs, links, representing relationships
and matter. Familiar examples of recycling of nutrients between species).
involve the so called detritus chain, which decomposes The simplest way of sketching ecosystems using
organic matter that is unusable for some organism to its graphs is the food web representation. In this way of
basic compounds that can be recycled into the grazing chain. drawing species relations, edges connect prey to their
This article provides an overview of cycles and cycling predators (see Figure 1).
indices in ecosystems ecology. Depending on the way of This food web representation can be associated with a
modeling ecosystems, cycles assume different meanings. matrix that expresses the relationships between species.
In what follows, a general definition of cycles, taken from This is the so called adjacency matrix, A. If the row
graph theory will be introduced. Then the concept of species is a food source of the column species then the
cycling is applied to (1) food webs (description of who corresponding coefficient will be 1. More generally, this
eats whom in the ecosystem) and (2) ecological networks relation is a consumerresource relation, as nodes can
(weighted, mass balanced versions of food webs). Simple represent nutrient pools, etc. Elsewhere, the coefficients
ways of computing cycling indices and removing cycles will be 0. The food web in Figure 1 can therefore be
will be provided. represented by this adjacency matrix:
Cycling and Cycling Indices 51

All kinds of paths, other than simple paths, contain at least

one cycle. For example, the graph in Figure 1 contains
just the simple cycle 2 ! 3 ! 4 ! 2. A graph containing
no cycles is said to be acyclic. Every pathway can be
4 3 classified according to its length that is given by the
number of nodes involved.

1 2 Cycles in Food Webs

Figure 1 Example of food web containing five species and Cycles in food webs can be divided into two main classes:
seven feeding relations (arrows, edges). feeding cycles and nonfeeding cycles. The former involve
species and their feeding relations (e.g., species A eats
species B; species B eats species A); cannibalism is a
0 1
0 0 1 1 0 simple kind of feeding cycle. The latter are typical of
B C food webs that comprise detritus compartments and
B0 0 1 0 0 C
B C nutrient pools: organic matter is recycled in the system
A B0 0 0 1 1 C
C via mineralization, creating a huge number of detritus
B0 1 0 0 1 C mediated cycles.
@ A
Feeding cycles are rare in published food webs. This
0 0 0 0 0
is mainly due to the fact that the resolution of food
webs is usually at the species/group of species level.
The adjacency matrix represents direct interactions The number of feeding cycles becomes more significant
between species. These direct interactions, however, when age structured populations are considered, espe
yield chains of indirect interactions. These will be cially in aquatic food webs. Nonfeeding cycles, on the
sequences of nodes and edges that are called paths. We other hand, are extremely abundant in published net
can discriminate between different kinds of paths: works, being several billion cycles for highly resolved
1. Open paths connect two different nodes. They can be ecosystem models.
subdivided into simple paths, containing no repeated
nodes (e.g., A ! B ! C, Figure 2a) and compound
paths, which contain repeated nodes (e.g., Structure of Cycles in Ecological
A ! B ! C ! B ! D, Figure 2b). Networks: Strongly Connected
2. Closed paths start and end at the same node. Also closed Components
paths can be divided into simple cycles, containing
no repeated nodes except the initial one (e.g., Two nodes A and B are said to belong to the same
A ! B ! C ! A, Figure 2c) and compound cycles, strongly connected component (SCC) if they are reach
representing repeated cycles (e.g., A ! B ! A ! B ! A, able from each other, that is to say if we can find a path
Figure 2d, where double arrows mean that the cycle is going from A to B and a path coming back from B to A.
traversed twice). If A and B belong to the same SCC, then they are
connected by cycles. A graph can be divided into its
SCCs, considering every node that is not involved in
cycles as an SCC by itself. Figure 3a represents the
(a) (b) Baltic Sea ecosystem. One can individuate 6 SCCs: 4 of
A B C A B C them are composed by a single node, while 2 of them
comprise more than 1 node (Figure 3b).
If we compact every SCC into a single node, we
(c) (d) produce an acyclic graph (Figure 3c). Further analysis
shows how one component contains just pelagic species
A C A B and the other one just benthic. Acyclic graphs can be
ordered so that all edges point in the same direction
Figure 2 Classification of pathways in (a) simple paths (open (from bottom to top in Figure 3c) using a procedure
pathways start and end at different nodes); (b) compound paths
(open pathways start and end at different nodes, contain
known as topological sort (or partial ordering). Acyclic
repeated nodes); (c) simple cycles (closed pathways start and graphs are therefore intrinsically hierarchical. In this
end at the same node); and (d) compound cycles (same cycle case, the flows find a sink in the benthic compartment,
traversed more than once). while the pelagic compartment acts as a bridge between
52 Cycling and Cycling Indices

(a) (b) (c)

12 Benthic

10 11

9 9 7 2

8 8

15 15 Pelagic

11 2 7 7 2

14 14

5 5

6 6

4 4

3 3



Figure 3 Schematic representation of Baltic Sea ecosystem (a). The boxes define different strongly connected components.
Condensing each box into a single node yields an acyclic graph (b). This graph can be sorted so that all arrows point in the same
direction, showing the underlying straight flow between compartments (c).

the primary producer 1 and the benthic compartment. describes the flow of energymatter from the row
The same structure was found for other aquatic net compartment (i ) to the column compartment (j ). An
works as well. Note that this feature depends drastically example of network and its matrix representation is
on the presence/absence of resuspension of nutrients. If given in Figure 4.
this is negligible, then the network presents several In order to show the computation of the Finns cycling
SCCs. When remineralization is strong, however, the index, it is necessary to introduce the concept of power of
process joins the benthic and pelagic components, thus adjacency matrices. Take as an example the matrix intro
forming a giant SCC. duced in the first section. If we square it, we obtain
0 1
0 1 0 1 2
B 0 0 1 1CC
Quantifying Cycled Fraction: Finns B C
A2 B
B0 1 0 0 1CC
Cycling Index B C
@ 0 1 0 0CA
Ecological networks are food webs where the edges are 0 0 0 0 0
quantified and represent exchanges of nutrients (usually
grams of carbon per m2 per year, but also nitrogen or This matrix shows the pathways of length 2 that connect
phosphorous) or energy. Moreover, inputs to the sys to two nodes. For example, there is just one path connect
tem and outputs from the system are explicitly ing node 1 to node 4 in two steps (the path 1 ! 3 ! 4),
represented by flows involving special compartments while there are two pathways connecting 1 to 5
(i.e., nodes that act as a source (imports) or sink (1 ! 4 ! 5 and 1 ! 3 ! 5).
(exports and respirations) for the system). Besides the In the same way, if we multiply this matrix with the
graph representation, a system can be described using adjacency matrix we get A3, which describes all the path
the so called flow matrix T, where each coefficient tij ways connecting two nodes in three steps; A4 will
Cycling and Cycling Indices 53

(a) 860

2003 Carnivores 203
11 184 Plants

feeders 1814
8881 200




Bacteria 255




0 11184 635 0 0 0 0 0 Imports

0 0 8881 0 0 0 300 2003 Plants
0 0 0 2309 5205 0 860 3109 Detritus
0 0 200 0 0 370 0 1814 Detritus feeders
0 0 1600 75 0 0 255 3275 Bacteria
0 0 167 0 0 0 0 203 Carnivores
0 0 0 0 0 0 0 0 Exports
0 0 0 0 0 0 0 0 Dissipations
Figure 4 Schematic representation of cone spring ecosystem (a). There are two imports (to Plants and Detritus), three exports (from
Plants, Detritus, and Bacteria) and five Dissipations (dashed arrows). The network can be associated with a matrix of transfers (b). The
first row represents imports, the last two columns stand for exports and dissipations, and the internal 5  5 part depicts
intercompartment flows.

similarly contain all the pathways of length 4, and so forth. For example, the G matrix for the network in Figure 4
The power Ax will contain all the pathways of length x. If would be
the food web contains no cycles, then for some x < n
0 1
(where n is the number of species) the matrix will contain 0 0:946 0:054 0 0 0 0 0
just zeros. If the food web contains cycles, on the other B C
B 0 0:794 0 0:027 0:179 C
0 0 C
hand, the powers never converge to 0. The pathways B C
enumerated in these matrices belong to all the different
B 0 0 0:201 0:453 0 0:075 0:271 CC
types that we illustrated in the first section. Now we can B0
B 0 0:084 0 0 0:155 0 0:761 C C
see how these considerations apply to quantified GB C
B 0 0:307 0:014 0 0 0:049 0:629 CC
networks. B C
Dividing each coefficient tij for the row sum produces
B 0 0:451 0 0 0 0 0:549 C C
the coefficients gij (of matrix G), which describe the frac B0
@ 0 0 0 0 0 0 0 C A
tion of flow leaving each compartment:
0 0 0 0 0 0 0 0
Multiplying G by itself, one obtains the fraction of flow
tij leaving the row compartment and reaching the column
gij P
k tik compartment in two steps (i.e., passing by an intermediate
54 Cycling and Cycling Indices

compartment). G3 will describe the exchanges in three steps, A particle entering compartment k will be recycled
and so forth. Summing over all possible powers of G, one lij1 times. The fraction of flow recycled is therefore
obtains the average number of visits a quantum of matter
leaving the row compartment will pay to the column com lkk 1
partment. This computation is made possible by the fact lkk k
that the power series of G converges to the so called
Leontief matrix L. G0 is defined as the identity matrix I: The recycled fraction for Bacteria (fourth compartment)
would be (1.018  1)/1.018 0.0172. The total flow
I G G 2 G 3 G 4    I G  1 L cycled C will be
The Leontief matrix for the network in Figure 4 would be C Rk Tk

0 1 which, computed for the example, will result in

1 0:946 0:946 0:202 0:440 0:031 0:120 0:880
B C 2777.23 units recycled.
B 1 0:958 0:199 0:434 0:031 0120 0:880 C
C The total fraction of recycled flow for the whole sys
B 0 1:207 0:251 0:547 0:039 0:117 0:883 C
tem will therefore be
B 0 0:186 1:039 0:084 0:161 0:018 0:982 C
B 0 0:374 0:092 1:169 0:014 0:085 0:915 C
B 0 0:545 0:113 0:247 1:018 0:053 0:947 C
B C which, for the network in Figure 4, would be 0.0654.
@ 0 0 0 0 0 1 0 C A
0 0 0 0 0 0 0 1 Limitations of FCI

In an acyclic network, the maximum coefficient of L will FCI considers only the diagonal coefficients of the
be 1 (i.e., a quantum of matter can visit another compart Leontief matrix, accounting therefore only for paths start
ment maximum once). This is because a particle of matter ing and ending at the same node.
leaving a compartment will never be recycled to the same Using the notation introduced above, we see that
compartment again. This is not true when cycles are FCI accounts for simple cycles and compound cycles,
present. In fact, when matter cycles in the network, a but does not consider the contribution of compound
particle can be recycled into the same compartment paths, as they never appear on the diagonal.
many times, raising the maximum value of the coefficients Compound paths, however, contain cycles that should
of the Leontief matrix. Therefore, the Leontief matrix of be included in the definition of cycling index.
an acyclic network would contain unitary coefficients on Unfortunately, there is no simple linear algebra tech
the diagonal for all compartments (a particle starting at nique that can account both for cycles and compound
any compartment will never come back). Consequently, a paths, and counting all the pathways in an ecological
simple way of estimating the cycled fraction would be to network is computationally very intense.
see how much these coefficients deviate from 1. This is at As an example of the limitation of the FCI, we see that in
the heart of the so called Finns cycling index (FCI). Figure 4 the pathway Plants ! Detritus ! Detritus
There are various formulations for this index, but here feeders ! Detritus ! Bacteria will not contribute to any
we present the simplest one, adapted from the one devel diagonal coefficient, even if it contains a cycle. Because each
oped in 1980 by J. T. Finn; the reader is referred to the quantum of matter can be recycled into the same compart
Further reading section for a complete account of the ment many times, it will also move around compound paths
possible variations. The following computation is valid many times. This may result in off diagonal coefficients in the
for steady state network only, that is, for networks where Leontief matrix that are greater than 1, stressing the need for
the input to any node equals the output from the same counting compound paths in the cycling process.
We will call Tk the sum of all flows entering the Number of Cycles in Food Webs
compartment k:
In order to quantify the abundance of simple cycles in
Tk tik food webs, one should know the maximum possible num
i ber of simple cycles. The maximum number of simple
cycles will be associated with a completely connected
For example, in Figure 4 the sum of the flows to the food web, that is, a food web whose adjacency matrix
Plants compartment T1 would be 11 184. contains just 1s.
Cycling and Cycling Indices 55

In order to count the maximum number of simple cycles, The total number of cycles is therefore given by the
we start from the ones with maximum length (in graph following formula:
theory they are called Hamiltonian cycles). In a completely !
n Xn n
connected food web composed of n species, the number of TotCycles C k;n k 1!
simple cycles of level (i.e., length) n is (n  1)!. This simple k 1 k 1 k
formula can be explained combinatorically using permuta The first 10 values are represented in Table 1. Note that this
tions: we can see a cycle of level n as a permutation of the n sequence is defined, in combinatorics, as logarithmic numbers.
labels of the nodes: for example, ABCD will represent the
cycle A ! B ! C ! D ! A. Now, the number of permuta
tions of n elements is n!. We note, however, that every cycle Finding Cycles in Ecological Networks
gives rise to n possible sequences (e.g., ABCD, BCDA,
CDAB, and DABC represent the same cycle of length 4). Finding cycles in graphs is a computationally difficult
Therefore, the total number of simple cycles of maximum task. Nevertheless, published ecosystems contain a few
length is n!/n (n  1)!. hundred nodes at most, and the low connectance (fraction
This is an enormous number, as soon as n becomes of realized connections) displayed by these systems
large. For example, in a 100 species food web, we can find ensures that the number of simple cycles is much lower
almost 10155 simple cycles of level n. than the theoretical case illustrated above, where all pos
Now that we know the total number of simple cycles of sible cycles are present.
level n in a completely connected food web, we can easily The idea behind most algorithms for cycle search is
derive the number of simple cycles of level (n  1). For simple: one should construct a path inside the network
each subgraph containing (n  1) species we will have until the same node is found twice. In this case the path is
(n  1)!/(n  1) (n  2)! simple cycles of length (n  1). either a cycle (the initial and final nodes do coincide) or a
The number of possible subgraphs containing (n  1) compound path (initial and final nodes are different).
species is given by the binomial coefficient Of the various possible ways of searching the cycles,
! backtracking based ones, such as depth first search
n (DFS) are surely the easiest to implement.
n 1

Therefore, the total number of simple cycles of level Removing Cycles in Ecological Networks
(n  1) in a completely connected food web composed of
n species is n(n  2)!. We have stated above that it is possible to enumerate all
Similarly, we can define the total number of simple the cycles in a food web. In an ecological network, how
cycles of length k in a completely connected food web of n ever, each cycle will also possess a weight, given by the
species as amount of flow passing through the cycle.
Some network analysis applications (e.g., the so called
n Lindeman spine) require an acyclic network as an input.
C k;n k 1! The removal of the cycles therefore becomes an impor
tant topic for network analysis.
Table 1 represents the number of cycles of level k (column) The current procedure requires the removal of cycles
for a completely connected food web of n species (rows). according to their nexus. Two cycles are in the same

Table 1 Number of simple cycles of length k (column) in a completely connected food web formed by n species (rows)

n 1 2 3 4 5 6 7 8 9 10 Total

1 1 1
2 2 1 3
3 3 3 2 8
4 4 6 8 6 24
5 5 10 20 30 24 89
6 6 15 40 90 144 120 415
7 7 21 70 210 504 840 720 2 372
8 8 28 112 420 1344 3 360 5 760 5 040 16 072
9 9 36 168 756 3024 10 080 25 920 45 360 40 320 125 673
10 10 45 240 1260 6048 25 200 86 400 226 800 403 200 362 880 1 112 083
56 Cycling and Cycling Indices

nexus if they share the same weak arc, defined as the The request for a quantification of cycling was then
smallest flow in the cycle. Cycles are then removed divid answered by the FCI illustrated above. Modified versions
ing the flow constituting the weak arc among all the cycles of the FCI, including biomass storage, utilizing the so called
sharing the same nexus. The resulting amounts are then total dependency and contribution matrices were published,
subtracted from each edge of the cycles. This process increasing the possibilities for modelers and therefore the
results in the removal of the weak arc. The procedure is number of applications of such indices to empirical studies.
then repeated until the resulting network is acyclic. Recently, it was pointed out how all these calculations
A nice by product of the procedure is the creation of a ignore some cycling that involves just off diagonal terms
network composed of all the cycles in the original network. in the Leontief matrix. Unfortunately, in order to com
This is usually referred to as aggregated cycles network in pute the exact amount of cycling in an ecosystem one
ecological literature. This network will receive no input, should utilize a computationally intensive method, which
produce no output, and will be balanced (i.e., incoming is therefore unfit to be applied to large ecosystem net
flows equal outgoing ones) for all nodes. If the resulting works. Fortunately, studies conducted on many small
aggregated cycle network is composed of several sub networks showed that the total amount of cycling and
graphs, each subgraph is a strongly connected component. the FCI seem linearly related, with the total cycling
Note that while some applications require acyclic net being around 1.14 times the FCI.
works, most of them are actually based on the fact that The relation between cycling and maturity of ecosystems
empirical networks contain millions of cycles. As was challenged by the work of Ulanowicz. He showed how
explained in the next section, in fact, cycles are among cycling could be inversely related to the developmental
the most important features of ecosystems. status of an ecosystem, and how perturbations could be
reflected into a higher cycling index. These considerations
suggest that cycling could be seen as a homeostatic response
Ecological Applications of Cycle Analysis to stress: impacts on ecosystems free nutrients from the
higher trophic levels; this freed matter is then recycled into
The recycling of energymatter is an important process that the system by microorganisms, generating cycles at the lower
occurs in every ecosystem. Cycling is believed to be a buffer trophic levels. In this view, responding to stress ecosystem
ing mechanism that allows ecosystems to face shortage of would show a decrease in cycle length and an increase in total
nutrient inflows. This process, however, has been neglected cycling. It is therefore important to know the distribution of
in many theoretical models, which concentrated on commu cycle lengths together with the total amount of cycling in the
nities rather than ecosystems, and which usually comprised ecosystem when one wants to assess the ecosystem status and
just a few species due to constraints of modeling techniques. maturity. Ulanowicz also presented important insights on
Food web ecologists always had an ambivalent attitude cycling as autocatalytic processes. The cycling feature of
toward cycling. For example, the first collection of food ecosystems is at the basis of the views of several authors on
webs published (which contained poorly resolved food webs ecosystem function and dynamics, such as, for example, the
with just a few nodes) showed that cycles are very rare. This work of Patten and colleagues.
was justified by the fact that cycles are likely to destabilize a Another aspect of cycling is represented by the compart
system, because they introduce positive feedbacks. This result mentalization into SCCs. Although ecosystems comprise
was, however, challenged by the discovery of many cycles in myriad interactions, they still can be divided into a few
larger food webs, and the role of cannibalism in age subsystems that are connected by linear chains of energy
structured population dynamics. In recent times, the impor transfers. In several aquatic food webs, SCC analysis shows a
tance of cycles in food webs has been reconsidered, thanks to subdivision into pelagic and benthic components of the
the switch of focus from local stability dynamics toward a ecosystem. This result is, however, dependent on the way
more comprehensive approach to ecosystems persistence and the ecosystem is modeled, with particular emphasis on the
nonlinear dynamics. Moreover, a greater attention has been importance of including several detritus compartments.
devoted to the microbial loop, which, in some aquatic eco Summarizing, cycling is an important aspect of ecosys
systems, receives more than 50% of the primary production, tem dynamics. Although cycles seem to be rare in published
remineralizes it and feeds it back to higher trophic levels. community food webs and models, their number is very
Ecosystem oriented modeling, on the other hand, large when detritus compartments are considered.
included cycles as the very foundation of the discipline. Moreover, it is important to stress that the role of the so
The first clear reference to the importance of cycling in called microbial loop, neglected in studies that concentrate
ecological network comes from the work of Lindeman on larger organisms, can dramatically change the cycling
who, in his seminal paper in 1942, described food webs performance of the system. These considerations lead eco
as cycling material and energy. Odum then included the system ecologists to the formulation of the amount of
amount of recycling as one of the 24 criteria for evaluat cycling in ecosystem networks. The FCI, even though it is
ing if an ecosystem is mature (i.e., developed). a biased count of the cycling in ecosystems, has found wide
Ecological Network Analysis, Ascendency 57

application in ecosystem studies. The problem of measuring Allesina S and Ulanowicz RE (2004) Cycling in ecological networks: Finns
index revisited. Computational Biology and Chemistry 28: 227 233.
the exact amount of cycling in an ecosystem is still an open De Angelis DL (1992) Dynamics of Nutrient Cycling and Food Webs,
problem, as it could be possible to ameliorate the algorithms 270pp. London: Chapman and Hall.
for finding and removing cycles. Finally, the network build Finn JT (1976) Measures of ecosystem structure and functions derived
from analysis of flows. Journal of Theoretical Biology 56: 363 380.
ing process is likely to determine the outcome in terms of Finn JT (1980) Flow analysis of models of the Hubbard Brook
cycling. It would therefore be important to have shared rules ecosystem. Ecology 61: 562 571.
for network building that would result in the comparability Patten BC (1985) Energy cycling in the ecosystem. Ecological Modelling
28: 1 71.
between different networks and ecosystems. Patten BC and Higashi M (1984) Modified cycling index for ecological
applications. Ecological Modelling 25: 69 83.
Ulanowicz RE (1983) Identifying the structure of cycling in ecosystems.
See also: Autocatalysis; Ecological Network Analysis, Mathematical Biosciences 65: 219 237.
Ascendency. Ulanowicz RE (1986) Growth and Development: Ecosystems
Phenomenology. New York: Springer.
Ulanowicz RE (2004) Quantitative methods for ecological network
analysis. Computational Biology and Chemistry 28: 321 339.
Further Reading
Allesina S, Bodini A, and Bondavalli C (2005) Ecological subsystems via
graph theory: The role of strongly connected components. Oikos
110: 164 176.

Ecological Network Analysis, Ascendency

U M Scharler, University of KwaZulu-Natal, Durban, South Africa
2008 Elsevier B.V. All rights reserved.

Introduction Ascendency Applications

Principle of Ascendency Further Reading

Introduction occurring in response to a cause. In ecosystems, it is

believed that no such direct, mechanistic cause and effect
In the search for a nonmechanistical explanation of ecosys behavior exists due to the interaction with other elements
tem behavior and development, Ulanowicz developed the which in turn influence the patterns of cause and effects
theory of ascendency. Direct cause and effect mechanisms, between pairs. Instead of the absolute probability, Popper
as known from the Newtonian world, are believed not to be introduces the term propensity, which describes a bias
sufficient to describe, or predict, the behavior of ecosystems. that events might (not will) happen. Popper therefore
Such mechanisms are inherently reversible and are not seen calls for a measure of such relative or conditional prob
to be sufficient to explain the behavior of single components abilities. Conditional probabilities are denoted by pai jbj ,
(e.g., species) within the context of the ecosystem. and are calculated by dividing the absolute probabilities
Ecosystems are believed to behave and evolve in a nonme p(ai,b j ) by the marginal probability p(ai), or the sum of all
chanistic fashion. The theory of ascendency tries to capture probable effects of one cause (Tables 13). The
this nonmechanistic behavior in a single index, indicative of conditional probability thus describes cause and effect in
ecosystem state and development, and of ecosystem health. the context of other absolute probabilities, considering
that a cause might have more than one effect. This elim
inates the pitfall of disregarding the influence of other
Principle of Ascendency interactions on the one in question. It is, of course, possi
ble to calculate the conditional probability of a
mechanical causeeffect pair, that is, the case of having
Conditional probabilities and ecosystem one cause and one effect. This turns out to be 1, or in
complexity other words, there is certainty that the effect in question
In a mechanistic world, the probabilities of events follow will follow the cause in question.
ing specific causes can be calculated by joint probabilities Since ecosystems are open, not all causes can be
p(ai,bj ). These describe an absolute probability of an effect accounted for. Some of them might originate outside the
58 Ecological Network Analysis, Ascendency

Table 1 Frequencies of joint To quantify growth and development, ecosystems are

occurrences of events (total 60)
portrayed as networks of material or energy exchanges.
These networks of feeding transfers are believed to ade
b1 b2 b3 b4 quately describe an ecosystem. It is assumed that other
significant aspects of ecological systems, such as beha
a1 4 5 7 9 vioral aspects, are in one form or another imprinted on
the amount of energy transferred, through their effect on
a2 2 4 2 1 population size and predator avoidance.
Ascendency describes both growth and development.
6 7 9 4 Growth of the ecosystem is measured as any increase
in total system throughput (TST), which is the sum of
all exchanges within the ecosystem and between the
Table 2 Joint probabilities (p(ai, bj )) and their column/row
system in question and its outside (imports, exports,
sums or marginal probabilities ( p(ai ), p(bj )) respirations). Total system throughput can rise either
by increasing the extent of the system (more species, or
b1 b2 b3 b4 p(ai) by extending ecosystem borders) or by an increased
activity of the system (e.g., during phytoplankton
a1 0.07 0.08 0.12 0.15 0.42 blooms).
p(ai ,bj) a2 Ecosystem development is quantified from the same
0.03 0.07 0.03 0.02 0.15
networks of material exchanges with the help of informa
a3 0.10 0.12 0.15 0.07 0.43
tion theory. In autocatalytic loops, the trend for
transferring material is as follows: those linkages which
p(bj) 0.20 0.27 0.30 0.23 1.00 are most rewarding to the loop will transfer more material
Values are obtained by dividing the number of occurrences (Table 1) than those which are not (compartments have in general
by the total number of observations (60). more than one outgoing link and can thus have pathways
to compartments outside the loop). The latter are not
Table 3 Conditional probabilities
necessarily discarded, but transfer only a small amount
of material. If a quantum of material sits in a compartment
b1 b2 b3 b4 in an autocatalytic loop, then it is therefore more likely to
be transferred along a route with high material transfer
than along a route with low material transfer. The prob
a1 0.33 0.31 0.39 0.64
ability that a quantum of material flows along the highly
p(ai |bj)
frequented routes is, therefore, higher compared to a net
a2 0.17 0.25 0.11 0.07
work where all routes transfer the same amount of
material. Conversely, the probability that a quantum of
a3 0.50 0.44 0.50 0.29 material flows along the less frequented routes will be
Values are obtained by dividing the values in the joint probability lower compared to a network where all routes transfer
matrix by the column sums (p(ai)) (see Table 2). the same amount of material. Such a change in probability
can be quantified with the help of information theory.
Information is defined as the agent that causes a change
system. Therefore, an open ecosystem can never evolve in probability. Ulanowicz uses the term information to
toward a mechanistic behavior of cause and effect. describe the effects of that which imparts order and
Ulanowicz states that autocatalysis, or indirect mutualism, pattern to the system.
is an important cause in ecosystem growth and develop In the calculation of information, the starting point is
ment. Autocatalysis is apparent when members of a to quantify ecosystem complexity. The complexity of a
feeding loop positively enhance the following member system is mirrored in the system configuration (amount of
of the loop, which eventually leads back to a positive links and distribution of transfers along those links).
enhancement of the starting member. Autocatalytic According to Boltzmann, the potential of each configura
loops exert a selection pressure on its members in that a tion contributing to systems complexity, s, can be
member of the loop might be replaced with a new con calculated as the negative logarithm of the probability
stituent who has a more positive effect. Autocatalytic that the event (the system configuration) will occur
loops exhibit a centripetality, which enables them to (s k log p, where k is a constant of proportionality,
attract more resources (available energy). These are rea i.e., a scaling factor). If a system configuration (the event)
sons for the growth and development of, or increase of will occur always (p 1), then the contribution to com
order in, ecosystems. plexity is diminished (log(1) 0, uncertainty is at its
Ecological Network Analysis, Ascendency 59

lowest) and the system behavior is simple (i.e., it always more material flows along some pathways, and less mate
behaves the same way). If a system configuration (or rial along others (Figure 1b). Thus, the most
event) occurs only rarely, then there is a large potential indeterminate network is one where all compartments
for complexity (i.e., it can behave in many different ways, are connected with each other and where, in proportion
uncertainty is high). Behavior of a truly complex system is to the compartmental throughput, equal amounts of
unique each time it functions (uncertainty is highest). material flow along the ingoing and outgoing pathways.
To calculate how much a rare configuration contri Quantifying the information which is gained by transfer
butes to system complexity, it is weighted by the (low) ring material along more and less frequented routes thus
frequency of its occurrence. The potential contributions gives a clue about the unevenness of material flowing
(or events) are averaged by the configurations of the along pathways.
system by weighting each si by its corresponding The change in probability from a situation where a
pi (Shannons formula: H K pi log pi ). In other quantum of material flows along an equiprobable pathway
words, each potential contribution of occurrence is and along a pathway which is not equiprobable is calcu
weighted by its corresponding probability that it will lated using conditional probabilities. To start with, the
occur, which is summed over all system configurations. uncertainty that an event occurs is
A high H corresponds to high uncertainty, complexity,
and diversity. H K log pai 1

and the uncertainty that an event occurs provided certain

information (bj) is available is
Average mutual information
The above discussion serves to illustrate how events can H K log p ai jbj 2
contribute to the complexity of a system. Next to consider
is whether these events contribute to an ordered pattern The information then is the a priori uncertainty minus the
in the system, or whether they contribute to random uncertainty if bj is known or
behavior. If all events are equiprobable, then the average   
uncertainty about what event will happen next is the I K log pai k log p ai jbj 3a
highest. This hypothetical situation can serve as a starting
point to calculate how much less uncertainty there is
under circumstances where not all events are equiprob  
I K log p ai jbj K log pai 3b
able. The decrease in uncertainty from a situation of
equiprobability to any other is called information. From or
an ecosystem perspective, a situation of equiprobability is   
one where material flows in equal amounts along all path I K log p ai jbj =pai 3c
ways (Figure 1a). One that is not equiprobable is where
I is not positive for all pairs of occurrences. The sum of all
I s which have been weighted by the corresponding joint
(a) probability turns out always to be positive, however. The
A joint probability of each occurrence serves, as in
12 12 Shannons formula, as a weighting for the frequency of
12 12
occurrence of each event (i.e., each co occurrence of ai
and bj). The result is called the average mutual informa
12 tion (AMI) or
12 XX     
AMI K p ai ; bj log p ai jbj =pai 4
i j

AMI is the amount of uncertainty reduced by knowing bj .
23 1
Results are in units of K.
As in the hypothetical example above, the a priori
1 23 uncertainty about where a quantum of material flows
23 in ecological networks is given by Shannons formula.
C B The additional information (bj) to calculate the condi
1 tional probability is the knowledge of the outputs from
Figure 1 (a) Hypothetical unconstrained network: low AMI. each compartment in the flow network a time step
(b) Hypothetical constrained network: higher AMI. earlier.
60 Ecological Network Analysis, Ascendency

Since, from an ecological network point of view, joint ascendency. Mutualism is furthermore not a result of
and conditional probabilities refer to transfers of material events elsewhere in the systems hierarchy but can arise
from compartment i to compartment j, the above formula at any level. Therefore it is theorized that in the absence
can be rewritten as of overwhelming external disturbances, the ascendency of
a system has a propensity to increase, that is, both activity
X Tij  Tij T::  (TST) and structure (AMI) increase. The theoretical
AMI K log 5
i; j
T :: Ti: T:j behavior of mutual information conforms to most of the
24 ecosystem properties originally put forward by Odum
where the joint probability of a quantum of material to characterize mature ecosystems.
(p(ai,bj)) flowing from species i to species j can be denoted Ascendency is limited by any constraints on the
as Tij/T.., remembering that the events in an events table increase in either TST or AMI. Limits to TST are set
are material flows in a system. T.. is the total system by the finite imports from outside system boundaries and
throughput, or the sum over all combinations of Tij . by the second law of thermodynamics, which requires that
The summation among all rows of the matrix is denoted a portion of the compartmental throughput be lost as
by the first dot, while the second dot stands for summation dissipation. Therefore the TST cannot increase indefi
among columns. nitely via recycling. The limits to the AMI, or system
The conditional probability development, are set by the flow structure. It limits the
extent to which the flows can be organized without a
p ai jbj p ai ; bj =pai change to the structure itself. Further limits to the AMI
in real networks are discussed in the section titled
can be rewritten as Tij/Ti. and the marginal probability Overhead.
(sum of all probable outcomes, p(ai)) as T.j/T... In theory, ascendency is higher when pathways are
To summarize, the AMI describes the information fewer in numbers (more specialization) and more articu
gained by knowing the outputs from each compartment lated (few pathways transport most of the material). The
in the flow network a time step earlier (bj) in addition highest theoretical value of ascendency is achieved when
to the a priori situation describing the flow of a quan all players in the system have one input and one output
tum of energy or material between two compartments only, and are thus joined in one big single loop. This
(ai). The uncertainty of where a quantum flows is configuration mirrors highest specialization, and in this
calculated through Shannons index of flow diversity. case AMI H (diversity of flows, see below). This situa
The uncertainty of where a quantum of material will tion cannot be achieved in real systems, due to reasons
flow by knowing bj is calculated by the conditional discussed in the section titled Overhead.

Ascendency Development Capacity

The scalar constant, k, has been retained throughout all As mentioned above, the limit to development is set by
calculations. To be able to combine growth and develop Shannons diversity index pertaining to the material
ment into one single index, k is substituted by the total transfers or flows. MacArthur applied Shannons diversity
system throughput or TST in order to scale the AMI to index to the material flows in an ecosystem to arrive at a
the size of the system in question. The resulting index is measure for the diversity of flows, H:
called ascendency and is denoted by
XTij  Tij 
XTij  Tij T::  H k log 7
A TST log 6a i;j
T:: T::
T:: Ti: T:j
where k is a scalar constant, and T.. is the TST, or the sum
or over all combinations of Tij .
X   H can, like the AMI, be multiplied by TST to scale the
Tij T::
A Tij log 6b diversity of flows to the system in question. TST  H is
Ti: T:j
i;j called the development capacity, or limit for develop
ment, C:
Besides indirect mutualism there are a number of
influences that can change the ascendency of a system. XTij  Tij 
C TST log 8a
These influences are thought to not have any favored i;j
T:: T::
direction of change, whereas indirect mutualism is
believed to drive development toward increased or
Ecological Network Analysis, Ascendency 61

Tij only one import path, then the overhead due to imports
C Tij log 8b
T:: is minimal and equals zero. From a systems point of view
it is regarded as counterproductive to minimize the mag
The development capacity is limited by two factors, nitude of the import, or to import only via one pathway.
namely TST and the number of compartments. The The insurance lies in being able to receive imports via
limits to TST are the same as in the case of ascendency. several pathways in case one is lost. In the case of
If a certain amount of TST is split between too many increased recycling within the system, the imports will
compartments, then some compartments will end up with occupy a smaller and smaller part of the TST. In this case,
a very small throughput. These are, in turn, prone to the development capacity will rise faster than the over
extinction should the system undergo disturbances. This head on imports.
process is believed to reduce the number of compart If the imports enter the system via fewer pathways or
ments and thus the number of flows. More stable compartments, then the ascendency will increase at the
systems are thus believed to show a higher C compared expense of the overhead. Systems are expected to pro
to systems undergoing frequent perturbations. gress toward fewer import pathways. The number of such
The initial complexity, H, consists of two elements. pathways can be changed should those links be disrupted
One is the AMI, describing the information gained by and others become necessary. Overall it is expected that
reducing the uncertainty in flow probability. It is an systems in a more stable environment rely on fewer
index of the organized part of the system. The other is import pathways compared to perturbed systems.
the residual uncertainty, or Hc (also called conditional The formula for the overhead on imports is as follows:
diversity). Thus, H AMI Hc. !
n T0j2
I T0j log 11
j 1
T0: T:j

Hc or Overhead where imports are assumed to originate in the fictitious

The residual uncertainty Hc, when scaled by TST is also compartment 0.
called the overhead. The overhead represents the unor
ganized, inefficient, and indeterminate part of the flow
structure and is considered an insurance for the system. Exports
Should the system become overly organized (high ascen Similar to the overhead on imports, the overhead on
dency), it will also be prone to perturbations. The exports depends on the amount of exporting pathways
overhead is split into four components: overhead due to leaving the system and the amount transferred along
imports, exports, respiration, and internal pathways. those pathways. The overhead due to export diminishes
The combined overhead is denoted by whenever there are fewer export pathways, lower magni
tude of transfers, or an uneven distribution of amounts
XTij  Tij2 transferred along the pathways. An increase in exports
Hc k log 9 becomes beneficial to the system whenever there is posi
T:: Ti: T:j
tive feedback via another system. The overhead on
Scaling Hc to the system by replacing k with TST exports is denoted by
yields X
n  2 
! E Ti;n1 log 12
X Tij2 i 1
Ti: T:;n1
 Tij log 10
Ti: T:j where exports are assumed to flow into a fictitious com
partment n 1.
The relationship between C, A, and  so becomes
C A: .
Imports Again, the overhead regarding the dissipations depends
The overhead due to imports is dependent on the number on the magnitude lost to the environment, on the number
of pathways originating outside the system, and on the of pathways, and the distribution of the magnitude trans
magnitude of the material transferred along those path ferred. Losses through dissipation are required by the
ways. If all sustenance is equally distributed among all second law of thermodynamics and are necessary to main
import pathways, then the contribution to the overhead tain metabolisms. The overhead on dissipation is
will be maximal. It will decrease when some pathways X
n  2 
import more and others less. It will also decrease if the D Ti;n2 log 13
overall magnitudes of the imports decrease. If there is i 1
Ti: T:;n2
62 Ecological Network Analysis, Ascendency

where respiration is assumed to flow into a fictitious or

compartment n 2.    
Tij Bi Bj
IB k log K log 15b
T:: B2
The fourth part of the overhead is that of internal trans or
fers and represents the extent of pathway redundancy.  
There are disadvantages to the system in maintaining Tij B 2
IB K log 15c
redundant, or parallel pathways. For one, there can be T:: Bi Bj
an increase in dissipations, whenever transfers occur not
only along the most efficient route, but also along leakier Summing over all realized combinations of i and j and
pathways. Also, the resource transferred along different weighted by the joint probability of occurrence, one
parallel pathways might not always end up at the right arrives at the biomass inclusive AMI, AMIB:
time at the consumer.  
An obvious advantage of parallel pathways is the insur X Tij Tij B 2
AMIB K log 16
ance of having more than one route of transfer in case of i;j
T:: T:: Bi Bj
disturbances of other routes. Redundancy is denoted by
! which is also called the KullbackLeibler information.
n X
X n Tij2
R Tij log 14 Scaling by the total system throughput gives the biomass
i 1 j 1
Ti: T:j inclusive ascendency, AB:
X Tij  
Tij B 2
AB TST log 17a
Biomass Inclusive Ascendency i;j
T:: T:: Bi Bj

The above indices were calculated on the trophic flows or

between compartments. It is also possible to calculate a
systems ascendency that embraces the connection Tij B 2
AB Tij log 17b
between biomass stocks and the trophic flows. This bio i;j
T:: Bi Bj
mass inclusive ascendency can be used as a theoretical
basis to derive element limitations for compartments, to AB is sensitive to changes in biomass and can thus show
identify limiting nutrient linkages, and to quantify the the sensitivity of the whole system to changes in stock of a
successional trend to include larger species with slower particular compartment.
turnover times. The above term can be split into the following terms:
Above, AMI was calculated as the difference between
two flow probabilities, the unconstrained or a priori joint X   X  
Tij T:: Ti: B
probability, and the constrained or a posteriori conditional AB Tij log Ti: log
Ti: T:j T:: Bi
probability. AMI can also be calculated between a bio i;j
X   i
T:j B
mass (unconstrained or joint) probability and the resulting T:j log 18
flow (constrained or conditional) probability, thereby j
T:: Bj
calculating a relationship between biomass and flows.
From the principal of mass action, the joint probability The first term is exactly the same as in the above
of whether a quantum of biomass leaves compartment i definition of the flow ascendency. Therefore, also the
(Bi/B) and enters compartment j (Bj/B) is BiBj/B 2. This biomass inclusive ascendency rises with an increased
expression constitutes the unconstrained joint probability number of compartments, increased specialization of
that a quantum flows from i to j. No constraining assump flows, and increased throughput. The second and third
tions are made about this exchange, with the exception of terms become zero whenever the proportional flow
the magnitudes of the stocks. The corresponding con through each compartment is the same as its proportion
strained distribution is taken as the conditional of the biomass. Only in this case would AB equal A. In all
probability of the actual flow from i to j or Tij/T. This other cases, AB will exceed A.
constraint is an addition to the probability calculated from
the stocks only, and therefore, structure and function are
tied together. The information gained is calculated as Limiting elements in compartments and limiting
follows: flows
     If one is interested in calculating a compartments
Bi Bj Tij contribution to the ascendency of a particular element k
IB K log K log 15a
B2 T:: (e.g., C, N, P, S, . . .) during a certain time step l, then one
Ecological Network Analysis, Ascendency 63

has to substitute into above equation the element and the value of ascendency were the decomposition activity and
time step: the capacity for resource exploitation. Ascendency was found
X   to be a useful indicator for the health assessment of marine
Tijkl B 2
AB Tijkl log 19 benthic ecosystems over space and time.
i;j ;k;l
T:: Bikl Bjkl Ascendency has also been applied to establish ecosys
tem responses to eutrophication and other anthropogenic
where Tijkl is the flow from i to j of element k during time
system alterations of carbohydrates, proteins, lipids, and
step l.
carbon biopolymers in various parts of the globe.
To show how the ascendency responds to turnover
times of various elements, the differential of AB regarding Whereas ascendency is, in general, believed to rise with
compartment p is given as eutrophication due to an increase in TST, this is not
always the case. Depending on the extent and frequency
qAB T:: 1 T:pk Tp:k of the eutrophication event, it might disturb the system to
2 20
qBpk B: 2 Bpk an extent where ascendency reflects a decrease in ecosys
tem stability through a decrease in AMI and TST. Another
Here the relative contributions of all elements investi
case of system perturbation was described for pesticide
gated to the systems ascendency can be calculated. Results
perturbed microcosms, using an index called scope for
will show that the system is most sensitive to the element
change in ascendency (SfCA). SfCA is an analogy to
with the slowest turnover rate. The element with the slow
scope for growth of an organism and is the balance of the
est turnover rate is also the element which enters the
ascendency of individual compartment inputs and outputs.
compartment in its least relative proportion. The last state
SfCA was hypothesized to decrease in the presence of a
ment accords with Liebigs law of the minimum for which
disturbance and was ultimately found to be a useful indi
ascendency provides a theoretical basis. The same results
cator for the short term assessment of perturbations in
could have been obtained by comparing elemental turnover
herbicide treated microcosms.
rates of all compartments. However, ascendency provides
Ascendency has also been used to assess the whole
yet another level of information, namely it identifies which
ecosystem impacts of severe freshwater abstractions from
source provides the limiting flow of the controlling element.
an estuarine catchment. The interdecadal comparison
To calculate this, the sensitivities of the individual bio
between light and severe freshwater abstraction and the
masses can be expanded to include the sensitivities of the
consequential reduction in sustained and pulsing freshwater
individual flows from source r to predator p. The following
inflow into the Kromme estuary revealed a decrease in
equation calculates the contributions of each flow:
ascendency under the present, freshwater starved condi
qAB Trp B 2 tion. The spatial comparison with other, similar, estuaries
log 21 that do not have such severe freshwater abstractions in the
qTrp T:: Br Bp
catchment shows a higher ascendency in estuaries with
The limiting source of the controlling element is the higher freshwater inflow that ensures sustained renewal of
one which is depleted fastest in relation to its available the nutrient pool to fuel primary production.
stock, that is, the one with the highest (Trp/Br). Knowing Since ascendency is very often influenced by a change
the sensitivity of the flow for each element and compart in the magnitude of TST, the organization of a system
ment, it is thus possible to pinpoint nutrient limitations is frequently reported as a ratio of ascendency/develop
and the limiting flows for each compartment in the food ment capacity (A/C), which cancels out the influence of
web. In ecosystems, not all species are limited by the same TST. Also the AMI is used as an unscaled index in a
nutrient. For instance when primary producers are lim comparative way. In general, it is advised to take the
ited by nitrogen, it does not necessarily mean that the behavior of other indicators of ecosystem health (e.g.,
entire food web is limited by nitrogen. exergy) into account in combination with ascendency to
arrive at a representative assessment of ecosystem state.
Ascendency has been shown to vary with the degree of
Ascendency Applications aggregation of the network. In general, ascendency
decreases in highly aggregated networks, even if the
Principles of ascendency, as they have been shown here, have TST is the same. The type of aggregation, that is, which
been applied to compare similar ecosystems (e.g., estuaries), compartments are aggregated, also significantly affects the
or the same ecosystems over a period of time including the value of ascendency. This is equally true for the aggrega
response of systems to disturbances. Examples of such appli tion of living and nonliving components of the network.
cations are the description of spatial and temporal change of The biomass inclusive version of ascendency and the
ascendency in marine microbial systems. They revealed that sensitivities of the individual flows were determined for
ascendency is strongly related to the functionality of the the Chesapeake Bay system to identify the limiting nutri
microbenthic loop. Important parameters determining the ent in the ecosystem and bottlenecks in carbon, nitrogen,
64 Ecological Network Analysis, Energy Analysis

and phosphorus transfers. The comparison over four sea Ulanowicz RE (eds.) Aquatic Food Webs, vol. 7, pp. 73 85. New
York: Oxford University Press.
sons revealed that, in general, the primary producers were Odum EP (1969) The strategy of ecosystem development. Science
nitrogen limited, which was in concordance with previous 164: 262 270.
studies on these groups. However, the nitrogen limitation Patrcio J, Ulanowicz RE, Pardal M, and Marques J (2006) Ascendency
as ecological indicator for environmental quality assessment at the
on the primary producer level was not propagated ecosystem level: A case study. Hydrobiologia 555: 19 30.
throughout the entire web, but all nekton was found to Popper KR (1982) A World of Propensities, 51pp. Bristol: Thoemmes.
be phosphorus limited. The type of nutrient limitation Rutledge RW, Basore BL, and Mulholland R (1976) Ecological stability:
An information theory viewpoint. Journal of Theoretical Biology
changed over the course of the year for a few primary 57: 355 371.
producers and invertebrates, but not for the nekton. It is Scharler UM and Baird D (2005) A comparison of selected ecosystem
attributes of three South African estuaries with different freshwater
important to note that nutrient limitations in a trophic
inflow regimes, using network analysis. Journal of Marine Systems
flow network are not determined by the type of limitation 56(3 4): 283 308.
of the primary producer, since the various organisms have Tobor Kaplon MA, Holtkamp R, Scharler UM, Bloem J, and de
Ruiter PC (2007) Evaluation of information indices as indicators of
different stoichiometric requirements. environmental stress in terrestrial. Ecological Modelling 208: 80 90.
Ulanowicz RE (1986) Growth and Development: Ecosystems
Phenomenology. New York: Springer.
Ulanowicz RE (1997) Ecology, The Ascendent Perspective. New York:
See also: Autocatalysis; Emergent Properties; Goal Columbia University Press.
Functions and Orientors; Indirect Effects in Ecology. Ulanowicz RE (2004) Quantitative methods for ecological network
analysis. Computational Biology and Chemistry 28: 321 339.
Ulanowicz RE and Abarca Arenas LG (1997) An informational synthesis of
ecosystem structure and function. Ecological Modelling 95: 1 10.
Further Reading Ulanowicz RE and Baird D (1999) Nutrient controls on ecosystem
dynamics: The Chesapeake mesohaline community. Journal of
Baird D and Heymans JJ (1996) Assessment of the ecosystem changes Marine Systems 19: 159 172.
in response to freshwater inflow of the Kromme River estuary, St.
Francis Bay, South Africa: A network analysis approach. Water SA
22(4): 307 318.
Fabiano M, Vassallo P, Vezzulli L, Salvo VS, and Marques JC (2004)
Temporal and spatial changes of exergy and ascendency in different Relevant Websites
benthic marine ecosystems. Energy 29: 1697 1712.
Genoni GP (1992) Short term effect of a toxicant on scope for change in Dipartimento di Scienze Ambientali.
ascendency in a microcosm community. Ecotoxicology and
Environmental Safety 24: 179 191. Ecopath with Ecosim.
MacArthur R (1955) Fluctuations of animal populations, and a measure Ecosystem Network Analysis.
of community stability. Ecology 36(3): 533 536. National Oceanic and Atmospheric
Morris JT, Christian RR, and Ulanowicz RE (2005) Analysis of size and
complexity of randomly constructed food webs by information Administration, Great Lakes Environmental Research
theoretic metrics. In: Belgrano A, Scharler UM, Dunne JA, and Laboratory.

Ecological Network Analysis, Energy Analysis

R A Herendeen, University of Vermont, Burlington, VT, USA
2008 Elsevier B.V. All rights reserved.

Introduction Indicators in Dynamic Systems

Level of Analysis Applications
Steady-State Analysis: Energy and Nutrient Intensities Further Reading
Steady-State Analysis: Other Indicators

Introduction oil. And economists have shown that demand for a shirt
produces a demand for steel. These are all examples of
Ecologists have long told us that all flesh is grass, which in indirect effects. The techniques used to quantify them
turn is sunlight. Thanks in large part to the oil embargo in span systems ecology, engineering, and economics a
1973, we appreciate that bread is not just sunlight, but also compelling example of cross disciplinary fertilization.
Ecological Network Analysis, Energy Analysis 65

Understanding them yields insights in diverse applica 3. Switching from throwaway to returnable beverage bot
tions, from bioaccumulation of pollutants in ecosystems tles saves energy and increases jobs.
to labor demand in economies.
A more recent example is that suburban living (sprawl) is
In principle, one could discern all aspects of indirect
only c. 10% more energy intensive than urban (compact)
ness from the full diagram of flows between
living. A biological example is the trophic cascade, exem
compartments in a system. In practice, we often desire,
plified by consequences of recent wolf reintroduction into
or accept, summary variables or indicators which are
Yellowstone National Park. Adding wolves has sup
specific to a particular application and convey the con
pressed elk activity, resulting in increased regeneration
cept more concisely, though often with a loss of details.
of browse vegetation.
In this article, such indicators are discussed, often
using explicit calculations applied to a simple, idealized
two compartment ecosystem. The indicators are energy
and nutrient intensities, trophic position (TP), path Steady-State Analysis: Energy and
length (PL), and residence time. Besides application to Nutrient Intensities
steady state, the concept is also extended to dynamic
ecosystems such as those responding to perturbations. The bookkeeping of energy analysis can be used to allocate
Finally, calculating energy intensity of goods and services many other kinds of indirectness. Starting with energy, we
in economic systems is discussed. The latter is a crucial extend the method to other entities. To illustrate this, a
step in determining the energy cost of living in a con hypothetical two compartment system at steady state is
sumer society, and has specific application in analyzing used (Figure 1). This is complex enough to allow feedback
consequences of an energy tax. (recycling), yet simple enough to allow using standard
algebra. All that is done here can be couched in matrix
notation, and shorthand is useful for systems with many
Level of Analysis compartments, but algebra is more transparent.
Figure 1 shows the flow of something, say energy, in a
In a multicompartment system, interactions can be ana two compartment system containing producers (e.g.,
lyzed at three levels of aggregation/detail: green plants) and consumers (e.g., herbivores). (See
Table 1 for definitions of terms.) The input to producers
1. Single compartment (isolated). This addresses direct comes from outside the system (the Sun), and the export
effects (e.g., an eagle eats mice but no grass). from consumers leaves the system. There is a variable
Traditional population biology often works at this amount of feedback from consumers to producers; it is
level, which includes no indirectness at all. possible for some plants to eat animals. In all diagrams of
2. Single compartment in system. This addresses direct plus this type we decide which flows convey the direct and
indirect effects (e.g., by eating mice which eat grass, the indirect influences we deem important; our judgement is
eagle is consuming embodied grass, which is embodied required. For example, the standard energy intensity con
sunlight). cept is that the energy losses (e.g., low temperature heat)
3. Whole system. This addresses system wide processes are assumed to be embodied in the remaining flows of
(e.g., to what extent is the entire system recycling high quality metabolizable biomass. This would give a
phosphorus vs. leaking it immediately?). modified diagram (Figure 2).
This article concentrates on level 2. The indicators cal The remaining flows thus convey the input, which we
culated are the property of a single compartment wish to account for. The missing losses are now implicitly
explicitly connected to other compartments in an eco embodied in the flows that remain, and the energy inten
system. Level 3 is beyond the scope of this article. sities carry this formally. The parallelism and difference
Level 2 analysis is the basis for most of the energy
analysis started in the early 1970s. It led to then surpris FEEDBACK EXPORTc
ing results such as these:
1. Only c. 60% of the energy to own and operate a car is INPUTp Producers Consumers
the fuel in the tank. Around 15% is required to pro INPUTc
duce the car, and c. 25% is for parts, maintenance,
insurance, registration, parking, etc. LOSSc
2. Only c. 10% of the energy to make the car is consumed
at the assembly plant. The remainder is consumed at Figure 1 Energy flows (cal/day) in a two-compartment system
the steel mill, glass works, iron mine, rubber plantation, at steady state. INPUTp is gross primary production of biomass
etc. by photosynthesis.
66 Ecological Network Analysis, Energy Analysis

Table 1 Definitions of terms

Symbol Definition Unitsa

tj Time step day

Ej Energy input to compartment j cal/day
"j Energy intensity of compartment js output cal/cal
ECOL Energy cost of living for a household Btu/yr
<"> Household average energy intensity ( ECOL/Y) Btu/$
"impj Energy intensity of imported material functionally identical to that produced by compartment j cal/cal
EXPORTj Export from compartment j cal/day
FEEDBACKj Flow from consumers to producers cal/day
gloof Generic term for system input which is allocated by method used in this paper ?/day
GPP Gross primary production, the energy fixed by plants cal/day
IMPORTj Import of material functionally identical to that produced by compartment j cal/day
INPUTj Input to compartment j cal/day
LOSSj Loss from compartment j cal/day
Nj Nutrient input to compartment j g/day
j Nutrient intensity of compartment js output g/cal
impj Nutrient intensity of import of material identical to that produced by compartment j g/cal
OUTPUTj Output of compartment j cal/day
PLj Path length of compartment j dimensionless
Sj Stock of compartment j cal
TPj Trophic position of compartment j dimensionless
tj Isolated-compartment residence time for compartment j day
j In-system residence time for compartment j day
Xij Flow from compartment i to compartment j cal/day
Xj OUTPUTj cal/day
Yi Annual household expenditure for consumption category i $/yr
Y Sum of all annual household expenditures $/yr
Z FEEDBACK cal/day
Grams, calories, and days are arbitrarily chosen as the units of mass, energy, and time for the examples in this article. Other units, as appropriate, are
used for the applications.

FEEDBACK * c EXPORTc * c gloof is used in Figure 3a. Figure 3b shows the balance
equation for energy. Figures 3a and 3b allow for imports
to inject embodied energy into the system. An example is
INPUTp Producers Consumers Howard Odums study of Silver Spring, Florida, where
tourists fed bread to fish whose normal food was plants
Figure 2 Embodied energy flows (cal GPP/day) in a two- growing in the spring. Another example of imported embo
compartment system. The energy intensities " (cal GPP/cal) died energy is Americas importing of clothes made in
convert energy flows (cal/day) of Figure 1 to embodied energy China.
flows (cal GPP/day). The energy intensities are obtained by solving the
(linear) equations implied by Figure 3b:
between Figures 1 and 2 crystallize the entire import of
indirectness presented in this article. The intensities are "i Xij "impj IMPORTJ Ej "j Xj 1
thus the conceptual, dimensional, and numerical bridge i 1
between system input and flows. In this system, the units
The sum is over all within system inputs. In Figure 3b E
of intensities will be calories of gross primary production
is the energy from the Earth, or at least external to the
(i.e., the sunlight fixed by photosynthesis, abbreviated
system. The implicit energy intensity of energy itself is
GPP) per calorie of producer biomass or consumer bio 1.0. Imported materials already have an energy intensity
mass, but diverse units are possible depending on the because of their production elsewhere. This is discussed
system and the question asked. For example, in economic in more detail in the section on dynamic indicators. Let us
systems, energy intensity is measured in Btu/dollar apply this to a specific set of flows shown in Figure 4a.
(1 British thermal unit 1055 J). For conciseness, feedback will be denoted by Z.
The key assumption is that embodied energy is Equation [1] gives one equation for each compartment:
conserved in every compartment: embodied energy in
embodied energy out. The general process is summarized
100 Z"c 10"p
in Figures 3a and 3b. Because the method can be used to 2
allocate many things besides energy, the generic term 10"p 5 Z"c
Ecological Network Analysis, Energy Analysis 67

(a) Z c 5 c

In-system: Output:
gloof embodied in j gloof embodied
incoming in-system in output flow Producers Consumers
flows 10 P
Source: 1
gloof generated
internally plus that (5 Z )/2 c
embodied in import Figure 5 Embodied nutrient flows (g/day) in system of
flow Figure 4a.  (g/cal) are the nutrient intensities.

iXij j i Xj
i = in-system inputs 2Z /(15 Z ) 10/(15 Z )

Ej + impj IMPORTj
Producers Consumers
Figure 3 (a) Generic scheme for allocating the embodied
(15 + Z )/(15 Z )
generic system input gloof for steady-state system. Gloof may be 1
energy, nutrient, or even time (for calculating residence time). In
economic applications, gloof may be money, labor, or pollution
(5 Z )/(15 Z )
assimilation. For each compartment j, embodied gloof
in embodied gloof out, by assumption. (b) Embodied energy Figure 6 Embodied nutrient flows (g/day) in system shown in
balance equation for compartment j, which follows from (a). Xij Figure 4a as a function of feedback, Z. Both compartments are in
are in-system inputs to j; Xj is js total output; Ej is the (direct) embodied nutrient balance, as is the entire system.
energy input to j.

balance, a consequence of the assumption that each indi

(a) vidual compartment is in balance.
Z Now suppose we want to track the flow of nutrient.
5 Nutrient intensities () will be expressed in grams of nutri
ent/cal of biomass. Here we will (arbitrarily) assume that
100 Producers Consumers
nutrient is taken up only by producers and passed on with
10 out loss to consumers, but that some nutrient is leaked from
consumers during metabolism. Specifically, assume that the
90 + Z 5Z
nutrient intensity of the consumer metabolic loss is half that
of the consumer export flow. This is shown in Figure 5.
The balance equations are then:
20Z 100 1 Zc 10p
5 Z 3
10p 5c Zc c
100 Producers Consumers
10(10 + 2Z ) which are solved to yield p (1/10)(15 Z)/(15  Z),
Figure 4 (a) Explicit energy flows (cal/day) for a two-
c 2/(15  Z). Using these intensities yields Figure 6
compartment system. Feedback, Z, can vary between 0 and 5 for embodied nutrient flow.
cal/day. Example is arbitrary, but the output/input ratio of 0.1 for The system is in embodied nutrient balance, but the
producers is appropriate for many real systems. (b) Embodied flows have a surprising feature: for Z > 0, the nutrient flow
energy flows (cal GPP/day) in system shown in (a) as a function of from producers to consumer (an internal flow) exceeds
feedback, Z. Both compartments are in embodied energy
balance, as is the entire system.
1 g/day, the systems input flow. Critics have called this
apparent contradiction a damning flaw of the method, but
actually it is to be expected. Feedback speeds up a sys
The solution is "p 10 2Z, "c 20, both expressed in tems flows: more molecules pass a given point per unit
cal GPP/cal biomass. Substituting these energy intensi time. Because here embodied nutrient is actual nutrient,
ties in Figure 3b gives the embodied energy flows shown the effect could be measured experimentally. This vali
in Figure 4b. The entire system is in embodied energy dates the method generally.
68 Ecological Network Analysis, Energy Analysis

25 0.25 An example is American household electronics, most of

Energy intensity (cal GPP/day)

Cons. energy intens.

which are made abroad. In Table 3, the specific equations

Nutrient intensity (g cal)

20 0.2 for the example system in Figure 4a and their solutions
Cons. nutr. intens. are listed.
15 0.15
Prod. nutr. intens.

10 0.1
Prod. energy intens.
Trophic Position
5 0.05
Trophic levels apply to a linear chain picture of feeding
0 0 patterns: A eats nothing but B; B eats nothing but C, etc. If
0 1 2 3 4 5 there are n compartments in the chain, then there are n
Feedback, Z integral trophic levels, and trophic level is the number of
Figure 7 Energy and nutrient intensities vs. Z. steps from the Sun 1. Thus for producers and consu
mers in a chain, trophic levels 1 and 2, respectively.
With omnivory and the resulting web interactions, this
Energy intensities and nutrient intensities as a function view breaks down unless nonintegral TPs are allowed.
of feedback, Z, are shown in Figure 7. Simply put, a compartments TP is the (energy) weighted
As feedback increases (and hence loss from consumers average of the TPs of each of its inputs plus 1. Caution:
decreases), the two intensities approach equality. When trophic interactions are always expressed in energy flows,
Z 5 cal/day, consumers have no losses and the two so here one must use energy flows only, not nutrients or
compartments have functionally merged. other flows. (There is also a dual approach, which results
in an infinite series of integral trophic levels, which is not
covered here.)
Steady-State Analysis: Other Indicators The standard convention of setting TPSun 0 is used.
From Table 3, the TPs are 1 2Z/100 and 2 2Z/100
Below are discussed three other indicators: TP, PL, and for producers and consumers, respectively. Feedback
residence time. In Table 2, the equations for each are increases TP of both. For Z 0, TPp 1, TPc 2, as
listed. The possibility of imports is explicitly allowed for. expected for a straight food chain.

Table 2 Explicit forms of the input and outputs in calculating several indicators for compartment j in steady-state analysis. For every
indicator, the equation to solve is Col A Col B Col C

B. Source term: C.
Internal or Output
Indicator A. In-system inputs term imported inputs term Comment
Energy "i Xij "impj  IMPORTj Ej "jXj Flows can have different units for different
intensity i in-system inputs compartments. Intensities will
(") P correspondingly have different units
Nutrient i Xij impj  IMPORTj Nj jXj Flows can have different units for different
intensity i in-system inputs compartments. Intensities will
() P correspondingly have different units
Trophic TPi Xij TPimpj IMPORTj 1 TPj For trophic position, all flows must be in
i in-system inputs
position P energy terms
i in-system inputs
Path PLi 1Xij None PLj Path length is almost the same as trophic
i in-system inputs
length P position. Flows need not be energy but
(PL) Xij IMPORTj Ej must have the same units for every
i in-system inputs
Residence i Xij tj j tj is the isolated compartment residence
i in-system inputs
time () P time for compartment j ( stock/
Xij IMPORTj Ej throughflow), assumed to be constant.
i in-system inputs
Flows need not be energy but must have
the same units for every compartment

Xij, flow of i to j; Ej, energy flow to j; Nj, nutrient flow to j.

See Table 1 for additional definitions.
Ecological Network Analysis, Energy Analysis 69

Table 3 Explicit balance equations and solutions for five indicators

Indicator Refer to figure Equations Solutions Units

Energy 4a 100 Z"c 10"p "p 10 2Z cal GPP/cal

intensity (") 10"p 5 Z "c "c 20

Nutrient 5 1 Zc 10p 1 15 Z g nutrient/cal

intensity () 10 15 Z
5 Z
10p 5c Zc c 2
2 c
15 Z
Trophic 4a TPc Z 2Z
1 TPp TPp 1
position (TP) 100 Z 100
TPp 10 2Z
1 TPc TPc 2
10 100
Path length (PL) 4a Z 2Z
PLc 1 PLp PLp
100 Z 100
PLp 1 PLc 2Z
PLc 1
Residence 4a Z Z Z day
c tp p p 1 tp tc
time () 100 Z 100 100
p tc c Z  
c 1 t p tc

Path Length isolated compartment residence times ti, typically defined

as the ratio of stock to throughflow. (Unlike all the indi
This can be expressed in two ways:
cators so far discussed, this requires that we know the
Looking backward in time. A molecule is just now leaving
compartment j. How many intercompartment transits
stocks at steady state.) The system residence time is a
function of these isolated compartment residence times
has it made between its entering the system and now? and the degree of connectedness of the compartments. In
Looking forward in time. A molecule is just now leaving words, residence time for compartment j is the weighted
compartment j. How many intercompartment transits average of the residence time of each input tj. From
will it make on average before exiting the system? Table 3, the residence times are (1 Z/100)tp (Z/
For a steady state system, these are equally easy to 100)tc and (1 Z/100)(tp tc) for producers and consu
calculate. For a dynamic system, the backward looking mers, respectively. Without feedback, the residence time
PL is preferable because it can be calculated without for producers,  p, is just tp because the only input is from
knowing the future. Therefore we calculate only the outside the system. Consumer residence time,  c, is tp tc,
backward looking PL. In words, PL is the weighted because a molecule leaving consumers has passed exactly
sum of the quantity (PL 1) for each input. Imports once through producers and consumers. TP and resi
do not figure in PL, which is based upon internal flows dence time are graphed versus Z in Figure 8.
only. The input flows need not be energy, but they
must all be in the same units so that the weighted
2.5 7
average can be calculated. Cons. res. time
PL is almost the same as TP. For a system with only 6
sunlight as energy input, TPi PLi 1. If there are other
Residence time (day)

Cons. trophic pos.

system energy inputs such as imported feed, the difference
Trophic position

between the two is more significant. As shown in Table 3, 1.5 4

PL 2Z/100 and 1 2Z/100 for producers and consu
mers, respectively. For Z 0, PLp 0 because there are 1 Prod. trophic pos.
no in system inputs to producers. 2
Prod. res. time 1

Residence Time () 0 0

0 1 2 3 4 5
As with PL, this can be expressed looking either backward Feedback, Z
or forward in time, but here only the former is treated: a Figure 8 Trophic position and residence time vs. Z. The
molecule is just now leaving compartment j. How long has isolated compartment residence times are 1 and 5 days for
it been in the system? It is assumed that we know the producers and consumers, respectively.
70 Ecological Network Analysis, Energy Analysis

Indicators in Dynamic Systems of stock and output are the same. Figures 10a and 10b
illustrate this in detail for energy intensity and residence
Calculating Dynamic Indicators
Most of energy analysis and systems ecology stressing The flows are multiplied by the time step t for
indirectness has assumed a steady state in which flows dimensional commensurateness with the stocks. Output
and stocks are constant over time. Yet real systems are can include a change in stock (inventory change in eco
almost always dynamic. All the indicators addressed in nomic terminology), so that the new stock is the old stock
this article can have a dynamic interpretation, as long as plus this change. Figure 10a shows that energy intensity
we use the back looking form. Any dynamic analysis must at time t t is a function of the flows at time t t, and
be explicit about stocks, flows, and time steps. The ele the stocks and energy intensities at time t. If one knows
ments of the dynamic view are shown in Figure 9. the initial energy intensities and stocks, and one has a
Figure 9 summarizes the assumption that in a time dynamic model to specify stocks and flows over time, one
step t, the gloof embodied in the inflows and in the can use the equation implied by Figure 10a to calculate
stock is distributed over the final stock and outflows. At dynamic energy intensities:
the end of the time step, there has been the mathematical X
equivalent of perfect mixing, so that the energy intensity "ti t Xijt t t "tj Sjt "timpj
IMPORTtj t t
i in system inputs

Ejt t t "tj t Xjt t t

"tj t Sjt 4
Gloof embodied in
incoming in-system Gloof embodied in
output flow during Similarly, from Figure 10b, one obtains for dynamic
flows during time
step t j time step t residence time:
Gloof embodied in Gloof embodied it t Xijt t t jt Sjt Sjt t
stock at time t in stock at iin system

Gloof generated time t + t

0 1
internally plus that
embodied in import jt t @ Xijt t t IMPORTj Ej A jt t Sjt
flow during time step t iin system

Figure 9 Generic scheme for allocating the embodied generic 5

system input gloof for the dynamic system. In the underlying
dynamics, stock changes over time as Stt St Figures 10a and 10b also demonstrate how intensity
OUTPUTtt INPUTStt t. (of anything) is injected into a system as a source term


it + t X ijt + t t
it + t X jt + t t
i = in-system

tj Stj jt + t S jt
+ t
timpj IMPORTj t + tt + Ej t + t t


it + t X ijt + t t Xij t + t t + IMPORTj + Ej

jt + t i = in-system
i = in-system
inputs inputs

t j Stj
jt + t S jt
Stj t

Figure 10 (a) Scheme for calculating dynamic energy intensity. The source term is energy itself plus energy embodied in imports. The
energy intensity of imports of type j, "impj, is specified exogenously. (b) Scheme for calculating dynamic residence time. The source term
is the aging of the existing stock in the time period t.
Ecological Network Analysis, Energy Analysis 71

and then allocated by internal flows. For energy inten change when feedback changes, while in the static model
sity, the source is the embodied energy in imports of used in the previous two sections all flows except feed
similar entities (competitive imports in economic ter back are assumed to remain constant.
minology) plus imports of different entities, here just
energy itself.
For residence time, the source term is just the aging
of the stock; external inputs do not contribute to resi Applications
dence time by definition. Similar comments apply to TP Ecological Example: Four-Compartment Food
and PL. Web
Figure 13 shows steady state energy flows and stocks in
a bog in Russia. The analysts disaggregated this ecosys
Simulations of Dynamic Indicators
tem into four compartments: plants, animals,
Figure 11 shows a two compartment dynamic model decomposers, and detritus. Detritus consists of undiffer
system. Initially the system is at steady state with no entiated dead material and therefore has no metabolic
feedback (Z 0), but feedback is switched on (Z 3) at losses. All other compartments contribute to detritus.
time 20 days, and then off again (Z 0) at 500 days. Additionally, animals and decomposers also eat detritus,
The details of the underlying model are not important resulting in two feedback flows and a web structure. All
here; it incorporates a nonlinear ratio dependent feeding indicators can be calculated from the equations given in
response by consumers to abundance of producers, and Table 2. The results are given in Table 4. Because
vice versa when feedback is on. Producer output depends plants have only a solar input, their energy intensity is
on light level, which is assumed constant, and producer quite low and TP 1. For the other compartments,
biomass. Simulation is performed using the modeling however, the energy intensities are higher and the TPs
software STELLA. are high. Decomposers have a TP 4.9, higher than the
Figures 12a12d show dynamic behavior of four of value of 4 which one would expect for a food chain
the indicators calculated here: energy intensity, TP, PL, instead of this web. Decomposers come out on top in
and residence time. On all graphs, the stock of producers both energy intensity and TP. Because this system has
and consumers is shown as well. Immediately after the only one input, PL is just TP1.
onset of nonzero feedback, producer stock increases as Table 4 shows that residence times are affected dra
more material now enters that compartment and consu matically by web structure. Isolated compartment
mer stock drops. But then consumer stock increases in residence times (stock/throughflow) are long for plants
response to increased producer stock, and both stocks and detritus, and short for animals and decomposers. The
asymptotically increase. This is reasonable, because longest, detritus, is 760 times the shortest, animals. In
along with increased feedback comes decreased loss, as contrast, in system residence times differ by only a factor
shown in Figure 11. of 4. Both animals and decomposers, which in isolation
Similar to the steady state calculations, energy inten would be fast, have large input flows from detritus (which
sity, TP, PL, and residence time all increase for both in isolation is slow). The consequence is that all three are
producers and consumers. However, the values are not comparably slow. This is one aspect of the notion that
given exactly by the static equations for Z 3 cal d 1. detritus links tend to slow down the response of ecosys
This is because in the dynamic model, all flows and stocks tems to perturbations.

Z [0]

f (Sun, Sp )[100] Producers Consumers

SP [100] SC [100]
f(SP, SC) [10]

0.90 * S P 0.05 * S C
[90] [5]

Figure 11 Model for dynamic simulation. Figures in square brackets are initial steady-state values, before feedback is started.
Figures within boxes are stocks; others are flows.
72 Ecological Network Analysis, Energy Analysis

200 60
987.55 4.1 Animals

Energy intensity (cal/cal)

50 1.25
150 Plants 16.74
Cons. stock
40 8490
Stock (cal)

100 Prod. stock 30 36.9 58.21
20 337.4
Cons. en. intensity Decomposers
50 78.81 584.9
Prod. en. intensity 279.9
0 0 305
0 200 400 600 800 1000
Time (day) Figure 13 Energy flows (g fixed carbon per m2 per year) in a
bog in Russia. Detritus is undifferentiated dead material, and
200 2.2
therefore has no metabolic loss. Numbers in compartments are
2 stocks (g fixed carbon per m2). From Logofet DO and Alexandrov
Cons. troph. pos. GA (1984) Modelling of matter cycle in a mesotrophic bog
150 Cons. stock 1.8

Trophic position
ecosystem. Part 1: Linear analysis of carbon environs. Ecological
Stock (cal)

1.6 Modelling 21: 247258.

100 Prod. stock
50 Energy/Economic Example: Energy Intensity of
1 Consumer Goods and Services, Energy Cost of
Prod. troph. pos.
0 0.8 Living
0 200 400 600 800 1000
Time (day) This topic is included to emphasize and illustrate the
(c) breadth of applicability of energy analysis and the analo
200 1.2
gies between ecological and economic systems. The
Cons. path length
1 question is how much energy is required to support,
150 directly and indirectly, human household consumption
Path length
Stock (cal)

Cons. stock 0.8

patterns. The approach is in two steps: (1) determine
100 Prod. stock 0.6 how much energy is needed, directly and indirectly, to
produce a product and (2) determine how much of it a
50 household consumes.
0.2 Consider a loaf of bread. The energy to grow the
Prod. path length ingredients, make the bread, and transport and market it
0 0
0 200 400 600 800 1000 can be determined by a detailed vertical analysis (also
Time (day) called process analysis), in which one sums:
200 15 1. the energy used in the supermarket;
2. the energy consumed in the bakery;
Residence time (day)

3. the energy consumed at the flour mill;
Cons. stock Cons. res. time 10 4. the energy used on the farm;
Stock (cal)

5. the energy for transport at every link; and so on.

100 Prod. stock
This process can even lead to cycles in systems with
5 feedback (e.g., cars require steel, but the steel industry
uses some cars), but the process usually converges to an
Prod. res. time
acceptable answer after just several steps.
0 0
0 200 400 600 800 1000
A vertical analysis is potentially accurate, but expen
Time (day) sive. Performing it for a wide range of products is
prohibitive. There is, however, a large database on the
Figure 12 Indicators in dynamic system. Initially system is at
steady state with feedback ( Z ) 0. Z is increased abruptly to a interactions of the sectors (c. 350500) of the US econ
steady value of 3 cal/day for days 20500, and then returned omy. This is the inputoutput (IO) table published by
abruptly to zero. (a) energy intensity; (b) trophic position; (c) path the US Department of Commerce. Many other countries
length; (d) residence time. have similar IO tables. With a number of fairly
Ecological Network Analysis, Energy Analysis 73

Table 4 Energy intensities, trophic positions, path lengths, and residence times for the Russian bog food web of Figure 13

Energy Trophic In-system residence

intensity, " position Path length Isolated-compartment residence time, 
Compartment (cal GPP/cal) (TP) (PL) time, t (years) (years)

Plants 2.60 1.00 0.00 8.60 8.60

Animals 9.56 3.43 2.43 0.017 20.5
Detritus 12.4 3.90 2.90 12.6 32.7
Decomposers 23.8 4.90 3.90 0.060 32.8

stringent assumptions, this table can be combined with Table 5 Energy intensities for household consumption
categories (Btu/$, 2003 technology, 2003 dollars)
direct energy use data for each sector to produce energy
intensities using the equation implied by Figure 3b. One 1. Residential fuel, electricity 139 300
such assumption is necessitated by the fact that the units 2. Vehicle fuel 94 300
3. Vehicle purchase, maintenance 5 400
in IO tables are monetary units per year, so one must
4. Food 6 100
accept dollars as an appropriate allocator of embodied 5. Alcohol, tobacco 3 700
energy. Because in the American energy industry, 6. Apparel 6 500
energy is usually measured in Btu, the energy intensity 7. Communication, entertainment 4 000
of goods and services is then expressed in Btu/$. IO 8. Health, personal care 2 400
9. Reading, education 3 000
based determination of energy intensities has been per 10. Insurance, pension 1 600
formed for c. 35 years. Under further assumptions, the 11. Contributions 3 800
intensities can be used to evaluate the energy impact of 12. Public transportation 21 200
different expenditure patterns. Doing this for a house 13. Asset gain 4 700
hold yields the so called energy cost of living. 14. Miscellaneous 4 200
15. Housing 5 100
The IO data are available, but gathering the asso
Direct energy ((1) (2)) 118 100
ciated direct energy data and performing the computation
Nonenergy (sum of (3)(15)) 4 700
is tedious, though todays computers make it increasingly
All personal consumption 11 100
easier. Solving 500 simultaneous equations of the form in Energy/GDP 8 900
eqn [1] is done by inverting a 500 rank matrix. Sprawl ((1) (2) (3) (15)) 15 700
Once we have the energy intensities, we need details Nonsprawl (sum of (4)(14)) 4 300
on how households spend their money over the range of Auto and related ((2) (3)) 21 200
consumer product categories, also known as their market
Shaded categories indicate an intensity greater than the energy/GDP
basket. This information is collected by the US Bureau of ratio. Sprawl contains housing and auto ownership and operation.
Labor Statistics. Putting the two together yields the Source: Authors calculations based on Carnegie Mellon University
energy cost of living (ECOL): data.

ECOL "i Yi 6
i all expenditure categories Btu/yr) using eqn [6] and intensities such as those shown in
Table 5. In Figure 14a, we see that of the average house
where Yi is the households annual expenditure for holds expenditures of $49 300 in 1973, only 6.4% was
expenditure category i. Applying eqn [6] allows one to for direct energy (residential fuel and electricity and
analyze the effect of overall spending and the mix in the auto fuel). After conversion to energy requirements
market basket. The latter will be significant only if the (Figure 14b), this portion was 63% of the total impact of
energy intensities are different for different expenditure 604 million Btu. The total is roughly the energy equivalent
categories. of 100 barrels of oil. Figures 15a and 15b show the energy
Table 5 shows IO based energy intensities deter pie for the lowest expenditure decile ($11 500/yr, 241
mined by Carnegie Mellon University for 1997, and million Btu/yr) and highest decile ($140 200/yr, 1233
updated and aggregated by the author into 15 categories million Btu/yr). The direct fraction is largest, 79%, for
covering all household expenditures. The intensities are the lowest decile, and lowest, 47%, for the highest decile.
indeed different, especially energy itself and service Figure 16 shows a statistical fit to energy versus
industries such as health care. expenditures for a representative sample of several thou
Figure 14 shows the result of transforming of a house sand American households. It confirms that because the
hold market basket (in dollars/yr) to its energy impact (in mix changes, energy is not a linear function of total
74 Ecological Network Analysis, Energy Analysis

(a) (a)
Asset gain
Health care Other 0.1% Health care
12.5% Public trans. 1.5% Food/alc./tobacco
5.9% 0.8%
Asset gain 0.6% 6.6%
Apparel Auto purch. maint.
6.8% Apparel
2.5% 0.5% 0.6%
Infrastructure Infrastructure
0.9% Auto fuel 1.6%
Other 15.3% Housing
19.6% 9.1%

Public trans.
Auto purch. maint.
13.8% Residential
Auto fuel energy
2.9% 3.5%

(b) energy
Asset gain Health care Food/alc./tobacco
Other 2.5% 1.0% 6.3% (b)
4.6% Apparel
Health care Food/alc./tobacco
1.4% Apparel
Public trans. 1.0% 5.4%
Asset gain 1.7%
1.5% 10.0%
Housing 0.5%
Auto purch. maint. Other
6.2% 7.3%

Public trans. Housing

Auto fuel 2.4% 14.6%

Auto purch. maint.

Residential 10.1%
40.3% Residential
Figure 14 For the average American household in 2003: 28.2%
(a) expenditures, which total $49 300; (b) energy impacts, which Auto fuel
totaled 604 million Btu. Source: Unpublished calculations by 18.8%
R. Shammin, R. A. Herendeen, M. Hanson, and E. Wilson.
Figure 15 2003 household energy impacts for (a) lowest
income decile ($11 500; 241 million Btu), (b) highest income decile
($140 200; 1233 million Btu). Source: Unpublished calculations by
R. Shammin, R. A. Herendeen, M. Hanson, and E. Wilson.
expenditures, but rather bends down and away from a
straight line through the origin. The reason is that direct
energy (auto fuel and residential fuel and electricity)
tends to level out as household expenditures increase. of the total for less affluent households, regressive impacts
Expenditures increase for other products, but these of an energy price increase would be expected if one
tend to be less energy intensive. For developed countries, ignored the indirect portion.
the shape of Figure 16 seems robust: studies of However, because the total energy curve in Figure 16
Norway, the Netherlands, and Australia have found a bends down, some regressiveness is still expected. To
similar result. compensate, one could design an income tax rebate to
even out the impacts over income classes. Figure 16 is the
key, as follows.
Energy/Economic Example: Regressive Effects
Suppose that fossil energy at the wellhead or mine
of an Energy Tax
mouth is taxed at rate of p dollars per Btu. Assume that
Concern with global warming, energy security, and pol economic sectors maintain their patterns of using inputs
lution strongly implies that fossil energy is too cheap to to produce inputs, that is to say, technology is constant.
compensate for its drawbacks. Energy taxes of various Assume further that each sector can successfully pass on
sorts have been proposed to stimulate more efficient use its increased costs to the consumers of its output. Then a
and to fund alternatives. In the debate, equity issues households market basket, if unchanged, will now cost an
quickly surface. Because direct energy is a larger fraction additional amount p  ECOL. The fractional increase
Ecological Network Analysis, Energy Analysis 75

2500 Table 6 Consequences of a $4 per million Btu mine mouth

fossil energy tax, based on assumptions in text
Total energy
2000 Direct energy Expenditure level
Indirect energy
Energy in MBtu/yr

Lowest Highest
5000 decile Average decile

Market basket expenditure 11.5 49.3 140.2

(thousand $/yr)
Total energy (million Btu/yr) 241 604 1233
500 h"i (thousand Btu/$) 21.0 12.3 8.8
Market basket price 964 2416 4932
increase ($/yr)
0 Market basket price 8.4 4.9 3.5
0 50 100 150 200 250 increase (%)
Total expenditure in 1000 US $/yr

Figure 16 Household energy impact vs. total expenditures.

Direct energy is auto and residential fuel and electricity.
Total energy direct energy plus energy impact of all other Further Reading
purchases. Source: Unpublished calculations by R. Shammin,
R. A. Herendeen, M. Hanson, and E. Wilson. Bullard C, Penner P, and Pilati D (1978) Net energy analysis: Handbook
for combining process and input output analysis. Resources and
Energy 1: 267 313.
Burns T (1989) Lindemans contradiction and the trophic structure of
is this quantity divided by the total market baskets origi ecosystems. Ecology 70: 1355 1362.
Fath BD and Patten BC (1999) Network synergism: Emergence of
nal cost, denoted by Y. then positive relations in ecological models. Ecological Modelling
107: 127 143.
Fract:incr:in mkt: basket cost ph"i 7 Finn JT (1976) Measures of ecosystem structure and function derived
Y from analysis of flows. Journal of Theoretical Biology 56: 115 124.
Hannon B (1973) The structure of ecosystems. Journal of Theoretical
where h"i is the average energy intensity of the market Biology 41: 535 546.
basket. The average is just energy/expenditure at the Herendeen R (1989) Energy intensity, residence time, exergy, and
appropriate point on Figure 16; it is the slope of a straight ascendency in dynamic ecosystems. Ecological Modelling
48: 19 44.
line connecting the origin and the point. Herendeen R and Fazel F (1984) Distributional aspects of an energy
As an example, consider a tax of $0.50 per gallon of conserving tax and rebate. Resources and Energy 6: 277 304.
gasoline equivalent. This is about $4/million Btu, or $24/ Herendeen R, Ford C, and Hannon B (1981) Energy cost of living, 1972
1973. Energy 6: 1433 1450.
barrel of oil. The latter is about 25% of the world crude Lenzen M, Wier M, Cohen C, et al. (2006) A comparative multivariate
oil price as of 7 November 2007. analysis of household energy requirements in Australia, Brazil,
Using Figures 1416, we perform the calculation in Denmark, India, and Japan. Energy 31: 181 207.
Logofet DO and Alexandrov GA (1984) Modelling of matter cycle in a
Table 6. The increase in market basket price ranges from mesotrophic bog ecosystem. Part 1: Linear analysis of carbon
3.5% for the highest expenditure decile to 8.9% for the environs. Ecological Modelling 21: 247 258.
lowest. For full equity, income tax rebates, or other mea Odum HT (1996) Environmental Accounting. New York: Wiley.
Ulanowicz R (1986) Growth and Development: Ecosystems
sures, could address this differential. Phenomenology. New York: Springer.
Needless to say, a proper calculation is much more
involved than this one, but the idea of indirectness will
pervade it.
Relevant Website

See also: Cycling and Cycling Indices; Ecological Economic Input Output Life Cycle
Network Analysis, Ascendency; Ecological Network Assessment, Carnegie Mellon University.
Analysis, Environ Analysis; Indirect Effects in Ecology.
76 Ecological Network Analysis, Environ Analysis

Ecological Network Analysis, Environ Analysis

B D Fath, Towson University, Towson, MD, USA and International Institute for Applied System Analysis,
Laxenburg, Austria
2008 Elsevier B.V. All rights reserved.

Introduction Network Properties

Theoretical Development of Environ Analysis Summary
Data Requirements and Community Assembly Rules Further Reading
Methods and Sample Network

Introduction fact, one of the three foundational principles in his seminal

paper introducing the environ theory concept. The neces
Environ Analysis is in a more general class of methods sary boundary demarcates two environments, the
called ecological network analysis (ENA) which uses net unbound external environment, which indeed includes
work theory to study the interactions between organisms all spacetime objects in the universe, and the second
or populations within their environment. Bernard Patten internal, contained environment of interest. This quantifi
was the originator of the environ analysis approach in the able, internal environment for each system object is
late 1970s, and he, along with his colleagues, has termed environ, and is the focus of environ analysis. An
expanded the analysis to reveal many insightful, holistic objects environ stops at the system boundary, but as eco
properties of ecosystem organization. ENA follows along systems are open systems, they require exchanges across
the synecology perspective introduced by E. P. Odum the boundary into and out of the system environs.
which is concerned with interrelations of material, Therefore, input and output boundary flows are necessary
energy, and information among system components. to maintain the systems far from equilibrium organiza
ENA starts with the assumption that a system can be tion. Objects and connections that reside wholly in the
represented as a network of nodes (compartments, ver external environment are not germane to the analysis.
tices, components, storages, objects, etc.) and the Another foundational principle of environ analysis
connections between them (links, arcs, flows, etc.). In theory is that each object in the system itself has two
ecological systems, the connections are based on the environs one receiving and one generating interactions
flow of energy, matter, or nutrients between the system in the system. In other words, an objects input environ
compartments. If such a flow exists, then there is a direct includes those flows from within the system boundary
transaction between the two connected compartments. leading to the object, and an output environ, those flows
These direct transactions give rise to both direct and emanating from the object back to the other system
indirect relations between all the objects in the system. objects before exiting the system boundary. This alters
Network analysis provides a systems oriented perspec the perception of a system component from internal
tive because it uncovers patterns and relations among all external to receivinggenerating. Thus, the object, while
the objects in a system. Therefore, showing how system distinct in time and space, is more clearly embedded in
components are tied to a larger web of interactions. and responsive to the couplings with other objects within
the network. This shifts the focus from the objects them
selves to the relations they maintain; or from parts to
Theoretical Development of Environ processes (or what Ilya Prigogine called from being to
Analysis becoming).
The third foundational principle is that individual
Patten was motivated to develop environ analysis to environs (and the flow carried within each one) are
answer the question, What is environment?. In order to unique such that the system comprises the set union of
study environment as a formal object, a system boundary is all environs, which in turn partitions the system level of
a necessary condition to avoid the issue of infinite indir organization. This partitioning allows one to classify
ectness, because in principle one could trace the environ flow into what have been called different
environment of each object out in space and back in time modes: (1) boundary input; (2) first passage flow received
to the big bang origins. The inclusion of a boundary is, in by an object from other objects in the system (i.e., not
Ecological Network Analysis, Environ Analysis 77

boundary flow), which has not cycled; (3) cycled flow, y2

which returns to a compartment before leaving the sys
z1 x2
tem; (4) dissipative flow that it has left the focal object f21 f42
not to return, but does not directly cross a system y4
boundary (i.e., it flows to another within system object); x1 x4
and (5) boundary outflow. The modes have been used to y1 f43
understand better the general role of cycling and the flow
f31 f54
contributions from each object to the other, which has had x3
application in showing a complementarity of several of the
holistic, thermodynamic based ecological indicators. f53
f15 y5
Data Requirements and Community
Assembly Rules Figure 1 Sample network with five compartments used to
demonstrate environ analysis notation and methodology.
Network environ analysis could be referred to as a
holistic/reductionistic approach. It is holistic because it
considers simultaneously the whole influence of all sys Compartments are connected by transaction of the
tem objects, yet it is reductionistic in that the fine details energymatter substance flowing between them. These
of all object transactions are entailed in the analysis. In pairwise couplings are the basis for the internal network
other words, it is the opposite of a black box model. The structure. A structural connectance matrix, or adjacency
network data requirements are considerable, which matrix, A, is a binary representation of the connections
include the complete flowstorage quantities for each such that aij 1 if there is a connection from j to i, and a 0
identified link and node (note flow and storage are
otherwise (eqn [1]):
interchangeable as determined by the turnover rate).
2 3
Data can be acquired from empirical observations, lit 0 0 0 0 1
erature estimates, model simulation results, or balancing 6 7
6 0 0 0 077
procedures, when all but a few are unknown. This diffi 6 7
culty in obtaining data has resulted in a dearth of A6
61 1 0 0 077 1
6 7
available complete network data sets. Due to this lack 60
4 1 1 0 075
of requisite data for fully quantified food webs, research
0 0 1 1 0
ers have developed community assembly rules that are
heuristics to construct ecological food webs. Assembly Storage and flows must have consistent units (although it
rules are in general a set of rules that will generate a is possible to consider multiunit networks). Typically,
connectance matrix for a number of species (N). units for storages are given in amount of energy or bio
Common assembly rules that have been developed are mass per given area or volume (e.g., g m 2), and units for
random or constant connectance, cascade, niche, modi flows are the same but as a rate (e.g., g m 2 d 1). The
fied niche, and cyber ecosystem, each with its own intercompartmental flows for Figure 1 are given in the
assumptions and limitations. In all but the last case, the following flow matrix, F:
assembly rules construct only the structural food web
topology. The cyber ecosystem methodology also 2 3
0 0 0 0 f15
includes a procedure for quantifying the flows along 6 7
6 f21
6 0 0 0 0 7
each link. It uses a metastructure of six functional groups: 6 7
producer (P), herbivore (H), carnivore (C), omnivore F6
6 f31 f32 0 0 0 7
7 2
6 7
(O), detritus (D), and detrital feeders (F), within which 60
4 f42 f43 0 0 7
random connections link species based on these defini
tional constraints. Flows are assigned based on realistic 0 0 f53 f54 0
thermodynamic constraints.
Note that the orientation of flow from j to i is used
because that makes the direction of ecological relation
Methods and Sample Network from i to j. For example, if i preys on j, the flow of
energy is from j to i. All compartments experience
To demonstrate basic environ analysis, it is best to pro dissipative flow losses (yi, for i 15), and here the
ceed with an example. Consider the network in Figure 1, first compartment receives external flow input, z1
which has five compartments or nodes (xi, for i 15). (arrows not starting or ending on another compartment
78 Ecological Network Analysis, Environ Analysis

represent boundary flows). For this example, these can Indirectness originates from transfers or interactions that
be given as occur nondirectly, and are mediated by other within
system compartments. These transfers could travel two,
y y1 y2 y3 y4 y5  3 three, four, or many links before reaching the target
destination. For example, the flow analysis starts with
the calculation of the nondimensional flow intensity
matrix, G, where gij fij/Tj. The generalized G matrix
2 3
z1 corresponding to Figure 1 would look as follows:
6 7
607 2 3
6 7
6 7 0 0 0 0 g15
z6 7 4 6 7
607 6 g21 0 0 0 0 7
6 7 6 7
607 6 7
4 5 G6 g32 0 0 0 7 6
6 g31 7
0 6 7
6 0
4 g42 g43 0 0 75
0 0 g53 g54 0
Total throughflow of each compartment is an impor
tant variable, which is the sum of flows into,
P P These values represent the fraction of flow along each link
Tiin zi nj fij , or out of, Tiout yi nj fji the ith normalized by the total throughflow at the donating com
compartment. At steady state, compartmental inflows partment. These elements give the direct, measurable flow
and outflows are equal such that dxi/dt 0, and therefore, intensities (or probabilities) between any two nodes j to i.
incoming and outgoing throughflows are also equal: To identify the flow intensities along indirect paths (e.g.,
Tiin Tiout Ti . In vector notation, compartmental j ! k ! i), one need only consider the matrix G raised to
throughflows are given by the power equal to the path length in question. For exam
2 3 ple, G2 gives the flow intensities along all paths of length
T1 2, G3 along all paths of length 3, etc. This well known
6 7
6 T2 7 matrix algebra result is the primary tool to uncover system
6 7
6 7 indirectness. In fact, it turns out that due to the way in
T6 7 5
6 T3 7 which the G matrix is constructed, all elements in Gm go
6 7
6 T4 7 to zero as m ! 1. Therefore, it is possible to sum the
4 5
T5 terms of Gm to acquire an integral flow matrix (called N),
which gives the flow contribution from all path lengths:
This basic information regarding the storages, flows, and
boundary flows provides all the necessary information to X
N G0 G1 G2 G3    Gm I G 1 7
conduct environ analysis. Environ analysis has been clas m 0
sified into a structural analysis, dealing only with the
network topology, and three functional analyses (flow, where G0 I, the identity matrix, G1 the direct flows,
storage, and utility) which requires the numerical values and Gm for m > 1 are all the indirect flows intensities.
for flow and storage in the network (Table 1). Note, that the elements of G and N are nondimensional;
The technical aspects of environ analysis are explained to retrieve back the actual throughflows, one need only
in detail elsewhere, so rather than repeat those here, the multiply the integral matrix by the input vector: T Nz.
remainder of the article highlights some of the important In other words, N redistributes the input, z, throughout
results from environ analysis. But first, one issue that must each compartment to recover the total flow through that
be covered is the way in which network analysis identifies compartment. Similarly, one could acquire any of the
and quantifies indirect pathways and flow contributions. direct or indirect flows by multiplying Gmz for any m.

Table 1 Basic methodologies for network environ analysis

Structural analysis Functional analyses

Path analysis Flow analysis: gij fij/Tj

Enumerates pathways in a network (connectance, cyclicity, etc.) Identifies flow intensities along indirect pathways
Storage analysis: cij fij/xj
Identifies storage intensities along indirect pathways
Utility analysis: dij (fij fji)/Ti
Identifies utility intensities along indirect pathways
Ecological Network Analysis, Environ Analysis 79

A similar argument is made to develop integral storage The implications of this important result are clear in
and utility matrices: that each compartment is embedded in and dependent
1 on the rest of the network for its situation, thus calling for
storage : Q P0 P1 P2 P3    Pm I P 1 a true systems approach to understand such things as
m 0
feedback and distributed control in the network.
utility : U D0 D1 D2 D3    Dm I D 1 Network Homogenization
m 0
9 The homogenization property yields a comparison of
resource distribution between the direct and integral
where pij (fij/xj)t, and dij (fij  fji)/Ti. flow intensity matrices. Due to the contribution of indir
ect pathways, it was observed that flow in the integral
matrix was more evenly distributed than that in the
direct matrix. A statistical comparison of resources dis
Network Properties
tribution can be made by calculating the coefficient of
variation of each of the two matrices. For example, the
Patten has developed a series of ecological network proper
coefficient of variation of the direct flow intensity
ties which summarize the results of environ analysis. The
matrix G is given by
properties have been used to assess the current state of
ecosystem networks and to compare the state of different Pn Pn  2
j 1 i 1 gij gij
networks. Furthermore, while interpreting some of the prop CVG 11
n 1g
erties as ecological goal functions, it has been possible to
identify the structural or parametric configurations that posi Network homogenization occurs when the coefficient of
tively affect the network property values as a way to detect or variation of N is less than the coefficient of variation of G
anticipate network changes. For example, certain network because this says that the network flow is more evenly
alterations, such as increased cycling, lead to greater total distributed in the integral matrix. The test statistic
system energy throughflow and energy storage, so one could employed here looks at whether or not the ratio CV(G)/
expect that if possible ecological networks are evolving or CV(N) exceeds 1. The interpretation again is clear that
adapting to such configurations. This leads to a new area of the view of flow in ecosystems is not as discrete as it
research on evolving networks. In this section, a brief over appears because in fact the material is well mixed (i.e.,
view is given for four of these properties: dominance of homogenized) and has traveled through and continues to
indirect effects (or nonlocality), network homogenization, travel through many, if not, most parts of the system.
network mutualism, and environs.
Network Mutualism
Dominance of Indirect Effects
Turning now to the utility analysis, the net flow, utility
This property compares the contribution of flow along matrix, D, can be used to determine quantitatively and
indirect pathways with those along direct ones. Indirect qualitatively the relations between any two components
effects are any that require an intermediary node to in the network such as predation, mutualism, or competi
mediate the transfer and can be of any length. The tion. Entries in the direct utility matrix, D, or integral
strength of indirectness has been measured in a ratio of utility matrix, U, can be positive or negative (1 dij,
the sum of the indirect flow intensities divided by the uij < 1). The elements of D represent the direct relation
direct flow intensities: between that (i, j) pairing; for the example in Figure 1,
this produces the following:
i;j 1 nij gij ij 2 3
Pn 10 0
f21 f31
i;j 1 gij 6
6 T2 T3 T5 7 7
6 7
6 f21 f32 f42 7
where ij, the Kronecker delta, is 1 if and only if i j and 6 0 0 7
6 T2 T3 T4 7
is 0 otherwise. When the ratio is greater than 1, then 6 7
6 f31 f32 f43 f53 7
dominance of indirect effects is said to occur. Analysis of D6 0 7 12
6 T3 T3 T4 T5 7
many different models has shown that this ratio is often 6 7
greater than 1, revealing the nonintuitive result that indir
6 f42 f43 f54 7
6 0 0 7
ect effects have greater contribution than direct effects. 6 T4 T4 T5 7
6 7
4 f15 f53 f54 5
Thus, each compartment influences each other, often 0 0
significantly, by many indirect, nonobvious pathways. T5 T5 T5
80 Ecological Network Analysis, Environ Analysis

Table 2 Direct and integral relations in sample network from Figure 1

Direct Integral

(sd21, sd12) (, ) ! exploitation (su21, su12) (, ) ! exploitation

(sd31, sd13) (, ) ! exploitation (su31, su13) (, ) ! exploitation
(sd41, sd14) (0, 0) ! neutralism (su41, su14) (, ) ! mutualism
(sd51, sd15) ( , ) ! exploited (su51, su15) (, ) ! mutualism
(sd32, sd23) (, ) ! exploitation (su32, su23) ( , ) ! competition
(sd42, sd24) (, ) ! exploitation (su42, su24) (, ) ! exploitation
(sd52, sd25) (0, 0) ! neutralism (su52, su25) (, ) ! mutualism
(sd43, sd34) (, ) ! exploitation (su43, su34) (, ) ! exploitation
(sd53, sd35) (, ) ! exploitation (su53, su35) (, ) ! exploitation
(sd54, sd45) (, ) ! exploitation (su54, su45) (, ) ! exploitation

The direct matrix D, being zero sum, always has the same distinct environs, there are in fact 2n environs in total.
number of positive and negative signs: The output environ, E, for the ith node is calculated as:
2 3 E G IN i 15
0 0
6 7
6 0 07 where Ni is the diagonalized matrix of the ith column of
6 7
6 7 N. When assembled, the result is the output oriented flow
sgnD 6 7 13
6 0 7 from each compartment to each other compartment in the
6 7
60 0 7
4 5 system and across the system boundary. Input environs
0 0 are calculated as
E9 N i9G9 I 16
The elements of U provide the integral, system deter
mined relations. Continuing the example, and now
where g9ij fij/Ti, and N9 (I  G9) 1. These results
including flow values derived from 10% transfer effi
comprise the foundation of network environ analysis
ciency along each link (gij 0.10, if aij 1, and gij 0
since they allow for the quantification of all within system
otherwise), we get the following integral relations
interactions, both direct and indirect, on a compartment
between compartments:
by compartment basis.
2 3

6 7
6 77
6 7 Summary
sgnU 6 7 14
6 7
6 7
6 7 A practical objective of ENA in general, and environ
4 5
analysis in particular, is to trace material and energy

flowstorage through the complex network of system
Unlike, the direct relations, this is not zero sum. interactions. The network environ approach has been a
Instead, we see that there are 17 positive signs (includ fruitful way of holistically investigating ecological sys
ing the diagonal) and 8 negative signs. If there are a tems. In particular, a series of network properties such
greater number of positive signs than negative signs in as indirect effects ratio, homogenization, and mutualism
the integral utility matrix, then network mutualism is have been observed using this analysis, which consider
said to occur. Network analysis demonstrates the posi the role of each entity embedded in a larger system.
tive mutualistic relations in the system. Specifically, See also: Cycling and Cycling Indices; Ecological
here, we can identify two cases of indirect mutualism, Network Analysis, Ascendency; Ecological Network
seven of exploitation, and one of competition Analysis, Energy Analysis; Emergent Properties; Indirect
(Table 2). Effects in Ecology.

Environ Analysis Further Reading

The last property mentioned here is the signature prop Dame RF and Patten BC (1981) Analysis of energy flows in an intertidal
erty, the quantitative environ, both in the input and oyster reef. Marine Ecology Progress Series 5: 115 124.
Fath BD (2007) Community level relations and network mutualism.
output orientation. Since each compartment has two Ecological Modelling 208: 56 67.
Indirect Effects in Ecology 81

Fath BD and Patten BC (1998) Network synergism: Emergence of Jrgensen SE, Fath BD, Bastianoni S, et al. (2007) Systems Ecology: A
positive relations in ecological systems. Ecological Modelling New Perspective. Amsterdam: Elsevier.
107: 127 143. Patten BC (1978) Systems approach to the concept of environment.
Fath BD and Patten BC (1999) Review of the foundations of network Ohio Journal of Science 78: 206 222.
environ analysis. Ecosystems 2: 167 179. Patten BC (1981) Environs: The superniches of ecosystems. American
Fath BD, Jrgensen SE, Patten BC, and Straskraba M (2004) Zoologist 21: 845 852.
Ecosystem growth and development. Biosystems 77: 213 228. Patten BC (1982) Environs: Relativistic elementary particles or ecology.
Gattie DK, Schramski JR, Borrett SR, et al. (2006) Indirect effects and American Naturalist 119: 179 219.
distributed control in ecosystems: Network environ analysis of a Patten BC (1991) Network ecology: Indirect determination of the life
seven compartment model of nitrogen flow in the Neuse River environment relationship in ecosystems. In: Higashi M and Burns TP
Estuary, North Carolina, USA Steady state analysis. Ecological (eds.) Theoretical Ecosystem Ecology: The Network Perspective,
Modelling 194(1 3): 162 177. pp. 288 315. London: Cambridge University Press.
Halnes G, Fath BD, and Liljenstrom H (2007) The modified niche model: Whipple SJ and Patten BC (1993) The problem of nontrophic processes
Including a detritus compartment in simple structural food web in trophic ecology: Towards a network unfolding solution. Journal of
models. Ecological Modelling 208: 9 16. Theoretical Biology 163: 393 411.
Higashi M and Patten BC (1989) Dominance of indirect causality in
ecosystems. American Naturalist 133: 288 302.

Indirect Effects in Ecology

V Krivtsov, University of Edinburgh, Edinburgh, UK
2008 Elsevier B.V. All rights reserved.

Introduction Problems and Implications for Environmental

Basics Management
Examples of Occurrence and Importance of Indirect Current and Further Directions
Effects Further Reading
Approaches and Techniques Used to Detect and
Measure Indirect Effects

Introduction technological progress. It should also be noted that the

boost of the growing appreciation of indirect effects in
Interrelations among ecosystem components and pro twentieth century was partly initiated by Vernadskys
cesses can be subdivided into direct (i.e., those which are fundamental theories about the biosphere, the noo
restricted to the direct effect of one component/process sphere, and interrelations between biota and
on another, and are attributable to an explicit direct geochemical cycling. Popularization of these views 50
transaction of energy and/or matter between the compo years later (e.g., by Lovelocks Gaia theory) stimulated
nents in question) and indirect (i.e., those that do not investigations of indirect effects even further.
comply with the above restriction). The history of natural
sciences is inseparable from the gradually increasing
awareness and understanding of indirect effects. By nine Basics
teenth century the significance of indirect interactions was
well realized, and was (sometimes implicitly) accounted There have been many definitions of direct and indirect
for in the classic studies of Darwin, Dokuchaiev, effects. Information on indirect interactions is scattered in
Gumboldt, Engels, and many other scientists. In the twen the literature, and may appear under various terms. For
tieth century, however, appreciation of indirect effects in example, among ecological phenomena which may
nature received considerable acceleration, predominantly (depending on the exact definition) be regarded as indir
due to the accumulating interdisciplinary knowledge of ect effects are exploitative and apparent competition,
natural ecosystems, the development of appropriate facilitation, mutualism, cascading effects, tri trophic
mathematical techniques, and the urgent necessity to level interactions, higher order interactions, interaction
resolve the growing problems of environmental damage, modification, nonadditive effects, etc.
resulting, ironically, from the uncurbed expansion of the First of all, it is important to distinguish between direct
human population backed by the advances of the and indirect effects. Usually, the interactions between two
82 Indirect Effects in Ecology

components not involving direct transfer of energy and/ example, demographic changes in the population struc
or matter are viewed as indirect, while those that involve ture, changes in the genotypic composition, and changes
an explicit direct transaction are viewed as direct. The in behavior (e.g., searching rates, antipredator behaviors),
literature is inconsistent on the definitions of indirect morphology, biochemistry (e.g., nutrient content, toxin
effects, and one way to clarify the problem is to stress concentration), or physiology.
the difference between a transaction and a relation. A
simple transaction between two ecosystem components
is always direct since it is the transfer of matter and/or Most Commonly Studied Indirect Effects
energy, whereas a relation is the qualitative type of inter
action. Relations include predation, mutualism, Among a plethora of possible indirect effects, there are
competition, commensalism, ammensalism, etc. Hence a five that have been studied most commonly. Their
direct relationship is the one which is based on a direct essence is depicted in Figure 1 and is briefly explained
(i.e., unmediated by another ecosystem component) trans below.
action only. For example, the classic predation (not to be
mistaken with, for example, keystone predation, indirect Interspecific competition
predation, etc.) is direct, and so is the nutrient uptake by Interspecific competition (also called exploitative compe
plants, algae, and bacteria, whereas mutualism and com tition) takes place whenever two (or several) species
petition are always indirect, as they result from the compete for the same resource. In Figure 1a, an increase
combination of a number of simple transactions. It is in Component 1 will lead to the increased consumption of
worth pointing out that the observed patterns of inter the shared resource (Component 2), and consequently to
relations between ecosystem components (e.g., the decrease in a competitor (Component 3). Examples of
correlation between abundance indices) frequently result this include, for example, two predators sharing the same
from a combination of direct and indirect effects, as each prey, or two microbial species whose growth is limited by
component is involved in a large number of pathways. the availability of the same nutrient.
Furthermore, if a direct relationship between two ecosys
tem components (say A and B) is modified by a third Apparent competition
ecosystem component, attribute, or forcing function (the Apparent competition occurs when two species have
two latter notions will include, for example, such modi a common predator. In Figure 1b an abundant
fiers as sunlight, temperature, pH, external and internal population of species 1 sustains a high density population
concentrations of alternative nutrients) then the of predator 2, who, in turn, may limit the population of
indirect relationship between the modifying agent and another prey species 3. From practical point of view, it is
the first two components (i.e., A and B) becomes worth noting here that this situation sometimes happens
superimposed upon the direct relationship between the as an unwanted result in biocontrol, when a biocontrol
components A and B. Consequently, the observed pattern agent (species 2), specifically introduced to control a
of interrelation (e.g., correlation between the abundance target (species 1), may increase the risk of a nontargets
data) between A and B will in this case result from the (species 3) extinction.
combination of direct and indirect effects.
Examples of factors known to modify the strength of Trophic cascades
density mediated indirect interactions include differ Trophic cascades involve propagation of the effect along
ences in the specific growth rates (important, for a vertical trophic chain consisting of three or more com
example, for apparent competition), density dependence ponents connected by grazing or predation. In Figure 1c,
of the transmitting compartment, and the possibility of an increase/decrease in Component 4 will lead to the
stochastic physical disruption. On the other hand, issues decrease/increase in Component 3, increase/decrease in
important in determining the manifested strength of the Component 2, and decrease/increase in Component 1.
behavior mediated indirect interactions involve ability of These effects are particularly well studied in aquatic
a focal species to detect changes in factors which matter food chains (see examples below), but have also been
for energetic costs and benefits of its behavior, sensitivity studied in terrestrial systems.
of its optimum behavior to these costs and benefits, and It is worth pointing out, however, that the structure
available behavioral options. of real ecosystems hardly ever fits tidily into the con
For density mediated effects, presence and strength of cepts of simple trophic levels (e.g., omnivory is
indirect interactions can be determined by analyzing par widespread in nature), and trophic cascades, therefore,
tial derivatives of the abundance of a species on the are often complicated by the interlinks within and
abundances of other (not immediately connected) species. among trophic levels (e.g., in terrestrial ecosystems
However, indirect interaction may involve ecologically insectivorous birds prey on predatory, herbivorous, and
important changes other than changes in abundance, for parasitoid insects, and the resulting effect of birds on the
Indirect Effects in Ecology 83

(b) Apparent
(a) Interspecific competition competition (c) Trophic cascade

1 3 4

2 1 3
(d) Indirect mutualism
involving exploitative
competition (f) Interaction modification
(e) Indirect mutualism involving 2
1 5 interference competition

1 4 3

2 4
2 3
Figure 1 Diagrams of the most commonly studied indirect affects. Direct effects are shown using solid lines, while indirect effects
(only the effects relevant to the accompanying discussion are illustrated) using dotted lines. Interaction modification is illustrated using a
dashed line. Numbers in the compartments are used solely for labeling to distinguish between different compartments, and do not relate
to any kind of hierarchy. Likewise, the box sizes do not bear any relevance to the sizes or significance of the compartments drawn, and
the relative size of the arrows relates neither to the effects strength no to the preferential directionality. See further explanations in the
text. (a) Interspecific competition; (b) apparent competition; (c) trophic cascade; (d) indirect mutualism involvingb9000exploitative
competition; (e) indirect mutualism involving interference competition; (f) interaction modification. Modified from Wootton JT (1994) The
nature and consequences of indirect effects in ecological communities. Annual Review of Ecology and Systematics 25: 443466.

primary producers and their damage by herbivory may, where birds increase the abundance of acorn barnacles
therefore, depend on the specific species and the condi by consuming limpets that otherwise dislodge the
tions involved). In particular, proper consideration of young barnacles off the rock.
detritus contributions to the energy flows may prove
the trophic cascade simplification unsuitable, as the Interaction modification
detritus compartment often has direct links to a number Interaction modification occurs when the relationship
of trophic levels. between a species pair is modified by a third species
(Figure 1f). Examples include positive effects of macro
Indirect mutualism and commensalism algae on zooplankton through interference with the
Indirect mutualism and commensalism involve a con hunting potential of fish and changing of a chemicals
sumerresource interaction coupled with either bioavailability due to the activity of a species, when the
exploitative (Figure 1d) or interference (Figure 1e) chemical in question is important for the functioning of
competition. For instance, starfish and snails reduce another species (e.g., acids produced by one microbial
the abundance of mussels, a dominant space occupier, population may increase bioavailability of compounds
and increase the abundance of inferior sessile species. that are bound or unaccessible for another microbial
The presence of grazers on oyster farms in Australia population).
increases oyster recruitment by removing algae, who It is worth pointing out that interaction modification
otherwise preempt the available spaces. In Figure 1d, is often, and quite rightly, considered as a principally
an increase in species 1 should lead to a decrease in different type of indirect effect. By coupling interaction
species 2 and an increase in species 3. The latter posi modifications with other types of relationships (e.g.,
tive effect would propagate up the right branch of the trophic), one may arrive at possibilities of numerous
diagram, increasing the abundances of species 4 and 5. (including very complex) relationships. One of the more
This situation arises when, for example, planktivorous simple of such combinations may be exemplified
fish preferentially feeding on large zooplankton indir (Figure 2) with an indirect effect of grazers and certain
ectly increase the abundance of small zooplankton. agricultural practices on the population density of foxes
Cases involving interference competition are well (Vulpes vulpes) and the rodent Marmota bobac in Eastern
known from, for example, the intertidal environment, Europe (V. Takarsky, personal communication): lower
84 Indirect Effects in Ecology

ecosystem components simultaneously take part in a

multitude of interactions, and it is therefore appropriate
to name it an interaction web. In fact, the number of
Grazing/hay making possible kinds of indirect effects is likely to be limited
only by the number of system components considered.

Vulpes vulpes Classifications of Indirect Effects

Although detailed analysis of various possible classifica
tions would be outside the scope of this publication, it is
worth mentioning, however, that indirect effects can be
characterized in a number of ways, related, for example,
to the characteristics of exerting, receiving, and transmit
ting compartments, presence/absence of a lag phase
Marmota bobac before the manifestation of a response, strength of the
interaction (particularly in relation to the direct interac
Figure 2 Diagram illustrating a positive indirect effect of tions) and its directionality (e.g., whether it is isotropic or
grazing on Marmota bobac population resulting from a anisotropic), dependence on a specific ecosystem context,
combination of consumerresource relationships with an
interaction-modification relationship. See further explanations in
importance for the functioning of the compartments
the text. involved, importance for structural (e.g., successional or
evolutionary) changes in the populations involved and
the whole biological community, and significance for
grazing rates lead to a denser and taller grass cover,
overall ecosystem functioning. In the authors view, the
enabling more successful hunting of predators.
different ways to characterize indirect interactions are not
Conversely, higher grazing rates lead to a lower grass
contradictory, but rather complementary, and may con
cover, thus enhancing the detection of predators by the
veniently contribute to the toolbox for comparative
rodents. As a result, increase in grazing may have an
ecosystem analysis.
indirect positive effect on the Marmota bobac population,
and an indirect negative effect on the population of foxes.
It should also be noted that some of the known examples Indirect Effects
of ammensalism and commensalism do actually fit in the All the relations not restricted to the effects of a direct
description either of a simple interaction modification or transaction of matter and energy between the adjacent
interaction modification coupled with a number of tropic ecosystem components are treated as indirect. Hence, for
relationships. For instance, the bioavailability example the purpose of the forgoing sections, all types of indirect
described above has been quoted by Atlas and Bartha as interactions mentioned above will be considered as indir
an example of commensalism. If, however, the chemical in ect effects. However, the distinction between directly and
question is not nutritional, but harmful for the second indirectly mediated effects will be made where deemed
species, then the relationship fits the criteria for ammens appropriate. The terms relationship and interaction will
alism. In a similar vein, protocooperative and mutualistic be used interchangeably. Furthermore, although it is rea
relationships are easily envisaged from certain combina lized that for the purpose of quantitative assessment the
tions of interaction modifications and tropic relationships. distinction between the terms effect and interaction
It is worth pointing out that although the indirect may be helpful, no such distinction has been made in
relationships listed above are mainly studied in relation this article, as in many studies addressing indirect effects
to pairs of biological species, they are applicable to a these terms are used interchangeably.
wider range of system components. It should also be The definition of indirect effects given above is very
noted that many more types of indirect effects are easily encompassing, and will include some of the effects which
envisaged from various possible combinations between may fall into the category of direct under a different
interacting compartments, and quite a few have indeed definition. For example, it is useful to account for the
been observed in nature. For example, Menge distin distinction between those effects that are directly and
guished 83 subtypes of indirect effects. However, an indirectly mediated, since the latter ones are particularly
attempt to exemplify every possible type of indirect difficult to observe, especially if the cause and effect are
effects would be outside the scope of this article. The substantially separated in time.
readers could easily construct, for example, many further The directly mediated effects have previously been
types of indirect effects combining the most commonly regarded as direct (i.e., as regards to the properties of their
studied ones depicted in Figure 1. In a real world, propagation). Here, however, the directly mediated effects
Indirect Effects in Ecology 85

will be treated as indirect, and the definition of indirect monitoring or experimental results. For example, Hunt
effects will, therefore, include such effects as trophic cas and co authors found that the increase in net N miner
cades, top down and bottom up controls, etc. The alization with precipitation is a consequence of not only
classification of indirect effects into directly and indirectly the direct effect of moisture supply on decomposition, but
mediated is applicable to a wide range of environmental also an indirect effect of changes in substrate supply and
processes and bears certain similarities with the distinction quality. de Ruiter and co authors studied nitrogen miner
between interaction chains and interaction modifications alization conducted at a wheat field. The impact of
earlier recognized for purely biotic relationships. microfaunal functional groups on N mineralization was
evaluated by calculating the impact of group deletion.
The results showed that the effect of the removal of a
Examples of Occurrence and Importance group may exceed the direct contribution of this group to
of Indirect Effects N mineralization rather considerably, with amoebae and
bacterivorous nematodes having values of 18% and 28%,
Indirect Effects in Terrestrial Environment
and 5% and 12% for, respectively, direct contribution
Arguably, the awareness of natural scientists as regards toward and impact of deletion upon overall N mineraliza
indirect effects in the terrestrial environment can be tion. Influence of the transitions of soil microorganisms
traced back at least to the end of nineteenth century, between dormant and active stages was studied by
when the school of thought founded by Dokuchaiev had Blagodatsky and co authors. Such transitions were shown
developed a theory that soil was a product of complex to be important for biogeochemical cycling and the rate of
interactions between climate and geological and biologi organic matter decomposition.
cal components of the terrestrial landscape. To date, the A combination of a detailed monitoring program, and
importance of indirect interactions in the terrestrial statistical and simulation modeling has been used in a
environment is well recognized. Indirect effects in terres study of ecological patterns in the Heron Wood
trial ecosystems relate, for instance, to the dependence of Reserve, located at the Dawyck Botanic Garden in
plant nutrient supply on mineralization of nutrients by Scotland. The suite of statistical techniques included
soil biota, and to the propagation of these effects through ANOVA, ANCOVA, correlation analysis, CCA, factor
the food chain. Soil fauna may help to disperse microor analysis, and stepwise regression modeling. The study
ganisms crucial for plant functioning and biogeochemical revealed a number of indirect effects resulting from a
cycling, and physically modify the habitat, thus changing complex multivariate interplay among ecosystem compo
environmental conditions for all the biological commu nents. For example, the results suggested that both direct
nity. Plants, in turn, modify the habitat for other negative and indirect positive effects of the microarthro
organisms, for example, by producing litter, providing pod community on specific fungal groups appeared to
shade, shelter, etc. All in all, indirect effects in the terres take place. The relatively high local abundances of the
trial environment are widespread; below are just a few dominant collembolan Folsomia might have caused local
examples of their recent studies. declines in ectomycorrhizal fungi, reflected, in turn, in the
A number of studies conducted in the terrestrial envi increase in pH (Whist this work was in press, Dr. Peter
ronment (this includes both field experiments and soil Shaw has checked identification of the dominant
microcosms) adopted experimental approach focusing Folsomia species (previously referred to as F. candida)
on the density manipulation experiments followed by from the Dawyck ecosystem study, and has shown that
analysis of the results obtained using parametric (e.g., it appears to fit the description of F. inoculata.) However,
ANOVA, Tukeys HSD) and nonparametric (e.g., for those samples where the dominant Folsomia were less
KruskalWallis and MannWhitney U tests) statistical abundant, overcompensatory fungal growth due to graz
tests. For instance, Miller used exclusion experiments to ing by mites and other collembola was implicated.
elucidate direct and indirect species interactions in a field Complex effects were also shown for bacteria, nematodes,
plant community. Experimental results were analyzed by protozoa, plants, and soil properties.
parametric and nonparametric techniques, which yielded
interesting information on the ecological characteristics
Indirect Effects in Aquatic Systems
of the species involved. Particularly, it was established
that species with a large competitive ability due to direct Awareness of indirect interactions in aquatic environment
effects generally had almost as large indirect effects, so has rather a considerably long history, and clearly pre
that the two effects almost cancelled each other. sented examples can be found in works (among others)
A number of terrestrial studies used various mathema of, for example, Mortimer, Hutchinson, and Reynolds. In
tical methods to investigate indirect interactions. In particular, in an earlier review by Abrams it was even
particular, a good insight into specific indirect effects suggested that most studies specifically addressing behav
was gained using simulation modeling to interpret ior mediated indirect effects tend to be conducted in
86 Indirect Effects in Ecology

freshwater ecosystems, while many of the early demonstra work on Rostherne Mere and suggest that the underlying
tions of density mediated indirect effects were done in mechanism might be a common inverse relationship
community studies in marine habitats. Likewise, much of between spring diatom and summer cyanobacterial
the knowledge related to indirect ecological interactions blooms resulting from the fact that the biogeochemical
has been contributed through the development and appli cycles of Si and P in the aquatic environment are coupled
cations of the methods of simulation modeling and network via the dynamics of primary producers (i.e., increased
analysis in relation to aquatic environment. Consequently, concentrations of Si in spring lead to an increase in a
simulation models capable of demonstrating indirect inter spring diatom bloom, and an increase in the removal of
actions in aquatic biogeocenoses (e.g., the Lake 2 model of P, N, and microelements from the water column with
J. Solomonsen) are widely used for teaching in the educa easily sedimenting biomass at the end of the bloom; con
tional establishments across the world. sequently, this may lead to a decrease in the summer
Recent studies of indirect effects in aquatic environ cyanobacterial development).
ment variously involved a combination of the empirical
approach and an application of statistical techniques,
Role of Abiotic Components
methods of network analysis, simulation modeling using
What if scenarios, and sensitivity analysis. One of the Although the importance of abiotic ecosystem compo
perhaps most frequently addressed examples of indirect nents is commonly recognized, most of the ecological
effects in aquatic environment relate to trophic cascades, studies (including those addressing the indirect effects)
which involve propagation of the effect along a vertical tend to study in detail only relationships among biota.
trophic chain consisting of three or more components The restriction of the integrative synthesis to species
connected by grazing or predation. For instance, as was interaction only cuts off a plethora of useful environmen
recently investigated by Daskalov, a decrease in the top tal studies related, for example, to issues of global climate
predators population in the Black Sea due to overfishing change. It should be noted, however, that the science of
resulted in a trophic casade, leading to an increase in the ecosystem dynamics is highly interdisciplinary, and the
abundance of planktivorous fish, a decline in zooplankton information relevant to the present discussion can, there
biomass, and an increase in phytoplankton crop. fore, be found not only in ecology and biology, but also
The previously made statements regarding the abiotic virtually in any section of natural and environmental
components (see above) can be emphasized with exam sciences, with geography, palaeontology, geoecology,
ples related to the importance of detritus. For instance, and climatology comprising the most obvious candidates.
Carrer and Opitz found that in the Lagoon of Venice In ecology, it is widely recognized that species inter
about half of the food of nectonic benthic feeders and action can be mediated by a nonliving resource, and that a
nectonic necton feeders passed through detritus at least species can potentially exert a selective force on another
once, while there was no direct transfer of such food species through nontrophic interactions. It should also be
according to the diet matrix. Whipple provided an analy noted that in nature many species are very well adapted
sis of the extended path and flow structure for the well to modify their community and habitat (e.g., beavers by
documented oyster reef model. Few simple paths and changing the habitats hydrological regime, humans by
large number of compound paths were counted. The initiating dramatic changes in global climate and geo
study provided structural evidence for feedback control chemical fluxes, earthworms by increasing aeration and
in ecosystems, and illustrated importance of nonliving redistributing organic matter in soil, etc.). Changes in
compartments (in this case, detritus) for the ecosystems physical characteristics of a habitat caused by the activity
functioning. Even for the model with a low cycling index of so called ecosystem engineers may be regarded as an
(i.e., 11%) multiple cyclic passage paths provided a con extreme case of such nontrophic interactions. Often, how
siderable (22%) flow contribution. Therefore, it was ever, even if abiotic components are considered in terms
envisaged that for ecosystems with higher cycling indexes of detrital pathways and/or nutrient cycling, the effects
the patterns observed should be even more pronounced. studied in detail are mostly confined to trophic interac
Another noteworthy illustration of indirect effects in tions only. Furthermore, many indirect interactions occur
aquatic ecosystems relates to the interdependency of bio between different stages of ecosystem development and
geochemical cycles. For example, Dippner concluded that are therefore easily overlooked and understudied. In eco
indirect effect of the silicate reduction in coastal waters logical literature these interactions are sometimes called
causes an increased flagellate bloom, due to a high avail historical effects, priority effects, or indirect delayed
ability of riverborne nutrient loads. In a study of lake regulations. Consideration of these effects is particularly
Suwa (Japan), Naito and co authors have shown that the important for the correct understanding of an overall
physiological parameters of the diatom Melosira were the ecosystem functioning. Hence, if one abstracts from the
important sources of the cyanobacterium Microcystis pro labels given to different branches of science, the impor
duction variability. These results agree well with our tance of abiotic ecosystem components and physical
Indirect Effects in Ecology 87

environment for ecosystem dynamics and evolutionary still one of the most discussed topics in ecology and
development becomes increasingly obvious. environmental sciences in general. While the detailed
review and the lively controversy of the discussions
related to this topic is outside the scope of this publica
Indirect Effects of Global Relevance
tion, it is worth pointing out that the absolute majority of
Indirect relationships important on the global or subglobal studies dealing with it also inevitably deal with indirect
scale are often separated from their cause spatially and/or effects (although the exact term is often not mentioned).
temporally. For example, the dramatic increase in volcanic
activity (possibly caused by the impact of an asteroid) at
Indirect Effects and Industrial Ecology
the end of the Mesozoic era is thought to have led to the
extinction of dinosaurs, which arguably stimulated the This article would be incomplete without mentioning of
eventual evolution of mammals (including humans). The studies and methods used in industrial ecology.
increased production and use of fertilizers in the 1950s led Industrial ecology is based on the analogy between
to the increased phosphate inputs, eutrophication, and natural and industrial ecosystems, and aims to facilitate
decrease in water quality in many lakes, ponds, and reser the development of industrial recycling and cascading
voirs during the subsequent decades. The increased cooperative systems by minimizing the energy consump
consumption of fossil fuels in the twentieth century led to tion, generation of wastes, emissions, and input of raw
the increased emissions of carbon dioxide, which were materials. Complex interplay among system components
eventually followed by global warming and an apparent has been taken into account in a large number of
increase in the frequency of natural disasters. This climate waste management and industrial ecology studies.
change was probably accelerated by the depletion of the Consequently, throughout the second half of the last and
planets ozone layer due to the CFC (chlorofluorocarbon) the beginning of the present century, some substantial
containing deodorants and refrigerants. progress has been made in various aspects of industrial
It should be noted that indirect relationships are not ecology, and in particular in understanding and account
related just to the activities of humanity, but have been ing for indirect effects.
important throughout the history of our planet. For exam One of the commonly used methods of industrial ecol
ple, a gradual development of the modern atmosphere ogy is life cycle assessment (LCA). It studies the
was largely due to the activity of cyanobacteria, which environmental aspects and potential impacts throughout
were among the first organisms to produce oxygen as a a products life (commonly referred to as cradle to grave
by product of their metabolism. The indirect implications approach), from raw material acquisition through produc
of the atmospheric oxygen enrichment were far reaching, tion, use, and disposal, and the same methodological
and led not only to profound global biological and geo framework allows analysis of the impacts associated with
chemical changes, but also ultimately enabled the physical products (e.g., cars, trains, electronic equipment),
development of Homo sapiens and its current civilization. and services such as waste management and energy sys
Last century, the line of thought started by Vernadsky tems. Similar to LCA, but usually with considerably
has eventually led to the creation of a new integrative narrower system boundaries, are methods of energy
branch of natural sciences, sometimes referred to as global analysis, including, for example, energy footprinting
ecology. Essentially, global ecology encompasses meth (which, effectively, constitutes calculations of how much
ods and scope of virtually all other environmental energy is spent and saved/recovered in all the processes
disciplines, and is predominantly concerned with the included within the chosen system boundary) and net
dynamics (including past and future) of the global ecosys energy analysis (which in addition to the detailed energy
tem the biosphere. As an example, it is worth mentioning budgeting involves calculation of indicators such as incre
the now classic climatological research carried out by mental energy ratio and absolute energy ratio). For
Budiko and co workers, which led to the creation of a example, on the basis of the energy budget estimates
half empirical model of the thermal regime of the atmo for case studies from the UK and Switzerland it has
sphere. This model was subsequently used to simulate past been argued that increasing recycling rates for plastic
and future dynamics of the atmosphere, and changes and glass would improve the energy budget of waste
between glaciation and interglacial periods. Furthermore, management programmes, and, therefore, benefit the
the results obtained aided interpretation of human evolu corresponding industrial ecosystems. Further modifica
tion, and led to further research aiming to counteract tions of the energy analysis methods make fruitful use of
possible global change, for example, by injecting certain emergy and exergy budgets.
substances into the stratosphere, and direct and indirect Another method popular in industrial ecology is eco
consequences to which such manipulations may lead. logical footprinting. Basically, the method estimates the
Currently, global climate change (principally related area necessary to support (i.e., in terms of, for example,
to the increased concentrations of greenhouse gases) is production of food, energy, processing of wastes) current,
88 Indirect Effects in Ecology

past, or probable future functioning of particular geogra Within the theoretical approach, observations (and/or
phical (often administrative, for example, countries, carefully considered experimental data) are used together
counties, towns) units. Despite numerous logistical pro with theoretical considerations to construct a model cap
blems of interconversions, system boundary definitions, able of investigating interactions among the components
and coefficient estimates, application of this method is incorporated in the model structure. This model is sub
very useful and illustrative. For example (as illustrated by sequently used to examine indirect effects between the
Herendeen), out of all Western industrialized countries, components. There are a number of drawbacks of this
only the ecofootprints of Australia and Canada appear to fit approach, for example, difficulties related to obtaining
inside their borders (the rest of the developed countries sufficient details about the components represented in
appear to live on the expense of other territories). the models, unavoidable uncertainty as regards fluxes,
parameters, initial values, etc. This uncertainty may
mask the significance of the relationships studied, includ
Evolutionary Role of Indirect Effects
ing indirect effects. Furthermore, as it is impossible to
It has been postulated by a number of authors, and has reproduce all the complexity of a real ecosystem, any
been proved mathematically by Fath and Patten, that model is a simplification of reality. Therefore, some of
indirect effects often promote coexistence and the role the potentially important interactions may be lost just by
of indirect effects should, in general, increase in the defining the model structure, while the importance of the
course of evolution. For example, in grassland commu others may be considerably altered.
nities containing Rumex spp., insect herbivory (by Within the experimental approach, densities of indivi
Gastrophysa viridula) appears to be a cost inherent in the dual species are manipulated (e.g., by total removal) in
development of plants resistance to pathogenic fungi microcosms or experimental plots, and statistical analysis
(Uromyces rumicus). Another example relates to the fact (e.g., ANOVA, ANCOVA) are subsequently applied to
that infection of plants with endophytic fungi often estimate the magnitude of indirect effects of manipula
enhances plants competitive abilities via deterring gra tions on densities of other species. It has been argued that
zers by production of toxic compounds (as a result, some this approach is best applied using a factorial design,
plants might have coevolved together with their endo where the densities of a number of components (e.g.,
phytes, for example, coupled evolution of Festuca and species or trophic groups) are changed both alone and in
Acremonium spp.). combination. If implemented properly, this approach
It should be noted, that indirect effects are important leads to a straightforward estimation of net effects.
for the evolution of not only natural, but also industrial However, there is always a danger that some of the
ecosystems. Traditionally, human society has developed indirect interactions have not manifested owing to un
without the necessary due respect to the rules and pro avoidable time constraints of any experiment. Also,
cesses governing the stability of its environment. partitioning of the registered net effects may be subject
However, by analogy with natural ecosystems (i.e., as to speculation. Experiments are often costly and by defi
regards recycling and cascading networks) industrial eco nition are limited by their design and the hypotheses
systems should aim to facilitate the development of tested. The simplicity of the experimental design may
recycling and cascading cooperative systems by minimiz mask the significance of the relationships studied for
ing the energy consumption, generation of wastes, trait mediated effects; measurements of population abun
emissions, and input of raw materials. dances may need to be supplemented by behavioral
observations, and/or biochemical, physiological, genetic,
and other analyses. Furthermore, there is always a big
Approaches and Techniques Used to question mark how applicable are the results obtained to
Detect and Measure Indirect Effects the processes happening in the real world.
Among mathematical methods which have been used
Detection and measurements of indirect effects are often far in studies of indirect effects in natural ecosystems are
from straightforward, and are mostly based on the intuition, statistical methods (e.g., regression and correlation analy
common sense, and prior knowledge of any particular sys sis, PCA, factor analysis, CCA, ANCOVA, ANOVA),
tem. Abrams and co authors described two major simulation modeling (e.g., using what if scenarios, sensi
approaches adopted in ecological studies, namely theoreti tivity and elasticity analysis), and methods of network
cal and experimental. They stated that in practice, the analysis. In particular, indirect interactions have often
theoretical and empirical approaches may be regarded as been analyzed using methods of network analysis. For
endpoints of a methodological continuum. Recently, how example, Fath and Patten used methods of network
ever, we have argued that the methodological continuum to analysis to show that, in the ecosystem context, direct
study indirect interactions is best represented by a triangle, transactions between organisms produce integral effects
with observational, experimental, and theoretical nodes. more positive than a simple sum of direct effects. This was
Indirect Effects in Ecology 89

in line with the view that mutualism is an implicit con seemingly appropriate environmental management meth
sequence of indirect interactions and ecosystem ods, and in particular biomanipulation, resulted in failure.
organization, and that the contribution of positive rela
tionships should increase along the course of evolution
Separation in Time
and ecological succession.
It should be noted that all the methods so far applied to Another problem related, in particular, to the detection of
investigations of indirect effects have both advantages and indirect effects, is the fact that they often occur after a
limitations. Many of these have been previously addressed considerable time lag. Unfortunately, research grants are
and no attempt to discuss the benefits and disadvantages of typically no longer than 3 years (often less). Hence it is
the techniques used to investigate indirect interactions has likely that in many studies the potential for indirect
been done in this article. Neither was it intended to address effects has been considerably underestimated (if not
any controversy and related discussion resulting from spe totally overlooked).
cific applications (and/or implications of such applications) Not only are cause and effect often separated in time,
of any particular method. It should be noted, however, that but they also may occur at different stages of the systems
the methodological framework of comparative theoretical succession. This, however, may be turned to human
ecosystem analysis (CTEA) (see below) suggests that the advantage, and used as a complementary measure of
mathematical techniques may be best used in concert, thus environmental control. For example, it has been shown
allowing a detailed complementary insight into complex that in freshwater lakes and reservoirs biogeochemical
patterns of mechanisms underpinning dynamics of natural cycle of P may be regulated by alterations in the biogeo
ecosystems. chemical cycle of Si, and stimulation of the spring diatom
growth may help to alleviate nasty cyanobacterial blooms
in summer. Clearly, due to inherent problems with eco
Problems and Implications for system complexity and uncertainty any application of
Environmental Management environmental management measures utilizing time lags
of indirect effects should be done with extreme caution.
There are many problems associated with studies of
indirect effects. Here we list the most general ones, in
Separation in Space
the authors view, resulting from the very nature of such
relationships, and the complexity of natural environment. Geographical separation between the cause and the effect is
We also emphasize the potential of using indirect effects another inherent problem of indirect effects. For economic
in environmental management and caution as regards and societal reasons, it is particularly well documented in
their misuse and careful consideration. studies and practical applications of pollution and biologi
cal control, but numerous examples are readily available
from virtually any areas of ecology and environmental
Complexity and Uncertainty
sciences. As regards pollution damage, the well known
Although the characteristics of indirect effects are examples relate to coastal and benthic effects of oil spills,
fairly readily established in a controlled laboratory experi consequences of lake acidification in Scandinavia due to the
ment involving a very limited number (typically <5) of transboundary transfer of sulphur dioxide since the begin
interacting components, in the natural environment the ning of industrial revolution, bioaccumulation of
complexity of interactions renders their characterization, radionuclides in the food webs of, for example, British
in particular in practical terms as regards the outcome for pastures following the Chernobyl fallout, etc.
any specific ecosystem, rather extremely challenging. In ecology, there is a growing awareness of placing the
Using the inverse of 100 community matrices (elements dynamics of a local biological community in the broader
of such an inverse matrix specify how the change in density landscape and biogeocoenological context. For example,
of a particular species affects the density of the other in biocontrol applications, failure to anticipate indirect
species, with species pairs determined by the elements effects resulting from the community openness and
position in the matrix) obtained by randomly changing exchange between localized subpopulations has led to a
the species specific parameters, Yodzis has shown that a number of costly failures and unexpected outcomes.
large number of predictions for overall interactions Following introduction, migrations of biocontrol agents
between separate species were directionally undetermined, were shown to be capable of suppressing nontarget spe
that is, positive in some cases and negative in others. Owing cies in spatially removed areas. Certain traits of the agent
to such an uncertainty, some theoreticians have even ques were shown to be related to elevated risks of nontarget
tioned whether credible predictions of the overall outcome species extinction, for example, the risk of extinction of a
of indirect interactions could be made at all. As a conse nontarget species increases with the decrease of its growth
quence, there are numerous cases where the application of rate and the increase of the agents attack rate compared
90 Indirect Effects in Ecology

to the target species. Hence the threat of extinction algae, infection of grass cultivars by endophytic fungi in
appears to be particularly acute for a nontarget species turf industry, etc. It should be reiterated, however, that a
occupying habitat fragments co mingled with the habitats sound knowledge of the systems natural history is abso
occupied by target species, and when the target species is lutely indispensable for an application of any
relatively productive and the agent is only moderately environmental control measure. The problems related to
efficient at limiting the target species numbers. the detection and investigations of indirect effects (with the
major ones listed above) are likely to provide challenging,
Defining System Boundaries and often unexpected complications, frequently after a
considerable time period. Hence, a combination of empiri
Closely related to the separation problem is the problem of cal and theoretical work should precede any practical steps,
drawing the systems boundaries. By narrowing the bound and any desk study should be backed up by a thorough
aries of the system considered, many of indirect effects may monitoring plus (where necessary) experimental program.
be reduced to the effects attributable to the variation in
external forcing. This is often convenient, in particular, if
the aim is to study responses in a controlled environment.
However, the scope for investigations of indirect effects in a Current and Further Directions
system with narrow system boundaries inevitably becomes
limited; by widening the boundaries and including more Further investigations of indirect effects are important both
components the potential for indirect effects is greatly for enhancing our understanding and therefore improving
enhanced, but as the systems complexity increases, so management of specific ecosystems, and for general
does the associated uncertainty of any interpretations!
development of ecology. Due to the increasing pace of
technological progress, collection of monitoring data
Implications for Environmental Management using automated techniques, and in particular remote sen
sing, is becoming increasingly easy. Combined with rapidly
Qualitative and quantitative account of indirect effects is
increasing computing power and progressive development
(albeit often implicitly) becoming a common part of
of mathematical methods, this may provide the necessary
environmental management, and is indispensable for suc
basis for a dramatic acceleration of investigations of indir
cessful application of, for example, landscape engineering,
ect effects, in particular the ones manifesting on a global
biomanipulation, biogeochemical manipulation, strategic
level, and in cases when the geographical separation is an
environmental assessment (SEA), and environmental
issue. With progressive accumulation of long term data
impact assessment (EIA). In particular, mathematical
methods can be helpful in this respect. For example, sets, it is also likely that the effects occurring after a time
Ortiz and Wolff used Ecopath with Ecosim software to lag will become more readily discernable. To maximize the
study benthic communities in Chile. They found that a benefits (i.e., for investigations of indirect effects) from the
simulated harvest of the clam Mulinia generated a com technological development, however, it should be supple
plex interplay involving direct and indirect effects, and mented by contemporary advances in the methodology of
drastically changed the properties of the whole system. their investigations.
McClanahan and Sala used a simulation model of the It has been argued that analysis of indirect relation
Mediterranean infralittoral rocky bottom to study possi ships in ecosystems may be greatly enhanced by the
ble effects of various management options. Running a application of a specialized methodological framework,
number of what if scenarios they concluded that many called CTEA. CTEA is aimed at bringing together sepa
of potential changes are likely to be indirect effects caused rate lines of current investigations, hence combining them
by changes in trophic composition. For example, if inver in an integrative approach (see the section titled Further
tivorous fish were removed as part of a management reading). Further development and systematic applica
scenario, sea urchins would reduce algal abundance and tion of CTEA is vital for improving the accuracy of
primary production, leading to competitive exclusion of ecological forecasting, and has, therefore, potential socie
herbivorous fish. Although similar interactions were tal benefits related to issues of EIA and sustainable
known from tropical seas, these results were not antici development. It is suggested that further developments
pated by previous field studies in the Mediterranean. should pay much attention to similarities and differences
In some cases appreciation and reliance on indirect of the indirect effects revealed in various types of ecosys
effects may form the basis of environmental management tems, or at different stages of the ecosystem development,
measures. To date, there are numerous relevant examples, and that the characteristics (e.g., magnitude, sign, etc.) of
including, for example, biocontrol, bio and biogeochem indirect interactions should be increasingly used for
ical manipulation or application of chemicals to reduce describing differences in ecosystem state, structure, and
sediment P release for subsequent control of blue green overall functioning. For example, analysis of specific
Emergent Properties 91

ecosystems may benefit from answering the following (to Fath BD and Patten BC (1998) Network synergism: Emergence of
positive relations in ecological systems. Ecological Modelling
name but a few) questions: 107: 127 143.
Fath BD and Patten BC (1999) Review of the foundations of network
What types of indirect effects are important for the overall
functioning of an ecosystem under investigation?
environ analysis. Ecosystems 2: 167 179.
Fleeger JW, Carman KR, and Nisbet RM (2003) Indirect effects of
How does the importance of indirect effects compare
with the importance of direct effects?
contaminants in aquatic ecosystems. Science of the Total
Environment 317: 207 233.
Herendeen RA (1998) Ecological Numeracy. Quantitative Analysis of
the pattern of indirect effects relatively constant, or
to (system specific) seasonal and longer term
Environmental Issues. Toronto: Wiley.
Kawanabe H, Cohen JE, and Iwasaki K (1993) Mutualism and
Community Organisation: Behavioral, Thoretical and Food Web
changes? Approaches. Oxford: Oxford University Press.
How does the pattern of indirect interactions change
due to pollution, disturbance, and various management
Korhonen J (2001) Industrial ecosystems Some conditions for
success. International Journal of Sustainable Development and
World Ecology 8: 29 39.
practices? Krivtsov V (2004) Investigations of indirect relationships in ecology and
Do indirect interactions predominant in an ecosystem
help to stabilize this ecosystem?
environmental sciences: A review and the implications for
comparative theoretical ecosystem analysis. Ecological Modelling
174(1 2): 37 54.
What is the relative contribution of indirect interac
tions to resistance, resilience, and facilitation of
Menge BA (1995) Indirect effects in marine rocky intertidal interaction
webs Patterns and importance. Ecological Monographs 65: 21 74.
successional changes? Miller TE and Travis J (1996) The evolutionary role of indirect effects in
communities. Ecology 77: 1329 1335.
How have the indirect effects changed during the
evolution of a particular ecosystem, and what was
Patten BC, Bosserman RW, Finn JT, and Gale WG (1976)
Propagation of cause in ecosystems. In: Patten BC (ed.) Systems
their contribution toward the driving forces of this Analysis and Simulation in Ecology, pp. 457 579. New York:
Academic Press.
evolution? Schoener TW (1983) Field experiments on interspecific competition.
American Naturalist 122: 240 285.
Strauss SY (1991) Indirect effects in community ecology Their definition,
study, and importance. Trends in Ecology and Evolution 6: 206 210.
Further Reading Wardle DA (2002) Communities and Ecosystems. Linking the
Aboveground and Belowground Components. Princeton: Princeton
Abrams PA, Menge BA, Mittelbach GG, Spiller DA, and Yodzis P (1996)
University Press.
The role of indirect effects in food webs. In: Polis G and Winemiller K
Wootton JT (1994) The nature and consequences of indirect effects in
(eds.) Food Webs: Integration of Patterns and Dynamics,
ecological communities. Annual Review of Ecology and Systematics
pp. 371 395. New York: Chapman and Hall (also see other papers in
25: 443 466.
this book).
Wootton JT (2002) Indirect effects in complex ecosystems: Recent
Budiko MI (1977) Globalnaya Ekologiya (Global Ecology). Moscow:
progress and future challenges. Journal of Sea Research
Misl(in Russian).
48: 157 172.

Emergent Properties
F Muller, University of Kiel, Kiel, Germany
S N Nielsen, Danmarks Farmaceutiske Universitet, Copenhagen, Denmark
2008 Elsevier B.V. All rights reserved.

Introduction Classification of Emergent Properties

The History of the Concept Quantifying Emergence
Emergence and Hierarchy Further Reading
How Emergence Emerges

Introduction or quantities, beyond easily measurable or predictable phy

sical parameters. The resulting properties have been
Many biologists will recognize the statement that the whole described on many levels of the biological hierarchy, from
is more than the sum of the parts as a commonly used (but simple physical systems, like laser beams, to the organization
hardly understood) phrase. This formulation refers to the of the whole biosphere within the Gaia concept of
idea that there are systems which possess additional qualities J. Lovelock. The emergent property at one level is in general
92 Emergent Properties

finding its causality at the subsystem components and the bound to happen that are not easy to predict from the
interaction between them. For example, an organized form basic knowledge of the system, no matter how extensive
of cell functions, stemming from self organized transforma this knowledge is.
tions of cellular compounds, known as hypercycling may be Highly relevant to biology and ecology is the question
considered as an emergent entity; at the physiological levels when an emergent property appears. This leads to
we are, for example, dealing with the mating behaviors of the distinction of primary and secondary emergence,
organisms as results of hormone interactions. Similarly, primary emergence being the first time an emergent prop
motion, feelings, or intelligent behavior occur as a conse erty appears. To be conserved the property can
quence of special couplings of neurons. In addition, the be reproduced again and again but in this case it is nomi
patterns in the development of ecosystems may not be nated as a secondary emergence. Recent approaches to
predictable from knowledge of organisms alone. Therefore, emergence have come up with three further notions
emergent properties are not unusual phenomena, but simply of emergence: computational emergence, thermodynamic
consequences of hierarchical organizations. emergence, and emergence relative to a model. The com
The concept of emergence found its way into ecology putational emergence deals with the patterns produced by
through the proposal of E. P. Odum, who suggested that the different computer programs, for example, cellular auto
study of emergence should lead to a new integrative dis mata systems developing complex distributions out of
cipline. This idea was due to the fact that studies of simple rules from game theory. Thermodynamic emer
complex systems had shown that the investigations of the gence covers the establishment of highly complex, self
details alone were not adequate in predicting ecosystem organized structures and their relations to the nonlinear,
function and behavior. Neither were they sufficient to far from equilibrium thermodynamics. Emergence relative
explain a more advanced pattern like behavior and perfor to a model defines emergence as the deviation of the actual
mance of ecosystems, for example, during the succession behavior of a physical system in comparison with an obser
from young systems toward more mature states. vers model of it.
Summarizing these historical notions of emergence,
the following features can be stated:
The History of the Concept
Emergent properties are properties of a system which
are not possessed by component subsystems alone.
The concept of emergent properties originates in the
nineteenth century, finding the primary roots back in The properties emerge as a consequence of the inter
actions within the system.
Kantian philosophy. The term was coined by G. H.
Lewis as far back as 1875. A common definition from Two fundamental types of interactions are found that
may be characterized as intra and inter connected
that time states that emergence is the denomination of
ness, that is, connections within and between levels,
something new which could not be predicted from the
including controls. This point does not consider the
elements constituting the preceding condition.
direction of the intra level interactions. Emergence is
Throughout the last century, several scientists have
based on both, upward and downward causation.
addressed the concept from a more philosophical point of
view, resulting in the appearance of different descriptions The historically emerged properties are considered
new with reference to their primary appearance.
and explanations. The definitions have, in general, been
referring to subjective arguments, such as surprise, unex These new properties appear at one level of a sys
tem and are not immediately deducible from
pectancy, thus being clearly observer dependent. This has
observation of the levels or units of which the sys
strongly influenced the present approaches, and this com
tem consists.
prehension has often been connected with a flavor of
mysticism. Thus, the seriousness of the concept has
often been underestimated.
During the last decades, the use of the term emergent Emergence and Hierarchy
properties has found widespread use in biological
sciences, especially, because it is clearly connected with Emergence has been described at many levels of the
the growing implementation of the system approach in biological hierarchy. As argued above, the reason for
ecology. The need for a holistic concept was due to the emergence is to be found in the hierarchical organization
failure of the traditional reductionistic research strategies of the system and the quantitative and qualitative char
to explain the properties of ecosystems by the knowledge acters of the linkages within the structure. As biological
of the behavior and the properties of the ecosystem con structures are often complex, this makes it hard to deter
stituents alone. Ecosystems are highly complex middle mine the actual cause of emergence.
numbered systems dominated by nonlinear relationships Hierarchy theory (see Hierarchy Theory in Ecology)
between their constituents. In such systems, things are states that middle numbered systems such as
Emergent Properties 93

ecosystems can be comprehended if they are investi mentioned to characterize situations of primary
gated on different levels of integration. Broad scale emergence.
levels can be assigned to high spatial extents, and low Many examples found in the literature deal with the
typical frequencies, filtering the signals from lower formation of the earliest cells. Biochemical cycles, the
levels. These scale levels are spatially smaller and organization and exchange of information by DNA or
their typical frequencies are higher. They are not able RNA and the compartmentalization of material within
to filter constraints from the higher levels, but their membranes are but a few examples. Molecular comple
potentials and interactions are building up the material mentarity, defined as nonrandom, reversible coupling
basis and the coordination functions of the higher level. of the components of a system, has been argued to be a
Emergent properties are created by both types of non widespread mechanism in biological systems and impor
linear interactions. Therefore, the properties of specific tant for the understanding of the processes lying behind
levels can be termed hierarchical emergent properties. emergent properties. The seemingly (self)organization of
Of course, the interesting question is how these proper molecules observed in prebiotic systems, such as
ties emerge. Some examples might be helpful to Turing structures and autocatalytic hypercycles (see
illuminate this question. Autocatalysis), can be seen as emergent properties already
at a very low level of organization.

Prebiological Emergence
Emergence in Biological Systems
Several examples of emergent properties can be found
Emergent properties really come into play when biologi
in physical and chemical sciences. They form an impor
cal systems reach higher levels of complexity. This
tant prerequisite for protobiology and evolutionary
becomes evident already when cells or groups of cells
processes. Within the area of physics some examples
communicate with each other as in the case of hormones
are nearly classical: Water (e.g., its wetness), which is a
and natural neural networks. Organs are composed of
simple molecule with a rather complex behavior, that is
cells, their individual functions are important only to
unpredictable from knowledge about oxygen and hydro
the organism as a whole. A heart, kidneys or lungs, are
gen alone, has often been used to demonstrate emergent
vital but their function is not existent when they are on
properties. Similarly, the sense of colors by the eyes is
their own. Organisms interacting as populations or socie
not predictable by knowing a certain wavelength of
ties provide properties which cannot be explained by
properties of the individual organisms alone. They all
Two famous chemical examples related to self
go together in what we consider as ecosystems and thus
organized behavior of systems may also be mentioned,
are a part of the biosphere.
the Benard cells and the BhelusovZabotinsky (BZ) reac
Cellular level. In regarding the outcome of interacting cells
tion. In the case of Benard cells, during specific
many studies have been concentrating on the organization of
conditions, hexagonal, convective cells (the emerging
neuronal systems, which result in unexpected properties like
structures) form a fluid when a thermal gradient is
the ability to move, to sense, to be intelligent, and to emote.
imposed on the experimental setup containment. In the
The sensory systems, being connected to visual, auditory, or
BZ reaction a special ratio of chemicals causes a mixture
other communicative processes are all playing a major role
to perform a pulsing pattern in colors with a period of
in how successful living organisms are in performing specific
about one minute. The structure of these physico
life strategies. Reliable senses, and responding the right way
chemical processes, gradients resulting in convective
to the received stimuli are crucial to the existence of many
cells, pulsing patterns, together with other observations
life forms, in processes like finding food, knowing when and
like the occurrence of Turing structures in chemical fluids,
where to escape, or creating bonds with other members of
spontaneous formation of lipid coacervates, might be cru
the species, for example, during reproduction.
cial to our understanding of the emergence of life.
Neural networks, like in our brains, consisting of a huge
number of interconnected neurons, are so complex that
unforeseen patterns in responses are bound to occur and
Protobiological Emergence
have also been reported to exist. During the evolution of the
The appearance of the earliest life forms has often been brain, emergent properties, together with new cell types,
referred to as a primary emergence. Although many of the local and large circuits have added up to the increasing
properties occurring during this phase of evolution have complexity of brain function. Motor control, the control
been repeated, over and over again their appearance and coordination of motor activity are taken care of by our
still qualifies them as emergent properties. As examples, brain passing on signals to the limbs or organs involved.
the emergence of life, emergence of animals, or the Organ level. Numerous cells, often during morphogen
emergence of bird feathers from reptile scales can be esis differentiated in certain, specialized directions, form
94 Emergent Properties

organs, take up a particular task of the organism, like for explain this behavior. The resulting oscillations were
instance liver cells secreting enzymes, or kidney cells found to be emergent properties of the colony.
filtering and cleaning the coelom. Although the formal Ecosystem level. Ecosystems are inherently complex as they
layout for this functionality is existent in the genetic are composed of an embedded hierarchy of all the pre
material of all cells, the eventual determination occurs viously mentioned subsystems in close interaction with
during the development of the organism and the actual abiotic factors. Emergence is to be expected, but surprisingly
function of the organs may be viewed as emergent. The few reports exist at this level, before all analyzing micro
brain as an organ may serve as an example of cosms, forest ecosystems, predatorprey relationships, food
this emergence: Here differentiated cells, with highly webs, and the organization of aquatic communities.
specialized physiological properties, go together and cre Ecosystem behavior is often analyzed through model
ate activity patterns that are far more complex than ing studies. The relation to emergent properties becomes
expected from knowing the physiology of neural cells clear when looking at recent efforts of structural dynamic
alone. The whole becomes more than the sum of the parts. modeling, where the changes in ecosystem composition
Organism level. Complex behavior occurs among the indi and structure over time are analyzed. Another example is
vidual organisms that cannot be determined exclusively by the work of B. C. Patten on the propagation of matter
internal factors. The sending, reception and interpretation of energy through the ecosystem network, leading to the
signals from interagent organisms, the relationship(s) to the discovery of the importance of indirect effects, quantita
outside, and thus semiotics play an important role, creating tive and qualitative utilities of the system, results that
patterns impossible to foresee if only the subsystems are are highly surprising and unexpected, and as such are
known. For example, in trees, the formation of branches emergent properties (see Ecological Network Analysis,
and leaf mosaics have been studied in a number of recent Environ Analysis). Both examples link to higher level
investigations with modeling approaches as well as the allo information expressions such as ascendency, different
cation of resources between above and belowground kinds of entropy or information derived descriptors like
biomass and the related physiological mechanisms. A mod exergy (see Exergy).
eling study of this problem indicates a complex integrated The ability of the ecosystem to perform with systema
growth pattern which may only be understood as an emer tic directional changes in some macroscopic characters,
gent property as it is claimed to have no direct or indirect not predictable from knowledge about the single ecosys
mechanistic basis related to subcellular activities. In a similar tem members alone, has been discussed since 1967 on
manner it was shown that whole plant behavior is an emer the basis of the 24 principles of ecosystem development
gent property arising from a rule based model of the system. during succession in the second edition of E. P. Odums
Communications between individuals, that is, their social Fundamentals of Ecology. Many other factors, known as
interactions within a population, are important to the func indicators, orientors, or goal functions have been pre
tion of the organism as a whole and are indistinguishable sented since then (Table 1).
from the emergence of ethological features. Stressing the
importance of communication, may lead to an interpretation
of the communicative process as an emergent interpretation
of signs, which is described within the discipline of semiotics. How Emergence Emerges
Population level. Populations are composed of individual
organisms, interacting in various ways, differing in quan The concept of emergent properties refers very clearly to,
tity and quality, throughout the biological system. The and must be seen in tight connection with, at least two other
interactions may vary in character according to the com concepts often occurring in literature on modern ecosystem
plexity. At the one end of the spectrum, we find the single theory, the concepts of hierarchy and self organization.
cell organisms interacting mostly on a material basis In connection with hierarchy, the emergent properties
(matter fluxes). At the other end, there are colonial organ
are seen as outcomes of ecosystem organization where
isms forming complex societies, where brains, senses,
supersystems are formed with subsystems as constituents
memory, and thus informational interaction become
and where the properties are observable at the supersystem
dominating. Emergent properties as a result of individual
level only. Here the emergent property is an outcome of a
level behavior and interactions in populations of social
certain way of organization. To exemplify this point, we
insects have been argued in several studies. For instance,
might look at the following hierarchical features:
the distribution of food to larvae of the fire ant has been
argued as emerging from interactions between indivi 1. Individual level: individual nutrition budgets fora
duals, workers, and larvae. Cellular automata models ging strategies.
were used to study the short time oscillations in ant 2. Population level: species nutrition efficiencies intras
colonies. The nonlinear dependencies describing the rela pecific food competition.
tionships between, and the movement of, individuals 3. Ecosystem level: nutrient cycling food webs.
Emergent Properties 95

Table 1 Some ecosystem orientors

Immature state Mature state

Properties of the dominating species

Rapid growth Slow growth
R-selection K-selection
Quantitative growth Qualitative development
Small size Large size
Short life spans Long life spans
Broad niches Narrow niches
Properties of production
Small biomass Large biomass
High P/B ration Low P/B ratio
Low respiration High respiration
Small gross production Medium gross production
Properties of nutrient flows and cycles
Simple, rapid, and leaky Complex, slow, and closed cycles
Small storage Large storage
Extrabiotic Intrabiotic nutrient distribution scheme
Small amounts of detritus Large amounts of detritus
Rapid nutrient exchange Slow nutrient exchange
Short residence times Long residence times
Minor chemical heterogeneity High chemical heterogeneity
Loose network articulation High network articulation
Low diversity of flows High diversity of flows
Undeveloped symbiosis Developed symbiosis
Properties of the community
Low diversity High diversity
Poor feedback control Developed feedback control
Poor spatial patterns Developed spatial patterns
Thermodynamic and integrative system properties
Poor hierarchical structure Developed hierarchical structure
Close to equilibrium Far from equilibrium
Low exergy storage High exergy storage
Small total entropy production High total entropy production
High specific entropy production Small specific entropy production
Small level of information High level of information
Small internal redundancy High internal redundancy
Small path lengths High path lengths
Low ascendency High ascendency
Poor indirect effects Developed indirect effects
Small respiration and evapo-transpiration High respiration and evapo-transpiration
Small energy demand for maintenance High energy demand for maintenance

The features, that are optimized throughout natural successions, provide several characteristics of emergent properties: They are only observable at the
ecosystem level (which is the typical and the lowest logical level to describe, e.g., cycling phenomena), and they are based on self organized processes.
They can not be explained on the basis of knowledge of the parts alone, and the emergence creating processual linkages between the subsystems are
nonlinear processes. From the hierarchy based viewpoint also the additive features (e.g., size, biomass, life spans) can be categorized as emergent
properties because their extensions are dependent on the scale of observation and because they also are based on internal system interrelations.

4. Landscape level: lateral nutrient transfers food webs The existence of emergent properties is based on the
including large scale predators. systems organization (built up by structures and func
tions) whereby the interrelations (energy, matter, water
On the other hand, the ability of biological systems to
and information flows, communications) play an impor
arrange themselves in a special manner, for example, in
tant role. Some conditions of the systems state add up to
a hierarchical way, is in itself a property which emerges
the increased chances that emergent properties will
as a consequence of the properties of its constituents,
appear. For example, instabilities seem to be important
but the organization and the function for sure cannot
conditions that support emerging processes, especially
always be foreseen. Thus, the capability of self organi referring to evolutionary emergence. Stable periods may
zation can be seen as an emergent property itself lead to the emergence of new structures through bifurca
(Figure 1). tions. As systems move toward the state of minimum
96 Emergent Properties

Inter-level loose use of the concepts over a vast range of areas at

relation first glimpse simply shows confusion. However, it is pos
sible to establish some typology of the areas where, and
the ways in which, the concepts have been used, following
lntra-level Figure 2.
relation First, emergent properties might appear through evo
lution of the systems, primary emergence, hereafter only
being repeated. This characteristic may be called evolu
tionary emergence. As structures are integrated, new
organizational forms, as previously mentioned often hier
archical, occur (hierarchical emergence).
Taking the view that emergent properties do exist and
that the reductionistic approach to science will not (dis)
solve the problem so it eventually disappears may allow
Figure 1 Biological entities are often organized in a hierarchical us to establish a schematic relationship between the vari
manner, whereby the emergent properties of a certain level are
based on the interrelations between the lower levels, while both
ous categories of ecosystem properties.
are constrained from the highest level linkages. One major line follows a direction of research problems,
the search for the unexplained and not understood. This
lies close to using emergent properties as research strate
dissipation they are, at the same time, moving toward
gies, while the extreme leads to the reductionist approach.
bifurcation points with possibilities of further evolution
This is more or less the situation at the second line, where
to occur. Similarly broken symmetries, complementarity
properties are collective and additive, that is, that the
has been proposed as a global mechanism.
properties are the sum of the whole, and may be explained
at subsystem level, provided sufficient knowledge exists. At
the other end, the attitude that only holistic studies will
Classification of Emergent Properties lead to increased understanding might be taken.
Along the third line, we find the core of emergence,
From the presentation of the concepts above it can be and following the above points the respective features
seen that emergence and emergent properties will not may be divided in an evolutionary line and in a hierarchal
easily find a clear, consistent and unifying definition for line. Here emergence is basically represented as a func
covering all the cases described. The widespread and tion of time and space. The evolutionary process was

emergent prop.
Ecosystem Hierarchical
properties properties Thermodynamic
emergent prop.

Research problems Formalistic emergent prop.

Emergent prop.
relative to model
Ecosystem Emergent
properties properties

Additive relations Dynamic first time

emergent prop.
Collective Evolutionary
properties properties Secondary
emergent prop.

Figure 2 An attempt to form a typology of emergent properties.

Emergent Properties 97

described above and deals with primary and secondary Rather in ecology, we do possess some knowledge
emergence. The organizational, hierarchical line includes about the system and what we usually refer to are the
four areas described in the previous sections: global deviations in what we observe in the systems or models of
emergent properties as a function of local rules and systems compared with our expectations built on previous
local interactions, thermodynamic emergent properties knowledge. The way of quantifying emergence has to be
dealing with emergence as a consequence of mainly the built on the use of computers and models. If our knowl
second thermodynamic law, the emergence of (dissipa edge gained hitherto is synthesized and treated in a
tive) structures as a result of thermodynamic gradients. computer model (from traditional ecological science) is
Computational emergence is also based on global pat p, and the outcome of an experiment or observations of a
terns emerging form local rules. As mentioned above, system differs by p the emergent properties can be
emergent properties also appear as a result of models calculated by the following:
being used to analyze the problem, which is called emer X p
gent property relative to a model. Emergence p ln
which correlates emergence to the concept of exergy.
Emergence now is a consequence of information gained
Quantifying Emergence between observations.
The question is if emergence in this manner will, at the
Several authors argue that in any attempts to formalize or end, dissolve itself and disappear as knowledge increases,
quantify the concepts, true emergent properties should which refers to the above debate of reductionism versus
be observer independent. This does not necessarily holism.
mean that emergent properties should be observation Many of the concepts used to characterize ecosystems
independent. Observations undertaken by different are based on various numerical treatments of data
methods result in differences in acquired knowledge. observed in the ecosystem. Since the concepts are
This means that emergent properties can be defined as immediately deducible (calculable) from certain knowl
the differences in knowledge gained by the observation of edge about the components of the ecosystem, for
a system by two different methods. This is partly reflected example, numbers, species, biomass etc. such concepts
by the computational emergence. cannot be coined as emergent property but rather as a
It is this observer dependency that leaves a way open collective property of the system. An interesting corre
for the quantification of emergent properties. Emergent sponding analog in this context are the macroscopic
properties could then be expressed in a semiquantitative properties from thermodynamics such as entropy and
way by the use of an index derived of Kullbacks measure parallels in formulation of formulas. Reductionism cannot
of information (Figure 3). This involves moving the win the debate since it will be impossible to achieve
normal reference frame in information theory assuming enough knowledge. If not for anything else, then
the a priori knowledge of the system to be zero, which is for thermodynamic reasons, since the achievement of
not necessarily the case. more and more detailed knowledge becomes more and
more expensive in terms of not only energy but also
Subject/observer Object/system dissipation.
Meanwhile, what strikes is that such a traditional,
gain I *** vertical organization of systems is not mandatory in
order to produce emergent behavior. Vertical, here, refers
to levels being either higher or lower in the hierarchy.
I ** Rather only parts are needed, of which none have actual
Emergence regulatory functions and therefore should be evaluated or
ranked higher than the other(s). Emergent properties can
I* Information/level occur also in horizontally organized systems, emergence
levels appearing alone as a consequence of interactions at the
Xi1 same level. The study of these intra level relationships
I0 and their consequences to the higher levels in the hier
Figure 3 Quantification of emergence, based on Kullbacks archy may be important to investigate in the future.
measure of information, might be carried out from quantifying the
difference between actual observed, a posteriori, behavior or
composition of a system and what may be predicted from a priori
knowledge about subsystems. The analysis may be carried out at See also: Autocatalysis; Ecological Network Analysis,
various levels of hierarchy, differing in emergence value. Environ Analysis; Exergy; Hierarchy Theory in Ecology.
98 Self-Organization

Further Reading Afdeling for Litteraturvidenskab. Copenhagen: University of

Bhalla US and Iyengar R (1999) Emergent properties of networks of Morgan CL (1923) Emergent Evolution. Williams and Norgate.
biological signalling pathways. Science 283: 381 387. Nielsen SN and Muller F (2000) Emergent properties of ecosystems.
Breckling B, Muller F, Reuter H, Holker F, and Franzle O (2005) In: Joergensen SE and Muller F (eds.) Handbook of Ecosystem
Emergent properties in individual based ecological models Theories and Management, pp. 195 216. Boca Raton, FL: Lewis
introducing case studies in an ecosystem research context. Publishers.
Ecological Modelling 186: 376 388. Salt GW (1979) A comment on the use of the term emergent properties.
Cariani P (1992) Emergence and artificial life. In: Langton G, Taylor C, American Naturalist 113(1): 145 148.
Farmer JD, and Rasmussen S (eds.) Artificial life II, pp. 775 797. Wicken JS (1986) Evolution and emergence. A structuralist perspective.
Redwood City: Addison Wesley. Rivista di Biologia/Biology Forum 79(1): 51 73.
Conrad M and Rizki MM (1989) The artificial worlds approach to Wieglieb G and Broring U (1996) The position of epistemological
emergent evolution. BioSystems 23: 247 260. emergentism in ecology. Senckenbergiana maritima
Emmeche C, Kppe S, and Stjernfelt F (1993) Emergence and the 27(3/6): 179 193.
Ontology of Levels. In Search of the Unexplainable. Arbejdspapir.

D G Green, S Sadedin, and T G Leishman, Monash University, Clayton, VIC, Australia
2008 Elsevier B.V. All rights reserved.

Introduction Self-Organization in an Ecological Setting

Historical Comments Practical Considerations
Theories of Self-Organization Further Reading


Self organization is the appearance of order and pattern

in a system by internal processes, rather than through
external constraints or forces. Plant distributions provide
examples of both constraints and self organization. On a
mountainside, for instance, cold acts as an external con
straint on the ecosystem by limiting the altitude at which
a plant species can grow. Simultaneously, competition for
growing sites and resources leads to self organization Figure 1 Effect of competition on plant distributions on a
within the community by truncating the range of altitudes gradient. The two plant species shown are adapted to different
conditions, which are here found at either end of the slope. At left,
where plant species do grow. Self organization can also be
there is no competition, so the distributions merge into one
seen among individuals within a population (e.g., within another. At right, competition truncates the distributions, leading
an ant colony or a flock of birds) and within individuals to sharply defined altitudinal zones.
(e.g., among cells during development) (Figure 1).
A growing understanding of ways in which internal a system. For instance, although biomass production in a
processes contribute to ecological organization has pro forest is a global property, it is simply the sum total of
vided new perspectives on many phenomena familiar production by all the organisms within the forest. A
from traditional ecology. Self organization usually stampede, on the other hand, is behavior that emerges
involves interactions between components of a system, when panic spreads from one animal to another within a
and is often closely identified with complexity. Also asso herd.
ciated with self organization is the idea of emergence: Semantic and philosophical issues sometimes lead to
that is, features of the system emerge out of interactions, confusion about self organization. Self organizing sys
as captured by the popular saying, the whole is greater tems are usually open systems, that is, they share
than the sum of its parts. It is necessary to distinguish information, energy, or materials with their surroundings.
between emergent features and other global properties of However this does not necessarily mean that the external
Self-Organization 99

environment controls or determines the way they orga role of adaptation in self organization. He suggested that
nize. A growing plant, for instance, absorbs water, light, seven basic elements are involved in the emergence of
and nutrient from its environment, but its shape and form order in complex adaptive systems. These include four
are determined largely by its genes. properties aggregation, nonlinearity, flows, and diver
Also, in considering self organization, it is important to sity and three mechanisms tagging, internal models,
clearly identify the system concerned, and in particular, and building blocks. In contrast, Stuart Kauffmans work
what is external and what is internal? This issue arises in on autocatalytic sets within Boolean networks emphasizes
the difference between a community and an ecosystem. ways in which self organization may structure biological
For a community, which consists of the biota of an area, systems independent of selection. Likewise, embryologist
the effect of (say) soil is an external constraint. However, Brian Goodwin suggested that to understand macroevo
for the corresponding ecosystem, which would include lution, we require a theory of morphogenesis which takes
soils, the interactions between plants, microorganisms, account of physical, spatial, and temporal dynamics in
and soil formation are internal processes. Defining the addition to selection. The work of James Kay provided
physical limits of an ecosystem poses similar problems. an interpretation of life from a thermodynamic perspec
A lake, for instance, is not a closed ecosystem. Among tive, arguing that self organizing systems maximize the
other things, water birds come and go, removing some dissipation of gradients in nature. In particular, Kay
organisms and introducing others. argues that over time, ecosystems evolve to dissipate
energy more efficiently by becoming increasingly com
plex and diverse.
Historical Comments

Self organization as a widespread phenomenon first came Theories of Self-Organization

to the attention of researchers during the mid twentieth
Thermodynamic Basis
century. The interest in self organization comes from
many different fields of study. The biologist Ludwig von In physical terms, the phenomenon of self organization
Bertalanffy drew attention to the role of internal interac appears at first sight to be ruled out by the second law of
tions and processes in creating organization within thermodynamics, which states that in any closed system,
biological systems. His general systems theory drew entropy increases with time. In this sense, living systems
heavily on analogies to highlight the existence of common seem to fly in the face of thermodynamics by accumulat
processes in superficially different systems. Meanwhile, ing order. However, self organizing systems need not be
W. Ross Ashby and Norbert Wiener explored self closed. Open systems, including living things, share
organization from the perspective of communications energy and information with the outside environment.
and feedback in the control of systems. Ashby introduced In the late 1960s, Ilya Prigogine introduced the idea of
the term self organizing in 1947. Wiener coined the dissipative systems to explain how this happens. He
term cybernetics to refer to the interplay of control defined dissipative systems to be open systems that are
systems and information. In the 1950s, systems ecologist far from equilibrium. Dissipative systems have no ten
H. T. Odum collaborated with engineer Richard dency to smooth out irregularities and to become
Pinkerton to develop the principle of maximum power, homogeneous. Instead, they allow irregularities to grow
which states that systems self organize to maximize and spread. Physical examples include crystal formation.
energy transformation. Biological systems, including cells, organisms, and eco
During the 1970s and 1980s, increasing computing systems, are all examples.
power made it possible to use simulation to explore the
consequences of complex networks of interactions. By the
The Network Model
last two decades of the twentieth century, the nature and
implications of biological self organization were increas An important source of self organization is provided by
ingly being explored as a part of the complexity theory. the interactions and relationships between the objects that
The new field of Artificial life (Alife), initiated by pio comprise a complex system. Patterns of such relationships
neers such as Chris Langton, Pauline Hogeweg, and are captured by the network model of complexity.
Bruce Hesper, has produced a series of seminal models Networks capture the essence of interactions and rela
that demonstrate self organization in a variety of ecolo tionships, which is a fundamental source of complexity. A
gical and evolutionary contexts. Around the same time, graph is defined to be a set of nodes (objects) joined by
H. T. Odum introduced the systems concept of emergy edges (relationships) and a network is a graph in which
to represent the total energy used in developing a process. the nodes and/or edges have values associated with them.
By the 1990s, researchers were looking for broad based In a food web, for instance, the populations form nodes,
theories of self organization. John Holland stressed the and the interactions between them (e.g., predation) form
100 Self-Organization

the edges. In a landscape, spatial processes and relation instance, the descendents of a particular individual
ships create many networks. For instance, the nodes might animal (the root of the tree) form a hierarchy deter
be individual plants and the corresponding edges would mined by birth. Trees and hierarchies are closely
be any processes that create relationships between them, associated with the idea of encapsulation.
such as dispersal or overshading. In an animal social
group, the nodes would be individuals and the edges
A scale free network is a connected network in which
the degrees of the nodes follow an inverse power law.
would be relationships such as kinship or dominance. That is, some nodes are highly connected, but most
Nodes that are joined by an edge are called neighbors. have few (usually just one) connections.
The degree of a node is the number of immediate neigh
bors that it has. A path is a sequence of edges in which the Encapsulation
end node of one edge is the start node of the next edge, for
example, the sequence of edges AB, BC, CD, DE Encapsulation is the process by which a set of distinct
forms a path from node A to node E. A cycle is a path that objects combine to act as a single unit. Individual fish, for
ends where it starts, for example, AB, BC, CA. A example, form a school by aligning their movements with
network is called connected if, for any pair of nodes, their neighbors. Because smaller objects usually merge
there is always some path joining them (otherwise it is into larger wholes, encapsulation is often linked to ques
disconnected). The diameter of a network is the maxi tions of scale. Encapsulation is closely associated with the
mum separation between any pair of nodes. Clusters are idea of emergence. The whole emerges when individuals
highly connected sets of nodes. become subsumed within a group in relation to the out
The importance of networks stems from their univer side world. There are many examples in ecology.
sal nature. Network structure is present wherever a Ecosystems are communities of interacting organisms;
system can be seen to be composed of objects (nodes) populations are groups of interbreeding organisms; and
and relationships (edges). Less obvious is that networks schools, flocks, and herds are groups of animals moving in
are also implicit in the behavior of systems. In this respect, coordinated fashion. In all of these cases, the individuals
the nodes are states of the system (e.g., species composi may not be permanently bound to the group, unlike cells
tion) and the edges are transitions from one state to within the human body. Cellular slime molds present an
another. intermediate case in which cells sometimes act indepen
Sometimes, network structure plays a more important dently but at other times aggregate to form a multicellular
part in determining the behavior of a system than the individual.
nature of the individual components. In dynamic systems, Various ecological theories are based on the assump
for instance, cycles are associated with feedback loops. In tion that encapsulation plays an important role in
disconnected networks, the nodes form small, isolated ecosystem structure and function. The concept of ecosys
components, whereas in connected networks, they are tem compartments implies that a community is formed of
influenced by interactions with their neighbors. Self distinct groups (compartments) consisting of mutually
interacting species, but the interactions between the
organization in a network can occur in two ways: by the
groups are limited.
addition or removal of nodes or edges, or by changes in
the values associated with the nodes and edges.
Several kinds of network patterns are common and Connectivity and Criticality
convey important properties.
Criticality is a phenomenon in which a system exhibits
A random network is a network in which the nodes are
connected at random. In a random network of n nodes,
sudden phase changes. Examples include water freezing,
crystallization, and epidemic processes. Associated with
the degrees of the nodes approximate a Poisson dis every critical phenomenon is an order parameter, and the
tribution, and the average length (L) of a path between phase change occurs when the order parameter reaches a
any two nodes is given by L log(n)/log(d), where d is critical value. For example, water freezes, when its tem
the average degree. perature falls to 0
C. A wildfire spreads when fuel
A regular network is a network with a consistent pat
tern of connections, such as a lattice or cycle.
moisture falls below a critical level (else it dies out).
Changes in the connectivity of a network have impor
Small worlds fall between random networks and reg
ular networks. They are typically highly clustered, but
tant consequences and often underlie critical phenomena.
When a network is formed by adding edges at random to a
with low diameter. A common scenario is a system set of N nodes, a connectivity avalanche occurs when the
dominated by short range connections, but in which number of edges is approximately N/2. This avalanche is
some long range connections are also present. characterized by the formation of a connected subnet,
A tree is a connected network that contains no cycles. A
hierarchy is a tree that has a defined root node. For
called a unique giant component (UGC), which contains
most of the nodes in the full network. The formation of
Self-Organization 101

the UGC marks a phase change in which the network connectivity of the system. In this way, mutation, migra
shifts rapidly from being disconnected to connected. tion, and extinction could keep the system near the
Any system that can be identified with nodes and critical region, as the addition of new variation drives
edges forms a network, so the connectivity avalanche the ecosystem out of subcriticality, while extinction ava
occurs in many settings and is the usual mechanism lanches prevent supercriticality. Proponents of this idea
underlying critical phase changes. point to extinction events, whose distribution follows an
The connectivity avalanche has several important inverse power law, as supporting evidence. However,
implications. For interacting systems, it means that the other explanations of this pattern, such as cometary
group behaves either as disconnected individuals, or as a impacts, are also plausible.
connected whole. Either global properties emerge, or they
do not: there is usually very little intermediate behavior.
Landscape connectivity provides an important ecological
example of critical phase change. Feedback is a process in which outputs from a system
Phase changes in connectivity also underlie criticality in affect the inputs. Predatorprey systems are examples of
system behavior. The degree of connectivity between states negative feedback. For instance, any increase in the size of
of a system determines the richness of its behavior. Studies a predator population means that more prey are eaten, so
based on automata theory show that if connectivity is too low, the prey population decreases, which in turn leads to a
systems become static or locked in narrow cycles. If connec decrease in the predator population. Reproduction is an
tivity is too high, systems behave chaotically. The transition example of positive feedback: births increase population
between these two phases is a critical region, popularly size, which in turn increases the rate of reproduction,
known as the edge of chaos. It has been observed that auto which leads to yet more births. Feedback loops arise
mata whose state spaces lie in this critical region exhibit the when a sequence of interactions form a closed loop, for
most interesting behavior. This observation led researchers example, ABCA. Feedback loops play an important
such as James Crutchfield, Christopher Langton, and Stuart role in food webs and ecosystem stability. Time delays in
Kauffman to suggest that automata need to reside in the the response within a feedback loop often lead to cyclic
critical region to perform universal computation. More spec behavior (e.g., in predatorprey systems).
ulative is their suggestion that the edge of chaos is an essential Both positive and negative feedback are important in
requirement for evolvability (the ability to evolve) in complex self organization. By dampening changes, negative feed
systems, including living things. Others have proposed that back acts as a stabilizing force. It is one of the principal
living systems exploit chaos as a source of novelty, and that mechanisms of homeostasis, the maintenance of dynamic
they evolve to lie near the edge of chaos. These ideas are equilibrium by internal regulation. In contrast, positive
closely related to self organized criticality (SOC). feedback magnifies minor deviations. An example is com
petitive exclusion: any small decrease in size of a
competing population is likely to lead to further
Self-Organized Criticality
decreases, until it dies out (Figure 2).
SOC is a phenomenon wherein a system maintains itself in
a critical or near critical state. A classic example is the
pattern of collapses in a growing pile of sand. Because
information theory suggests that systems in critical states Stigmergy is a form of self organization that occurs when
are most amenable to information processing and complex parts of a system communicate by modifying their environ
ity, self organized criticality has been proposed as a ment. Many examples of stigmergy occur in the organization
component of collective behavior in ant colonies, societies, of eusocial insect colonies. For example, in ant colonies,
ecosystems, and large scale evolution. SOC is character objects such as food, larvae, and corpses are often stored in
ized by events whose size and frequency distributions discrete larders, nurseries, and cemeteries. Models show that
follow an inverse power law. However, it is often difficult this civic order can emerge through interactions between the
to distinguish genuine cases of SOC from simple cause and ants and their environment. In the model, ants pick up objects
effect processes that exhibit similar distributions. at random, and may drop them when they encounter similar
For example, ecosystems might tend toward critical objects. Over time, this process creates piles of similar objects.
states through the following mechanism. If new species or Positive feedback causes larger piles to grow at the expense of
mutations appear in an ecosystem occasionally, then as smaller ones (Figure 3).
the variation in the ecosystem increases over time, so does
the probability of forming destabilizing positive feedback
loops. Such destabilizing interactions could initiate ava
lanches of extinctions, and the probable size of such Synchrony can alter system level behavior by enhancing or
avalanches would be related to the preexisting dampening nonlinearities. For example, when predator and
102 Self-Organization

(a) (b) change in temperature or day length. Synchronized breed

ing conveys distinct advantages such as maximal
exploitation of resources and satiation of predators.
Different species often have co adapted simultaneous
seasonal behavior, such as birds that breed when butter
+ + flies emerge. However, both the environmental cues,
+ + and the physiological response, may differ among these
co adapted species. For example, great tits time their egg
+ laying by photoperiod. Winter moths are an important
+ + + food source during the breeding season, and they develop
+ + +
more quickly at higher temperatures. As a result, recent
+ +
warm springs in Europe caused by climate change have
disrupted the synchronization between these species,
Figure 2 The role of feedback in self-organization of a food
web. In this diagram, circles represent populations and arrows
reducing food availability for nesting great tits and poten
indicate the influence (positive or negative) of one population on tially destabilizing populations.
another. In a food web, circular chains of interaction between In other cases, synchronous behavior arises through
populations form feedback loops, as in the example shown here. social contagion, where individuals imitate others. The
(a) The initial food web contains both positive and negative dynamics of such behavior are similar to those seen in
feedback loops. Internal dynamics within the positive feedback
loops leads to the local extinction of several populations. (b) The
epidemiology. Social contagion can lead to coordinated
resulting food web contains only negative feedback loops, which group behavior such as flocking, as well as disparate phe
stabilize the community. nomena such as synchronized flashing in fireflies, and
fashions in mate choice among birds and fish. The emer
gence of synchronous behavior in these cases is highly
(a) (b) sensitive to the structure of social networks. Synchrony is
easily achieved when networks are highly connected (i.e.,
individuals can perceive a large number of other indivi
duals, or some individuals have very large influence).
However, in loosely connected networks, social contagion
can result in asynchronous waves or chaos.

Complex Adaptive Systems

Complex adaptive systems (CASs) consist of diverse,
Figure 3 The emergence of order by stigmergy and
locally interacting components that are subject to selection.
feedback in an ant colony. Given a random scatter of objects (a), Examples include learning brains, developing individuals,
ants sort objects by picking them up and dropping them again economies, ecosystems, and the biosphere. In such systems,
when they find a similar object. This process creates piles, which hierarchical organization, continual novelty and adapta
grow over time. Large piles grow at the expense of smaller ones tion, and nonequilibrium dynamics are known to emerge.
until only a few large piles remain (b).
As a result, the behavior of a CAS is characterized by
nonlinearity, historical contingency, thresholds, and multi
ple basins of attraction. A key question in current CAS
prey populations are tightly coupled to one another, a stable, research has been the relationship between resilience and
negative feedback relationship can result where an increase criticality. Some authors suggest that a CAS will generally
in prey causes increased predators and a subsequent evolve toward self organized criticality. By being main
decrease in prey. In this case, the ecological interaction tained near the edge of chaos, such systems might
acts like a thermostat regulating population size. However, maximize information processing. In this way, criticality
if the two populations respond at different rates, oscillations might enhance the ability of CASs to adapt to changing
or even chaotic behavior can occur instead. A classic exam environments and efficiently utilize resources, making sys
ple of such oscillations occurs in the interaction between tems become more resilient over time.
populations of hares and lynxes in the Arctic Circle.
Synchronized breeding behavior is common and includes
mass flowering in plants, mass breeding in birds, and mass Artificial Life
spawning among marine animals such as corals and squid. In The field of Alife uses simulation models to understand
these cases, synchrony is usually achieved by individuals biological organization by abstracting crucial features and
responding to a common environmental cue, such as a examining living systems as they could be.
Self-Organization 103

One of the most widespread representations used in maximum area that a sitting bird can defend without
Alife models is the cellular automaton (CA). This is a grid abandoning her nest. More complex coordinated group
of cells in which each cell has a state (some property of behaviors can emerge when individuals take on different
interest) and is programmed to behave in identical fash tasks and roles within groups. For example, within ant and
ion. Each cell has a neighborhood (usually the cells termite colonies, individuals can develop into a variety of
immediately adjacent to it) and the states of it neighbors castes, each with distinct roles such as foraging, nest
affect changes in a cells state. The most famous example defense, and nursing young. In honeybees, individuals
is the Game of Life, in which each cell is either alive or take on different roles at different life stages.
dead at any time. Despite its extreme simplicity, the In some cases, upper limits exist on the size that social
game showed that large numbers of interactions governed groups can attain and depend on interactions between the
by simple rules lead to the emergence of order within a animals. In apes, for instance, where social bonds are main
system. Cellular automata have been used to model many tained by grooming, troop sizes tend to be 3060 individuals.
biological and ecological systems. In models of fires, epi Larger troops tend to fragment. Among humans, social
demics, and other spatial processes, each cell represents a groups are usually much larger. The anthropologist Robin
fixed area of the landscape and the cell states represent Dunbar argues that this is a consequence of speech providing
features of interest (e.g., susceptible, infected, or immune more efficient social bonding than grooming, leading to a
organisms in an epidemic model). natural group size of 100150 individuals.
Other prominent ALife models include Tom Rays In most cases, group size may be the outcome of
Tierra model, which demonstrated adaptation within several interacting ecological and social factors. For
self reproducing automata. Craig Reynolds boids model example, although lions hunt cooperatively, prides and
demonstrated that flocking behavior emerges from simple hunting groups are usually larger than is optimal for
interactions between individuals. James Lovelocks hunting efficiency. Lionesses cooperate to defend cubs
Daisyworld model showed the potential for biotic feed against infanticidal males by forming cre`ches. In addition,
back and adaptation to stabilize the biosphere. hunters are vulnerable to attack by larger groups,
and territories are more effectively defended by larger
Self-Organization in an Ecological Setting The origin of cooperation among groups of cells and
Social Groups organisms can also be examined from the perspective of
self organization. The paradox of the evolution of coop
Relationships between individuals create several kinds of
eration is that (by definition) selfish individuals
organizations within groups of animals.
outcompete altruists, and therefore in a population of
Coordination between moving animals leads the for
self replicators, a selfish mutant should always spread at
mation of groups. Examples include swarms of insects,
the expense of altruists. Nonetheless, altruism does occur
flocks of birds, schools of fish, and herds of mammals.
among humans and cooperative behavior is often seen
Coordinated group movements, even in very large
among animals. Such cooperative behavior can self orga
groups, can be achieved by individuals obeying simple
rules, such as keep close, but not too close, to your nize when the network structure that governs interactions
neighbors and head in the same general direction as among individuals results in the same individuals encoun
your neighbors. tering one another repeatedly (e.g., when individuals are
Several mechanisms that channel aggressive behavior fixed in space, so that their only interactions are with their
create social organization. In social animals, dominance neighbors), or when their reproductive fate is very closely
hierarchies reduce the potential costs of conflict over tied to that of others (as is the case for cells within a
mates and food. Adominance hierarchy emerges when multicellular organism). Experimentally, the evolution
interactions between individuals result in physiological of cooperation has been induced in bacterial populations
and behavioral changes: for example, winning a contest by production of adhesive, causing individual cells to
may elevate testosterone, causing increased dominance clump together. Cooperation can also evolve in marginal
behavior, and evoking submissive behavior from indivi environments, where the evolutionary impact of compe
duals who have been less successful in the past. In this tition between individuals is outweighed by the need to
way, coherent transitive hierarchies can emerge even survive. Experimental studies of bacteria in marginal
when all individuals were initially equal. Similarly, terri environments show that complex spatial patterns and
toriality reduces the costs of conflict over resources by signaling behaviors can emerge as a result of this selec
partitioning a landscape among a population. tion. In theoretical models, the inclusion of policing
Territoriality often generates spatial patterns, such as behavior (punishing nonconformists) can also enforce
regular distances between nests in seabird colonies. In high levels of cooperation even when interactions occur
this case, the distance between nests is defined by the at random in large societies.
104 Self-Organization

Persistence and Stability in Ecosystems and photosynthesis. The outputs of material from one
organism often become inputs to other. This focus on
One of the most puzzling topics in systems ecology is how
what eats what led Elton to identify several patterns,
ecosystems emerge that are at once complex and stable.
notably the food chain and the food web, the food cycle,
Field studies suggest that the most complex (diverse)
ecosystems are also the most stable. However, this obser the ecological niche, and the pyramid of numbers.
vation runs counter to expectation from systems theory. It Self organization in ecosystems is evident in the struc
shows that the more components a dynamic system has, ture in food webs, networks that describe trophic
the more likely it is that a destabilizing interaction (such interactions among species. Within food webs, specific
as a positive feedback loop) will cause it to collapse and patterns of interaction may be prevalent. These patterns,
lose species. Consequently, systems theory suggests that termed ecological motifs, are thought to represent espe
simpler ecosystems should be more stable than complex cially stable interactions. The concept of keystone species
ones. The paradox implies that the complex, stable eco supposes that certain species play a crucial role in main
systems seen in nature are not random assemblages. taining the integrity and stability of an ecosystem.
Self organization in this case involves removal of desta Analysis of food webs suggests that a small world
bilizing positive feedback loops. structure is common. That is, most species interact with
only a small number of other species, but the connectivity
of the web as a whole is maintained by a few species that
Communities versus Assemblages interact with a large number of others. This observation
provides a theoretical basis for the idea of keystone spe
The question of how important self organization is in
cies. Functionally, small world networks are thought to be
ecosystems has long been debated in ecology. Are ecosys
robust to random loss of nodes (e.g., species), but vulner
tems communities of co adapted species, or are they
simply random assemblages? Some early theorists, such able to attacks that target their highly connected nodes
as Clements, believed that the groups of species found (e.g., keystone species).
together were specialized for living together, whereas
others, such as Gleason, stressed the importance of chance
and individuals. Spatial Patterns and Processes
The idea of succession concerns the patterns and pro
cesses involved in community change, especially after Spatial processes lead to the formation of distribution
disturbance. A form of self organization often associated patterns. Seed dispersal, for instance, often produces con
with succession is facilitation. That is, plants and animals centrations of seedlings around parent plants and leads to
present in an area can alter the local environment, thereby the formation of clumped distributions. When local dis
facilitating the appearance of populations that replace persal is combined with patchy disturbance, such as fire,
them. After a fire, for example, a forest will regenerate the result is a distribution composed of patches. When
with herbs and shrubs growing back almost immediately. combined with environmental gradients, such as soil
The first trees to reappear will be pioneer (disturbance) moisture, local dispersal can produce zone patterns, with
species, which disperse well, grow fast, and can tolerate different species dominating different areas (Figure 4).
open, exposed conditions. These trees create shade and
leaf litter, which favor slow growing, shade tolerant trees. (a) (b) (c)
Recent theoretical work (such as Hubbells neutral
theory of biodiversity and biogeography) emphasizes the
role of chance and spatial dynamics in generating ecolo
gical patterns. In these models, self organization is trivial
because all individuals and species are effectively identi
cal, and species abundances are driven by random birth,
migration, and death processes. Both neutral and self
organizing models have been successful in explaining
real relative abundance and speciesarea curves.
Figure 4 Emergence of spatial patterns from dispersal. This
CA model shows the hypothetical distributions of two plant
Food Webs populations that result in three different scenarios. (a) Global
dispersal, in which seeds can spread anywhere, results in
Species interactions lead to the flow of material within an random distributions of plants. (b) Dispersal from local seed
ecosystem. For animals the most common processes are sources leads to clumped distributions. (c) The combination of
eating, respiration, excretion, and egestion. For plants, local dispersal and environmental gradients (from top to bottom)
they are root uptake of water and nutrients, respiration, creates vegetation zones.
Self-Organization 105

Fragmentation is one of the most important conse and if this happened on a global scale, then the biosphere
quences of landscape connectivity. When the density of itself might behave as a self regulating system. However,
(randomly located) objects in a landscape falls below a evidence for Gaian processes in real ecosystems remains
critical density, they are mostly isolated individuals. tenuous and their theoretical plausibility is disputed.
When the density exceeds the critical threshold, they
become connected. The density at which the critical
threshold occurs depends on the size of the neighborhood
of the objects. There are many cases where landscape Self organization may play a prominent role in evolution,
connectivity plays an important role. Epidemic processes especially in the context of landscapes, which regulate
require a critical density of resources to spread. Instances interactions between individuals. One consequence is the
include disease outbreaks (susceptible individuals), fire evolution of cooperation in marginal and viscous habitat
spread (fuel), and invasions of exotic plants (suitable networks, whereas randomly interacting populations are
sites). Populations become fragmented if individuals can more dominated by intraspecific competition and there
not interact with one another. For instance, in wet years the fore more likely to behave selfishly.
water bodies of central Australia are essentially connected Landscape structure influences genetic diversity and
for water birds, which can fly from one body to another speciation. In connected landscapes, genes flow freely
almost anywhere in the continent. In dry years, however, throughout a species and speciation is inhibited.
many water bodies shrink or dry up and become too widely However, in fragmented landscapes, a species breaks
separated for birds to migrate between them (Figure 5). into isolated subpopulations. Fragmentation increases the
risk of inbreeding and loss of genetic diversity in these
subpopulations. Divergence between population fragments
Self-Organization in the Biosphere may also underlie adaptive radiations, in which many
Arguably the most ambitious ecological theory based on novel species suddenly emerge simultaneously.
self organization is the Gaia hypothesis, which postulates As species adapt to their environment, they are often
that the biosphere itself evolves to a homeostatic state. faced by tradeoffs in allocating resources for different
Lovelock suggested the Daisyworld model as an illustra purposes. These tradeoffs can lead to the evolution of
tion of how this process might occur. On the hypothetical distinct morphs within a species, or to speciation. For
Daisyworld, black and white daisies compete for space. example, many mangrove species face a conflict between
Although both kinds of daisies grow best at the same salt tolerance and competitive ability. Mangroves grow in
temperature, black daisies absorb more heat than white estuaries, where salinity varies along the gradient
daisies. When the Sun shines more brightly, heating the between land and sea. Mangroves growing landward will
planet, white daisies spread, and the planet cools again. be under strong selection for competitive ability, while
When the Sun dims, the black daisies spread, warming the those growing closer to the sea require better salt toler
planet. In this way, competitive interactions between daisies ance. The tradeoff, combined with local seed dispersal,
provide a homeostatic mechanism for the planet as a whole. can generate discrete banding patterns in the distribution
The idea behind Gaia is that ecosystems will survive of mangrove species, where each species is displaced by a
and spread more effectively if they promote the abiotic more salt tolerant one closer to the sea.
conditions required for their own persistence. If so, eco Contingency also plays a large part in the organization
systems might gradually evolve to be increasingly robust, of spatial distributions. Spatial dominance occurs when a
particular species is overwhelmingly abundant in a local
environment. In this situation, the species can resist inva
sion, even by a superior competitor, because its
propagules are much more numerous locally than those
of any other population. For the same reason, a mutation
that enables a species to exploit a novel environment may
result in it permanently excluding potential competitors
from that environment, even after they have evolved
Subcritical Critical Supercritical similar adaptations.
Figure 5 Critical phase changes in connectivity within a
fragmented landscape. In this CA model, grid cells represent
sites in a landscape. Gray and black cells represent vegetation
and white cells have no cover. The black cells show examples of
Practical Considerations
patches of vegetation sites that are connected, for example, by
spread of a fire ignited in the center of the grid. Notice that only a The insights provided by theories of self organization
small increase in the density of covered sites makes the have many practical implications, both for ecology and
difference between subcritical and supercritical. for conservation. The sharp end of the conservation debate
106 Ecological Complexity

often hinges on the question of which areas and which sites to studies of connectivity in landscapes, both field
to conserve. If ecosystems consist of random collections of based, and using data from remote sensing and
species, then one site in a landscape is as good as another. geographic information.
All that matters is to preserve representative populations
of each species. However, if the ecosystems consist of self
organized communities, in which the species are adapted
to depend on one another for survival, then whole com See also: Autocatalysis; Ecological Complexity; Emergent
munities need to be conserved. Properties; Hierarchy Theory in Ecology.
Closely related to the above issue is that the tendency
for randomly constructed food webs to be unstable raises
questions about the long term viability of artificially cre Further Reading
ated communities in which translocated species are
introduced into new areas. Self organization is evident Ball P (1999) The Self Made Tapestry: Pattern Formation in Nature.
Oxford: Oxford University Press.
even in artificial ecosystems. In biosphere 2, for instance, Camazine S, Deneubourg J L, Franks NR, et al. (2003) Self Organization
a closed, experimental environment designed to emulate in Biological Systems. Princeton: Princeton University Press.
natural ecosystems, the environment was found to favor Green DG, Klomp NI, Rimmington GR, and Sadedin S (2006)
Complexity in Landscape Ecology. Amsterdam: Springer.
species that collect more energy and internal processes Holland JH (1996) Hidden Order: How Adaptation Builds Complexity.
led to unexpected problems, such as runaway depletion of New York: Addison Wesley.
oxygen levels. Levin SA (1998) Ecosystems and the biosphere as complex adaptive
systems. Ecosystems 1(5): 431 436.
The need to understand self organization is impor Patten BC, Fath BD, and Choi JS (2002) Complex adaptive hierarchical
tant when considering altered ecosystems. For systems Background. In: Costanza R and Jrgensen SE (eds.)
instance, it is usually not possible to carry out experi Understanding and Solving Environmental Problems in the 21st
Century, pp. 41 94. London: Elsevier.
ments to determine the long term effects of current Prigogine I (1980) From Being to Becoming. New York: Freeman (ISBN
ecological management practices such as translocation 0 7167 1107 9).
of populations, controlled burning or allocation of Rohani P, Lewis TJ, Gruenbaum D, and Ruxton GD (1997) Spatial
self organization in ecology: Pretty patterns or robust reality? Trends
reserves and wilderness areas. This problem makes in Ecology and Evolution 12(8): 70 74.
simulation modeling a potentially crucial tool of eco Sole RV and Levin S (2002) Preface to special issue: The biosphere as a
logical theory and practice. New methods of field complex adaptive system. Philosophical Transactions of the Royal
Society of London B 357: 617 618.
observation are also appearing. For instance, the Watts DJ and Strogatz SH (1998) Collective dynamics of small world
need to understand landscape fragmentation has led networks. Nature 393(6684): 440 442.

Ecological Complexity
J L Casti, International Institute for Applied System Analysis, Laxenburg, Austria
B D Fath, Towson University, Towson, MD, USA and International Institute for Applied System Analysis,
Laxenburg, Austria
2008 Elsevier B.V. All rights reserved.

Complexity as a Systems Concept Would-Be Worlds

Surprise-Generating Mechanisms Conclusions
Emergent Phenomena Further Reading
Ecological Complexity

Complexity as a Systems Concept are difficult to understand. The behavior of national

economies, the human brain, and a rain forest ecosystem
In everyday parlance, the term complex is generally are all good illustrations for complex systems.
taken to mean a person or thing composed of many These examples show that there is nothing new about
interacting components whose behavior and/or structure complex systems. But what is new is that for perhaps the
Ecological Complexity 107

first time in history, we have the knowledge and the commonplace into a more formal, stylized language, one
tools to study such systems in a controlled, repeatable, in which intuition and meaning can be more or less
scientific fashion. So there is reason to believe that this faithfully captured in symbols and syntax. The problem
newfound capability will eventually lead to a viable the is that an integral part of transforming complexity (or
ory of such systems. anything else) into a science involves making that which
Prior to the recent arrival of cheap and powerful is fuzzy precise, not the other way around, an exercise we
computing capabilities, we were hampered in our ability might more compactly express as formalizing the
to study a complex system like a road traffic network, a informal.
national economy, or a supermarket chain because it was To bring home this point a bit more forcefully, let us
simply too expensive, impractical, too time consuming consider some of the properties associated with simple
or too dangerous to tinker with the system as a whole. systems by way of inching our way to a feeling for what is
Instead, we were limited to biting off bits and pieces of involved with the complex. Generally speaking, simple
such processes that could be looked at in a laboratory or in systems exhibit the following characteristics.
some other controlled setting. But with todays computers Predictable behavior. There are no surprises in simple
we can actually build complete silicon surrogates of these systems; simple systems give rise to behaviors that are
systems, and use these would be worlds as laboratories easy to deduce if we know the inputs (decisions) acting
within which to look at the workings and behaviors of upon the system and the environment. If we drop a stone,
the complex systems of everyday life. it falls; if we stretch a spring and let it go, it oscillates in a
In coming to terms with complexity as a systems con fixed pattern; if we put money into a fixed interest bank
cept, we first have to realize that complexity is an account, it grows to a predictable sum in accordance with
inherently subjective concept; what is complex depends an easily understood and computable rule. Such predict
upon how you look. When we speak of something being able and intuitively well understood behavior is one of the
complex, what we are really doing is making use of every principal characteristics of simple systems.
day language to express a feeling or impression that we Complex processes, on the other hand, generate coun
dignify with the label complex. But the meaning of terintuitive, seemingly acausal behavior that is full of
something depends not only on the language in which it surprises. Lower taxes and interest rates lead to higher
is expressed (i.e., the code), the medium of transmission, unemployment; low cost housing projects give rise to
and the message, but also on the context. In short, mean slums worse than those the better housing replaced; the
ing is bound up with the whole process of communication construction of new freeways results in unprecedented
and does not reside in just one or another aspect of it. As a traffic jams and increased commuting times. For many
result, the complexity of a political structure, an ecosys people, such unpredictable, seemingly capricious, behav
tem, or an immune system cannot be regarded as simply a ior is the defining feature of a complex system.
property of that system taken in isolation. Rather, what Few interactions and feedback/feedforward loops. Simple
ever complexity such systems have is a joint property of systems generally involve a small number of components,
the system and its interaction with another system, most with self interactions dominating the linkages among the
often an observer and/or controller. variables. For example, primitive barter economies, in
So just as with truth, beauty, good, and evil, complex which only a small number of goods (food, tools, weapons,
ity resides as much in the eye of the beholder as it does in clothing) are traded, seem much simpler and easier to
the structure and behavior of a system itself. This is not to understand than the developed economies of industria
say that there do not exist objective ways to characterize lized nations, in which the pathways between raw
some aspects of a systems complexity. After all, an amoeba material inputs and finished consumer goods follow
is just plain simpler than an elephant by whatever notion labyrinthine routes involving large numbers of interac
of complexity you happen to believe in. The main point, tions between various intermediate products, labor, and
though, is that these objective measures only arise as capital inputs.
special cases of the two way measures, cases in which In addition to having only a few variables, simple
the interaction between the system and the observer is systems generally consist of very few feedback/feedfor
much weaker in one direction than in the other. ward loops. Loops of this sort enable the system to
A second key point is that common usage of the term restructure, or at least modify, the interaction pattern
complex is informal. The word is typically employed as among its variables, thereby opening up the possibility
a name for something that seems counterintuitive, unpre for a wider range of behaviors. To illustrate, consider a
dictable, or just plain hard to understand. So if it is a large organization that is characterized by variables like
genuine science of complex systems we are after and employment stability, substitution of capital for human
not just anecdotal accounts based on vague personal opin labor, and level of individual action and responsibility
ions, we are going to have to translate some of these (individuality). Increased substitution of work by capital
informal notions about the complex and the decreases the individuality in the organization, which in
108 Ecological Complexity

turn may reduce employment stability. Such a feedback complexity as surprise generating mechanisms, whose
loop exacerbates any internal stresses initially present in quite different natures each lead to their own character
the system, leading possibly to a collapse of the entire istic type of surprise. Let us take a quick look at each of
organization. This type of collapsing loop is especially these mechanisms before turning to a more detailed con
dangerous for social structures, as it threatens their ability sideration of how they act to create complex behavior.
to absorb shocks, which seems to be a common feature of Paradox. Paradoxes arise from false assumptions about
complex social phenomena. a system leading to inconsistencies between its observed
Centralized decision making. In simple systems, power is behavior and our expectations of that behavior.
generally concentrated in one or at the most a few deci Sometimes these situations occur in simple logical or
sion makers. Political dictatorships, privately owned linguistic situations, such as the famous liar paradox
corporations, and the Roman Catholic Church are good (This sentence is false.). In other situations, the paradox
examples of this sort of system. These systems are simple comes from the peculiarities of the human visual system,
because there is very little interaction, if any, between the as with the impossible staircase shown in Figure 1, or
lines of command. Moreover, the effect of the central simply from the way in which the parts of a system are put
authoritys decision upon the system is usually rather together, like the developing economy discussed in the
easy to trace. preceding section.
By way of contrast, complex systems exhibit a diffu Instability. Everyday intuition has generally been
sion of real authority. Such systems seem to have a honed on systems whose behavior is stable with regard
nominal supreme decision maker, but in actuality the to small disturbances, for the obvious reason that unstable
power is spread over a decentralized structure. The systems tend not to survive long enough for us to develop
actions of a number of units then combine to generate good intuitions about them. Nevertheless, the systems of
the actual system behavior. Typical examples of these both nature and humans often display pathologically sen
kinds of systems include democratic governments, labor sitive behavior to small disturbances, as for example,
unions, and universities. Such systems tend to be some when stock markets crash in response to seemingly
what more resilient and stable than centralized structures minor economic news about interest rates, corporate mer
because they are more forgiving of mistakes by any one gers, or bank failures. Such behaviors occur often enough
decision maker and are more able to absorb unexpected that they deserve a starring role in our taxonomy of
environmental fluctuations. surprise.
Decomposable. Typically, a simple system involves weak Incomputability. The kinds of behaviors seen in models
interactions among its various components. So if we sever of complex systems are the end result of following a set of
some of these connections, the system behaves more or rules. This is because these models are embodied in
less as before. Relocating American Indians to reserva computer programs, which in turn are necessarily just a
tions produced no major effects on the dominant social set of rules telling the machine what bits in its memory
structure in New Mexico and Arizona, for example, since, array to turn on or off at any given stage of the calculation.
for various cultural reasons, the Indians were only weakly By definition, this means that any behavior seen in such
coupled to the dominant local social fabric in the first worlds is the outcome of following the rules encoded in
place. Thus, the simple social interaction pattern present the program. Although computing machines are de facto
could be further decomposed and studied as two inde rule following devices, there is no a priori reason to
pendent processes the Indians and the settlers. believe that any of the processes of nature and humans
Complex processes, on the other hand, are irreducible. are necessarily rule based. If incomputable processes
Neglecting any part of the process or severing any of the
do exist in nature for example, the breaking of waves
connections linking its parts usually destroys essential
on a beach or the movement of air masses in the atmo
aspects of the systems behavior or structure. You just
sphere then we could never see these processes manifest
cannot start slicing up systems of this complexity into
themselves in the surrogate worlds of their models. We
subsystems without suffering an irretrievable loss of the
very information that makes these systems a system.
(a) (b)
3 2
Surprise-Generating Mechanisms

The vast majority of counterintuitive behaviors shown by

complex systems are attributable to some combination of 2
the following five sources: paradoxes/self reference,
instability, incomputability, connectivity, and emergence.
With some justification, we can think of these sources of Figure 1 The impossible staircase.
Ecological Complexity 109

may well see processes that are close approximations to Recent work on harvester ants has shed considerable
these incomputable ones, just as we can approximate an light on the process by which an ant colony assesses its
irrational number as closely as we wish by a rational current needs and assigns a certain number of members to
number. However, we will never see the real thing in perform a given task. These studies identify four distinct
our computers, if indeed such incomputable quantities tasks an adult harvester ant worker can perform outside
exist outside the pristine world of mathematics. the nest: foraging, patrolling, nest maintenance, and mid
Connectivity. What makes a system a system and not den work (building and sorting the colonys refuse pile).
simply a collection of elements is the connections and So it is these different tasks that define the components of
interactions among the individual components of the the system we call an ant colony, and it is the interaction
system, as well as the effect these linkages have on the among ants performing these tasks that gives rise to
behavior of the components. For example, it is the inter emergent phenomena in the colony.
relationship between biota and abiota that makes an One of the most notable interactions is between
ecosystem. Each component taken separately would not forager ants and maintenance workers. When nest main
suffice. The two must interact for sustainable life to take tenance work is increased by piling some toothpicks near
place. Complexity and surprise often reside in these the opening of the nest, the number of foragers decreased.
connections. Apparently, under these environmental conditions, the
Emergence. A surprise generating mechanism depen ants engaged in task switching, with the local decision
dent on connectivity for its very existence is the made by each individual ant determining much of the
phenomenon of emergence. This refers to the way the coordinated behavior of the entire colony. Task allocation
interactions among system components generate unex depends on two kinds of decisions made by individual
pected global system properties not present in any of ants. First, there is the decision about which task to per
the subsystems taken individually. A good example is form, followed by the decision of whether to be active in
water, whose distinguishing characteristics are its natural this task. As already noted, these decisions are based
form as a liquid and its nonflammability, both of which are solely on local information; there is no central decision
totally different from the properties of its component maker keeping track of the big picture.
Figure 2 gives a summary of the task switching roles
gases, hydrogen and oxygen.
in the harvester ant colony, showing that once an ant
The difference between complexity arising from
becomes a forager it never switches back to other tasks
emergence and that coming only from connection pat
outside the nest. When a large cleaning chore arises on
terns lies in the nature of the interactions among the
the surface of the nest, new nest maintenance workers are
various component pieces of the system. For emergence,
recruited from ants working inside the nest, not from
attention is not simply on whether there is some kind of
workers performing tasks on the outside. When there is
interaction between the components, but also on the
a disturbance like an intrusion by foreign ants, nest main
specific nature of that interaction. For instance, connec
tenance workers will switch tasks to become patrollers.
tivity alone would not enable one to distinguish between
Finally, once an ant is allocated a task outside the nest, it
ordinary tap water involving an interaction between
never returns to chores on the inside.
hydrogen and oxygen molecules and heavy water (deu The ant colony example shows how interactions
terium), which involves interaction between the same among the various types of ants can give rise to patterns
components albeit with an extra neutron thrown in to of global work allocation in the colony, patterns that
the mix. Emergence would make this distinction. In prac could not be predicted or that could not even arise in
tice it is often difficult (and unnecessary) to differentiate any single ant. These patterns are emergent phenomena
between connectivity and emergence, and they are fre due solely to the types of interactions among the different
quently treated as synonymous surprise generating tasks.
procedures. A good example of emergence in action is Table 1 gives a summary of the surprise generating
the organizational structure of an ant colony. mechanisms just outlined.
Like human societies, ant colonies achieve things that no
individual ant could accomplish on its own. Nests are
erected and maintained, chambers and tunnels are exca Emergent Phenomena
vated, and territories are defended. All these activities are
carried on by individual ants acting in accord with simple, Complex systems produce surprising behavior; in fact,
local information; there is no master ant overseeing the they produce behavioral patterns and properties that just
entire colony and broadcasting instructions to the individual cannot be predicted from knowledge of their parts taken
workers. Somehow each individual ant processes the partial in isolation. These so called emergent properties are
information available to it in order to decide which of the probably the single most distinguishing feature of com
many possible functional roles it should play in the colony. plex systems. An example of this phenomenon occurs
110 Ecological Complexity

Environmental change Task switching

Midden worker Forager

Patrolling ant Forager

Nest maintenance worker Forager

Increase in food availability

Brood-care worker,
Nest mainten-
nest construction, seed
ance worker
storage, reserves

Detritus on surface of nest mound

Nest maintenance worker

Intrusion by foreign ants

Figure 2 Task switching in a harvester ant colony.

Table 1 The main surprise-generating mechanisms forming their average, the central limit theorem of prob
ability theory tells us that this average and the dispersion
Mechanism Surprise effect around it must obey the bell shaped distribution.
Paradoxes Inconsistent phenomena
Instability Large effects from small changes
Incomputability Behavior transcends rules Ecological Complexity
Connectivity Behavior cannot be decomposed into parts
Emergence Self-organizing patterns
Complexity research has discovered that many systems
display common structural and behavioral/dynamical
characteristics (Table 2). The interplay of these complex
when one considers a collection of independent random system characteristics entails systems to exhibit proper
quantities, such as the heights of all the people in New ties such as surprise, emergence, and power law scaling.
York City. Even though the individual numbers in this set Ecological complexity is the observation that ecological
are highly variable, the distribution of this set of numbers systems exhibit many of the same properties as physical
will form the familiar bell shaped curve of elementary complex systems, and thus an active research program has
statistics. This characteristic bell shaped structure can be arisen over the analysis of ecological data to see to what
thought of as emerging from the interaction of the com extent ecosystems share these common properties with
ponent elements. Not a single one of the individual other complex systems.
heights can correspond to the normal probability distri Ecosystems are composed of a large number of highly
bution, since such a distribution implies a population. Yet diverse components interacting with self stabilizing and
when they are all put into interaction by adding and self promoting feedback to produce emergent patterns. As
Ecological Complexity 111

Table 2 Some characteristics of complex systems

Structural characteristics Behavioral/dynamical characteristics

Large number of components Nonlinear

Large number of components Chaotic
High diversity of components and connections Catastrophic
Asymmetries Self-organization
Strong interactions Multiple steady states
Hierarchic organization Adaptive

such, ecological systems have been described as complex,

adaptive, hierarchical systems (CAHS) or self organized,
using knowledge of nonlinear phenomena to better
guide policy development for adaptation strategies
hierarchical open (SOHO) systems. Unlike with complex and mitigation to environmental change; and
physical systems, openness is a property that is required of
all ecological systems. This is because ecological systems
new tools and methods for studying ecological
are self perpetuating through means of capturing energy,
The emphasis on integrated natural and social systems
doing useful work (biochemical reactions, growth, and
addresses the growing interest to understand the role of
maintenance) to persist at least momentarily at a highly
human influence on the environment. There is a recent
organized state far from thermodynamic equilibrium this
awareness of the need to alter this influence in some fashion
is the metabolic process. A second ecological defining fea
both for ourselves and our environment. New tools and
ture is that organisms are able to replicate themselves such approaches incorporating self organization, emergence,
that the system outlives the constituent parts this is the and co adaptation are needed to improve our ability to
reproductive process. Therefore, it is often said of an manage and restore natural systems. These new approaches
ecosystem that, the whole is more than the sum of the to ecosystem management must also account for the natural
parts. dynamics and integrate concepts of sustainable develop
Two of the most pressing issues regarding ecological ment. Advances in ecological complexity science are
complexity are the need to develop appropriate measures essential in successfully navigating this transition.
to quantify the structural and behavioral complexity of
ecosystems, and to identify the underlying processes that
generate this complexity, through theory, analysis, mod
eling, and field studies. Would-Be Worlds
A new journal, Ecological Complexity (Elsevier), is one
forum for this research since 2004. The journal considers In the past few years, a number of electronic worlds have
papers dealing with biocomplexity related to the envi been created by researchers associated with the Santa Fe
ronment with an emphasis on interdisciplinary and Institute and elsewhere to study the properties of com
integrated natural and social systems science. plex, adaptive systems. The authors cite just three such
Topics typically found in the journal include: worlds here as prototypical examples of how to use the
computer as a kind of information laboratory to investi
aspects of biocomplexity in the environment and
gate such systems.
Tierra. This world, created by naturalist Tom Ray, is
systems; and biospheres as complex adaptive
ecosystems populated by binary strings that serve as electronic surro
gates for genetic material. As time unfolds, these strings
self organization of spatially extended ecosystems; compete with each other for resources, with which they
properties and structures of complex create copies of themselves. New strings are also created
by computational counterparts of the real world pro
ecological pattern formation in space and time; cesses of mutation and crossover. Over the course of
the role of biophysical constraints and evolutionary
attractors on species assemblages;
time, the world of Tierra displays many of the features
associated with evolutionary processes seen in the natural
ecological scaling (scale invariance, scale covariance,
and across scale dynamics), allometry, and hierarchy
world, and hence can be used as a way of experimenting
with such processes without having to wait millions of
theory; years to bring the experiment to a conclusion. But it is
ecological topology and networks; important to keep in mind that Tierra is not designed
studies toward an ecology of complex systems; to mimic any particular real world biological process;
complex systems approaches for the study of dynamic
humanenvironment interactions;
rather, it is a laboratory within which to study neo
Darwinian evolution, in general.
112 Ecological Complexity

TRANSIMS. For the past 3 years, a team of researchers behavior of these agents is determined by a set of extre
at the Los Alamos National Laboratory headed by Chris mely simple rules that constitute nothing more than
Barrett has built an electronic counterpart of the city of common sense rules for survival and reproduction. A typi
Albuquerque, New Mexico, inside their computers. The cal set of rules might be:
purpose of this world, which is called TRANSIMS, is to
1. Find the nearest location containing sugar. Go there,
provide a testbed for studying the flow of road traffic in an
eat as much as necessary to maintain your metabolism,
urban area of nearly half a million people. In contrast to
and save the rest.
Tierra, TRANSIMS is explicitly designed to mirror the
2. Breed if you have accumulated enough energy and
real world of Albuquerque as faithfully as possible, or at
other resources.
least to mirror those aspects of the city that are relevant
3. Maintain your current cultural identity (set of charac
for road traffic flow. Thus, the simulation contains the
teristics) unless you see that you are surrounded by
entire road traffic network from freeways to back alleys,
together with information about where people live and many agents of different types (tribes). If you are,
work, as well as demographic information about incomes, change your characteristics and/or preferences to fit
children, type of cars, and so forth. So here we have a in with your neighbors.
would be world whose goal is to indeed duplicate as With even such primitive rules as this, strange and won
closely as possible a specific real world situation. drous things begin to happen. A typical scenario is shown
Sugarscape. Somewhere between Tierra and in Figure 3, where we see the sugar marked by yellow
TRANSIMS is the would be world called Sugarscape, dots on the Sugarscape. The agents are initially distrib
which was created by Joshua Epstein and Rob Axtell of uted randomly on the landscape, red dots being agents
The Brookings Institution in Washington, DC. This that have a good ability to see food at a distance, blue dots
world is designed as a tool to study processes of cultural representing more myopic agents. It is reasonable to
and economic evolution. On the one hand, the assump expect that if no other considerations enter, natural selec
tions about how individuals behave and the spectrum of tion would tend to favor good vision over time. Indeed
possible actions at their disposal is a vast simplification of this is the case, as seen by the center panel in Figure 3,
the possibilities open to real people as they go through showing a preponderance of red agents in the population.
everyday life. On the other hand, Sugarscape makes fairly However, if the experiment is run again, giving agents the
realistic assumptions about the things that motivate peo possibility of passing wealth on to their offspring in the
ple to act in the way they do, as well as about how they go form of sugar, we find that inheritance has a pronounced
about trying to attain their goals. What is of considerable effect on survival. This is shown in the third panel of the
interest is the rich variety of behaviors that emerge from figure, in which many more agents having poor vision are
simple rules for individual action, and the uncanny able to survive by making use of sugar willed to them by
resemblance these emergent behaviors have to what is their parents.
actually seen in real life. Although this simple example is useful in illustrating
In order to conduct the kinds of repeatable, controlled the workings of the CompuTerrarium, it hardly suggests
experiments that natural scientists take for granted when
a revolution in our way of thinking about and studying
trying to understand and create theories of physical and
social structures. For that we need to add a lot more
engineering systems, Epstein and Axtell decided to grow
whistles and bells to the system. Epstein and Axtell have
a social order from scratch by creating an ever changing
done exactly this. When they add seasons so that sugar
environment and a set of agents who interact with each
concentrations change periodically over time, the agents
other and the environment in accordance with simple
begin to migrate. When a second resource, spice, is intro
rules of survival. An entire social structure trade, economy,
duced, a primitive economy emerges as a result of a new
culture then evolves from the interactions of the agents. As
elementary rule: Look around for a neighbor having a
Epstein remarks about social problems, You dont solve it,
commodity you need. Bargain with that neighbor until
you evolve it. Epstein and Axtell call their laboratory in
you reach a mutually satisfactory price. Trade at that
which societies evolve the CompuTerrarium. Here is how
it works. price. Figure 4 shows the effect of this type of trading
The interacting agents are each graphically repre
sented by a single colored dot on the landscape they
inhabit, which is called the Sugarscape. Every location
in the landscape contains time varying concentrations of
a food resource, called sugar. Each individual has a unique
set of characteristics; some are fixed like sex, visual range
Initial condition No inheritance With inheritance
for food detection, and metabolic rate, whereas others are
variable like health, marital status, and wealth. The Figure 3 Evolution on the Sugarscape.
Ecological Complexity 113

1 Agents forage for sugar and spice 2 If trade is allowed, they flourish 3 Without trade, many starve

Figure 4 The effects of trade in the Sugarscape.

economy. In the first part of the figure, agents are simply the creation of a viable theory of such processes. These
foraging independently for both sugar and spice. In the key components are given as follows.
middle panel we see the effect of beginning trade; now A medium sized number of agents. In contrast to simple
lots of agents flourish. Finally, the third panel shows the systems like superpower conflicts, which tend to involve
effect of turning off the trade. Without trade being a small number of interacting agents or large systems
allowed, many of the agents cannot survive. like galaxies or containers of gas, which have a large
There is certainly much more that can be said about enough collection of agents that we can use statistical
the social laboratory constructed by Epstein and Axtell. means to study them complex systems involve what
Issues involving the emergence of cultural groups, com we might call a medium sized number of agents. Just like
bat, institutional structures, and the like can all be Goldilockss porridge, which was not too hot and not too
introduced to study myriad questions of interest to cold, complex systems have a number of agents that are
social scientists. The interested reader will certainly not too small and not too big, but just right to create
want to consult the monograph by Epstein and Axtell interesting patterns of behavior.
that details these and many other matters. It is cited in Intelligent and adaptive agents. Not only are there a
the references for this article. Here we must content medium sized number of agents, these agents are intelli
ourselves with simply noting that the CompuTerrarium gent and adaptive. This means that they make decisions
offers a platform to study society from the bottom up. on the basis of rules, and that they are ready to modify the
With this view, we can explore social behavior that is rules they use on the basis of new information that
dynamic, evolutionary, and locally simple. What could becomes available. Moreover, the agents are able to gen
be better than to have a laboratory like this in which to erate new rules that have never before been used, rather
do such experiments? than being hemmed in by having to choose from a set of
The main point in presenting these discussions of preselected rules for action. This means that an ecology of
Tierra, TRANSIMS, and Sugarscape is to emphasize rules emerges, one that continues to evolve during the
two points: (1) we need different types of would be course of the process.
worlds to study different sorts of questions, and (2) each Local information. In the real world of complex sys
of these worlds has the capability of serving as a labora tems, no agent knows what all the other agents are
tory within which to test hypotheses about the doing. At most, each person gets information from a
phenomena they can represent. And, of course, it is this relatively small subset of the set of all agents, and
latter property that encourages the view that such com processes this local information to come to a decision
putational universes will play the same role for the as to how he or she will act. In the Sugarscape, for
creation of theories of complex systems that chemistry instance, what the traders adjacent to a given individual
labs and particle accelerators have played in the creation in the market are doing constitutes the local informa
of scientific theories of simple systems. Gleick has given a tion that the individual has available to help decide
fuller account of the technical, philosophical, and theore what to do next.
tical problems surrounding the construction and use of So these are the components of all complex, adap
these silicon worlds. tive systems like the Sugarscape, TRANSIMS, or
Tierra situations a medium sized number of intelli
gent, adaptive agents interacting on the basis of local
Conclusions information. At present, there appears to be no known
mathematical structures within which we can comfor
The key components in each and every complex, adap tably accommodate a description of any of these
tive system and a decent mathematical formalism to worlds. This suggests a situation completely analogous
describe and analyze them would go a long way toward to that faced by gamblers in the seventeenth century,
114 Hierarchy Theory in Ecology

who sought a rational way to divide the stakes in a Further Reading

game of dice when the game had to be terminated
Casti J (1992) Reality Rules: Picturing the World in Mathematics. I The
prematurely (probably by the appearance of the police Fundamentals, II The Frontier. New York: Wiley (paperback edition,
or, perhaps, the gamblers wives). The description and 1997).
analysis of that very definite real world problem led Casti J (1994) Complexification. New York: HarperCollins.
Casti J (1997) Would Be Worlds. New York: Wiley.
Fermat and Pascal to the creation of a mathematical Epstein J and Axtell R (1996) Growing Artificial Societies. Cambridge,
formalism we now call probability theory. At present, MA: MIT Press.
Gleick J (1987) Chaos. New York: Viking.
complex system theory still awaits its Pascal and Jackson E (1990) Perspectives of Nonlinear Dynamics. Cambridge:
Fermat. The mathematical concepts and methods cur Cambridge University Press, vols. 1 and 2.
rently available were developed, by and large, to Mandelbrot B (1982) The Fractal Geometry of Nature. San Francisco,
CA: W.H. Freeman.
describe systems composed of material objects like pla Nicolis J (1991) Chaos and Information Processing. Singapore: World
nets and atoms. It is the development of a proper Scientific.
theory of complex systems that will be the capstone Peitgen H O, Jurgens DH, and Saupe D (1992) Fractals for the
Classroom, Parts 1 and 2. New York: Springer.
of the transition from the material to the informational. Ray T (1991) An approach to the synthesis of life. In: Langton CG,
Taylor C, Farmer JD, and Rasmussen S (eds.) Artificial Life II, SFI
Studies in the Science of Complexity, vol. X, pp. 371 408. Reading,
MA: Addison Wesley.
Schroeder M (1991) Fractals, Chaos, Power Laws. New York: W.H.
See also: Emergent Properties; Hierarchy Theory in Freeman.
Ecology; Self-Organization. Stewart I (1989) Does God Play Dice? Oxford: Basil Blackwell.

Hierarchy Theory in Ecology

T F H Allen, University of Wisconsin, Madison, WI, USA
2008 Elsevier B.V. All rights reserved.

Need for Hierarchy Theory History of the Field

Hierarchy and Hypothesis Scale and Type
Hierarchical Levels Further Reading

Need for Hierarchy Theory the onset of hostilities when Nichols attacked Gleasons
paper at the 1926 International Congress of Plant
In ecology we need a body of theory to address relation Sciences. Hierarchy theorys focus on level of analysis
ships that are consequences of changing levels of analysis, offers such clarification. It lays subtle distinctions bare,
which call for altered definitions. For instance, the con so that definitions work for the ecologist instead of ecol
temporary fracas over definitions of plant competition ogists working for their definitions.
could benefit from recognizing differences in the scope If big, slow things were always on top, such that hier
and type of the framework used by the respective parti archical levels were only a matter of scale, the problem
sans. With distinctions between levels of analysis made would reduce to a straightforward technical scaling issue.
clear, each school may test their respective hypotheses in Not to underestimate the challenges of scaling in engi
peace aware of which theories actually compete and when neering, but that technical setting does not need
they merely address some other level of discourse. The something as grand as a theory to deal with hierarchies.
contentious literature surrounding overcompensation of But in ecology, scaling is complicated by higher ecologi
plants in response to losses to grazing was significantly a cal levels giving lower levels meaning. In ecology, the
matter of pulse versus press consumption in relation to move upscale to be more inclusive often changes signifi
different timescales for assessing recovery. The cance more than it invokes a change of size, and so we do
Clements/Gleason debate over the proper definition of indeed need a special body of theory to deal with differ
plant community might not have lasted the body of the ence of quality, not just quantity. Differences in scale
twentieth century had hierarchy theory been available at quickly become large enough to cause qualitative change
Hierarchy Theory in Ecology 115

in perception, which forces a change in the level of Ecology in particular invites many logical types because
analysis. Thus scale is soon embroiled in values, judg its hierarchies are so rich. A new type invokes new aggre
ment, and arbitrary choices, not just as an inconvenience, gation criteria, which come explicitly from observer
but as a necessity for proper understanding. While scaling decisions. Consider the difference between a community
as an engineering technicality actively ignores such messy conception of vegetation as opposed to the process
issues, hierarchy theory explicitly includes value based functional conception that prevails in ecosystem modeling.
decisions of the observer in creating hierarchies. A forest can be considered as a collection of trees on a tract
To control for observer values, technical measurement of land. Alternatively those same tree trunks may be aggre
and analysis in science keeps its criteria constant across the gated as a separate class from the leaves (Figure 1). If
local discourse. But large discoveries precisely amount to leaves in a forest are the production system independent
the recognition of a change in value. New scientific ideas of species, then the boles are part of the carbon storage
indicate a specific change in the preanalytic stage, before function. This assignment has the peculiar effect of unify
deciding what might be relevant data. In the terms of ing the tree trunks with soil carbon in a single carbon
Russell and Whitehead (made accessible by Gregory storage compartment. A community focus aggregates
Bateson), new scientific ideas amount to the definition of trees set in an environment of soil and atmosphere.
new logical types. Hierarchy theorys central activity is Meanwhile a flux process conception splits the trees into
recognizing logical type. Logical types are tied to some at least two parts, one of which aggregates with the soil. But
new level of inclusivity, a new hierarchical level with its the soil was part of the environment in the community
own meaning. Notice how left and right sides are possessed conception. Thus the same pieces of soil and plant biomass
by organisms at their own level of existence. Meanwhile, are aggregated into different higher units, depending on
the notions of up and down refer to a larger discourse that the type of system that is recognized as being in the fore
includes an environment, which is shared by many organ ground by the observer. Note how forests under either
isms. As a result, a mirror switches the image left and right, conception may be called forest ecosystems, suggesting
that one use of hierarchy theory is to untangle alternative
but with no switch in up versus down. The larger scope
meanings in commonplace ecological terminology. The
invoked by the idea of up introduces a new logical type,
difference between a process focused ecosystem and a
even though left and right may often be simply at right
community is a change in logical type.
angles to up and down. If left and right contrasted with up
and down can be problematic, ecosystems are a nightmare.
While exquisitely holding criteria constant in formal scien
tific calibration will help, it is insufficient for large
Hierarchy and Hypothesis
discoveries, which turn on recognizing when a new type
Hierarchy theory is a body of thought that relates chosen
is necessary to solve some puzzle.
levels of analysis to defined levels of organization, all in

Biota and Process

environs ecosystem

Community Environs

Carbon Carbon
storage capture
Species Species Soil Air

Tree Tree Soil

bole carbon Leaves

Leaves Bole Roots

Figure 1 A community conception leads to an expected situating of whole trees in an environment. But a process-functional
ecosystem conception of that same forest can lead to a hierarchy where tree boles are separated from the leaves, and then united with
soil elements in a carbon storage compartment. Each respective hierarchy takes its form from the purpose for which it is intended.
116 Hierarchy Theory in Ecology

the context of scaling. It advises the scientist of subtle but because of inequivalent size. Host versus parasite is the
crucial distinctions that follow from observing and mak basis of a hierarchy employing levels subtly different from
ing analytical decisions. A significant part of hierarchy those in the population/organism distinction. Hierarchy
theory is observation in relation to conceptions of order in theory places entities in levels, taking care to be explicit
complex systems that would otherwise invite confusion. It about the definitions that lead to those levels, and the
is a metatheory that guides the generation, fine tuning criteria that create order and linkage.
and testing of other bodies of thought, themselves more Scale versus definition has potential for generating
easily recognizable as theories in the conventional sense. different sorts of hierarchies. Some hierarchies focus on
Some theories in ecology are associated with answering size and containment, while others are control hierarchies
questions taken as given. Such theory is validated where upper level entities simply control lower levels. A
in testing hypotheses. Other ecological theories may Watt governor may be placed at a higher level in a control
fine tune questions, perhaps clarifying what is meant by hierarchy, while being smaller than the whole steam
competition, so that worthwhile hypotheses may be gen engine it controls. Whether it is a scalar or a control
erated. By contrast, hierarchy theory applies in the hierarchy depends on the use for which the hierarchical
preanalytical stage, when the questions are being framed conception is intended, something for which the observer
rather than when clarified or answered. In the preanaly must take responsibility. Time against space plots are
tical stage, the boundaries of things are established, and popular in landscape ecology. But such hierarchies can
structures are assigned to types or classes. With the dis miss out on the interesting situations where large space
course laid out unambiguously by hierarchy theory, other maps onto short time spans, or long time spans map onto
theories may come into play, testing explicit hypotheses small places (Figure 2). The globe is large enough to be
with measurement and models. Thus hierarchy theory the context of continental movement over hundreds of
does not have its own hypotheses per se, but rather opens millions of years, but at the same time the rotation of the
the way for subsequent testing of specific hypotheses. globe is also responsible for diurnal phenomena, at the fast
Like multivariate description in ecology, hierarchy the
end of ecological happenings. Surfaces arise when narrow
ory focuses on hypothesis generation and clarification.
space applies to large differences in time constants (strong
From a small number of first principles, it highlights
temporal connection within, but weak connections across
what would be otherwise taken for granted and then
surfaces). Ecotones would be a case in point, because
forgotten in the muddle that ensues. Hierarchy theory is
there is rapid exchange and fast process inside the abut
explicit, as it positions the tacit next to the focal. Its
ting ecosystems or communities areas, while the
precision is in thought and choosing definitions, more
exchanges across the narrow ecotone may be remarkably
than action in quantitative experimentation.
slow. Thus ecotones are spatially small, while

Hierarchical Levels Not often plotted Conventional linear plot

timespace for biota and
but common and climate
Entities in a hierarchy are recognized as belonging to important n
levels. Levels are sets, but the sets become levels because wide r t h s ro 00 km Ice age
Ea 35 0
of robust asymmetry between them in a hierarchy. 24 h
Mathematically, the asymmetry between levels makes
rs ars
hierarchies partially ordered sets. Hierarchy theory is r rido Community 6 ye
Co 10
the set theory that may precede network theory (see 6
Canopy & ait lia
km s Str stra
Ecological Network Analysis, Environ Analysis). A level 50 r e Au
of analysis assigns entities to levels, and is often explicit Tree To lates
iso Not often
about their relationship to other entities assigned to other Short
narrow Leaf plotted but
levels. There is a distinction between levels of observation ne still common
ce oto
and levels of organization. Levels of observation are Fen Ec and important
ordered relative to each other on matters of size and Fast Slow
scale. Meanwhile relationships between levels of organi Figure 2 A common graph appearing in landscape ecology
zation follow from definitions chosen by observers, plots increases in time against space, focusing on the
sometimes as a prelude to actual observation. For quasi-linear pattern of larger things seen as behaving over longer
instance, organisms are subsumed by populations only time periods. But such plots ignore potential control systems and
their hierarchies. Local intransigence can control large entities
by a definition. Hidden in the definition is a requirement (Wallaces realms where Australias fauna is isolated from Asia by
for equivalence between population members. the narrow Torres Strait, separating millions of years of
Meanwhile, a host and its parasite, while both organisms, evolution). Barriers and surfaces occur at the lower right, while
are generally not assigned to the same population, in part communication channels and corridors appear upper left.
Hierarchy Theory in Ecology 117

representing slow exchanges that might cause the ecotone operates largely in the realm of epistemology, as a theory
to move in a process of gradual encroachment. of observation and analysis. The discourse generally takes
Conversely, in a communication channel, small differ the position that hierarchies appear somewhere between
ences in time constants apply along the long connection. the material world and human understanding. If there are
Corridors would be an example here, where there is rapid complex material systems that are not hierarchic, we might
movement along the extended length of the corridor. In expect to have great difficulty in observing or understand
ecology, these special places, such as ecotones and corri ing them. There appear to be points of passage of
dors, are at least as interesting as situations where time information up hierarchies, where details are explicitly
and space widen in concert. Complexity in hierarchies lost. A military command is a favorite hierarchic example
arises from the challenge of mapping between levels, as of a human organization. There, details of how an indivi
scale and definition entwine. dual soldier observed local enemy concentrations fall away
as the intelligence passes up the command. Setting the
detail aside allows the top brass to make sweeping deci
sions, without being encumbered by a blizzard of local
History of the Field happenings. Not only must the general in command let
go of details of grains so as to get a handle on the wide
Hierarchy theory has its roots in economics and business extent, but so too must the observers of hierarchical struc
administration of the 1960s, suggesting that the world ture. To understand what a general is doing, the observer
appears nearly decomposable. We can decompose wholes of a command structure needs to integrate away the details
into parts, but only to a degree, in that parts communicate inside the army. By the late 1960s, the notion of hierarchy
and leak onto each other. Complete decomposability had moved beyond administration systems, and was being
would deny upper level structures existence. Completely taken up across a range of disciplines.
decomposed, the parts would not to be able to commu In the following decade, hierarchy theorists from phys
nicate with each other in making the larger whole. Parts ics addressed hierarchical complexity after Heisenberg,
have strong connections within, but weak connections invoking dualities, uncertainty, and complementarity
between, and those weak connections may be precisely between dynamics versus structure. Important develop
what links hierarchical levels (Figure 3). ments have turned on the tension across the dilemmas
While some practitioners in subsequent studies have presented by dual structures, as in the holon, a general
sought real hierarchies in an external world, much of the ized entity in a hierarchy. Holons have been equated with
early literature of business administration hierarchies is the concept of system, with the advantage that holon does
agnostic about the ultimate reality of hierarchical structure. not appear in common parlance. The holon can therefore
In this spirit, hierarchy theory in social organizations escape the reification and slovenly usage in the vernacular
where the model is mistaken for the materiality.
Levels and strength of connection Conceptual developments suggest that what is inside a
given holon is chosen by an observer. This emphasizes
N+3 that holons are abstractions more than material objects; a
N+2 point forgotten when system is used for holon. In the
N holon, the tension is between system and subsystem. But
the subsystem is a system in its own right, thus offering
some sort of dual existence that invites contradiction.
The concept of the holon takes the whole to be a
surface that integrates the parts to give a unified signal
N + 3 High Weak connection N+3
Weak relative to But strong relative
to the rest of the universe. At the same time, the holon is
N + 2 Middle strong below N+2
to weak above the surface that integrates the external environment for
N + 1 Low Strong connection the parts to experience. In ecology, the environment falls
Figure 3 In nested hierarchies, the bonds that unite members
away at the level of the holon when viewed from the
of the lowest level N are strong, as they make entities of level perspective of the parts. A forest raises humidity and
N 1. The bonds that create level N 2 are weaker. lowers temperature, thus allowing survival of some of its
Nevertheless, these weaker bonds appear as the strong bonds parts, tree seedlings that are its future. The parts are
making the entities at level N 2 when seen from level N 3. If protected. Conversely, viewed from the context of the
the two largest units in the figure are molecules, then the entities
at N 1 are atoms. Breaking their atomic bonds releases huge
hot, dry environment surrounding the forest, the contri
amounts of atomic energy as subatomic particles, N, are freed. butions of each tree to the water vapor inside the forest
Atomic bonds are stronger and release more energy when are lost in a more general flux of water from the canopy.
broken, compared to breaking the chemical bonds that make Thus, loss of information occurs with movement both up
molecules, the N 2 entities. and down the hierarchy. While the environment is too
118 Hierarchy Theory in Ecology

large to catch the details of the working of the parts, the particularly useful for exploration before firm criteria
parts themselves cannot span wide enough to see large, connecting levels have been established. Concomitantly,
slow differences in the larger context. In all this, we see non nested hierarchies are for mature ideas, where
again the tension embodied in hierarchical discussions focused sets of relationships are organized and abstracted
between scale, organization, and uncertainty in in a control system.
observation. In thermodynamic studies of ecological emergence,
Earlier, five general principles for ordering ecological nested hierarchies are essential, because otherwise the
hierarchies were recognized: bookkeeping of energy flow between the system and its
environment would not sum. In complexity theory,
(1) As to frequency of behavior, higher level holons
self organized emergence is a matter of thermodynamic
operate at a lower frequency, taking longer to exhibit
gradients being applied to material systems that are
returns in behavior than holons at lower levels.
pushed away from equilibrium. Thus, nested hierar
(2) Higher levels in a hierarchy constrain lower levels by
chies apply when self organization is invoked, when
displaying intransigent constancy. Deans constrain
holons emerge at a new level without any plan.
faculty by not changing the budget, except once a
Planned systems often yield to a non nested concep
tion. A surprising and important new turn in applied
(3) Higher levels in a hierarchy will be contextual to
ecology of human management systems links non
lower levels. The environment would be seen as
nested human socioeconomic hierarchies to nested
operating at a higher level.
thermodynamic hierarchies. The whole system is embo
(4) With regard to bond strength, higher level holons are
died in energy flow and control through the twinned
held together by weaker forces than those that inte
social and biogeochemical hierarchies.
grate lower level holons (e.g., chemical vs. nuclear
These thermodynamic approaches develop self
bonds) (Figure 3).
organizing holarchic open systems (SOHOs), using the
(5) As to containment, if higher level holons consist of
term holarchy for nested hierarchies. The word holar
lower holons, which they contain, then the hierarchy
chy appears in part to sidestep the political
is said to be nested. Not all the criteria apply to all
unacceptability of hegemonic hierarchical control.
hierarchies, but all five principles may apply
Using the SOHO approach, the full power of hierarchy
theory in solving real time problems has been developed
The distinction between nested and non nested hierar by Waltner Toews and colleagues at NESH, a Canadian
chies matters (Figure 3). In nested systems, upper level centered, complex systems group. They solved some
entities contain and consist of lower level entities. In non critical problems in Peru, Kenya, and Nepal. For
nested hierarchies, containment is not a criterion, but instance, a Kathmandu sewer had children playing
principles (1)(3) can still apply. In nested hierarchies, around slaughterhouse waste. By linking the social hier
containment applies even if aggregation criteria between archy to the ecological process hierarchy, NESH
levels change type. Western medicine generally uses identified that a street cleaner caste was being blamed
nested hierarchies for the human condition. Thus orga for things out of their control. Blaming scapegoats had
nelles may be aggregated into cells by biochemical led to inaction and paralysis, but once the street cleaners
interaction. Meanwhile, nesting of organs inside the were no longer held responsible, the SOHO thermody
whole body may invoke fluid mechanics as a principal namic methodology achieved significant rehabilitation
on which parts make the whole person. When whole as the social and ecological hierarchies began to function
humans nest inside groups, relationships may be in epi in concert.
demiological terms. In Western medicine, there are The earliest explicit introduction of hierarchy the
regular changes in aggregation criteria from biochemical, ory into ecology in the 1970s spoke of decomposability
through fluid dynamic, to epidemiological. Despite as an issue in some of the biomes studied in the
inconstant criteria for linking levels, the nesting keeps International Biological Program (IBP). At that time,
such hierarchies straight. But in non nested hierarchies, terms, such as environ, creaon, and genon were
such as food chains or pecking orders, the top dog neither coined as extensions of the concept of holon. Environ
contains nor consists of the subordinate individuals. addresses the environment acting as an integrated
Because there is no nesting to maintain order, non nested whole for its residents (see Ecological Network
hierarchies embody only one specific rule for moving Analysis, Environ Analysis). The inward direction
between levels. As a result, the criteria for moving up a toward the holon pertains to the creaon, whereas the
food chain must be consistently is eaten by, or conver outward direction pertains to the genon generating new
sely going down it is eats. In this way, the hierarchy is things and experiences for the environment and its
consistent top to bottom. Because of their robustness to residents. Holon remains the central concept. The first
changes in aggregation criteria, nested hierarchies are fully integrated treatments of hierarchies in ecology
Hierarchy Theory in Ecology 119

turned on epistemological implications of scale and may be variously inclusive of species across narrow or
dynamics. Following shortly, evolutionary ideas focused wider areas. Under the organism criterion, examples are
on the structural elements in hierarchies, in a more found from redwood trees to mites. An ecological hier
ontological spirit. The structural elements were cast as archy may change the scale and type at the same time,
a triadic view of holons, where the level above and the but it is fraught with conceptual danger. Indeed, hier
level below, as well as the level of the holon in archy theory is often invoked to clean up the mess in
between, are all required for an adequate treatment. the aftermath of scale and type being mixed together.
Recently, two more crucial levels were added: the There is no prohibition changing both together, but
level above the context keeps the context of the only so long as the relationships at each new level
holon stable, while the level below the parts provides are explicit. This matters because most descriptions of
stability for the material of which the parts are made. ecological material precisely do change type across
widening scalar levels, although most of them do not
follow the textbook ordering from organism to bio
Scale and Type sphere. For instance, in a forest community, a rotting
tree trunk may be considered an ecosystem, whose
Scale problems invited hierarchy theory into the disci upper surface is landscape, on which grows a commu
pline. Ecologists have long been aware of scale, nity of bryophytes.
investigating the properties of quadrats in obtaining esti The copious variety of materials, entities, and sizes in
mates of vegetation in the 1950s. Then change in variance ecology invites hierarchy theory into ecology. Indeed, it is in
across quadrat size was used to measure aggregation of ecology that hierarchy theory has been used most often to
plants on the ground. Hierarchy theory remains asso significant effect, as in the NESH studies mentioned above.
ciated with scale today. The observation protocol brings Hierarchy theory can capture a rich set of scaled examples
attention to a universe of a certain extent, while making a across a mixture of types. Ecology is a multiple scaled labyr
second distinction, the finest grain at which observation inth of types. Hierarchy theory is the ball of string that
units are distinguished from one another. Grain and we can trail behind, so that ecological scientists do not get
extent together characterize the scalar level in question lost.
in many ecological hierarchies. Grain and extent are
connected. Wider extents require coarser grains, if the
mass of data are to be remembered, analyzed, and under See also: Ecological Network Analysis, Environ Analysis.
stood. Modern computational power has widened the gap
between grain and extent, where remotely sensed areas
are captured in billions of pixels. Even so, explicitly link Further Reading
ing items in the grain across the extent becomes Ahl V and Allen TFH (1996) Hierarchy Theory, A Vision Vocabulary and
difficult, and generally impossible as the extent widens Epistemology. New York: University of Columbia Press.
by much. Allen TFH and Hoekstra TW (1992) Toward a Unified Ecology.
New York: University of Columbia Press.
In contrast to linking across scales, it is possible to Allen TFH, ONeill RV, and Hoekstra TW (1984) Interlevel relations in
unify ecology across types of ecological system that ecological research and management: Some working principles from
correspond to the main subdisciplines of ecology: hierarchy theory. General Technical Report R.M.110. Fort Collins:
USDA Forest Service (republished in 1987 in Journal of Applied
organism; population; community; ecosystem; land Systems Analysis 14: 63 79).
scape; biome; and biosphere. These types for Allen TFH and Starr TB (1982) Hierarchy: Perspectives for Ecological
ecological subdisciplines are explicitly not scale based, Complexity. Chicago: University of Chicago Press.
Kay J, Regier H, Boyle M, and Francis G (1999) An ecosystem approach
and so are not required to be assigned to level in the for sustainability: Addressing the challenge of complexity. Futures
order given in the previous sentence. When scale is 31: 721 742.
parsed away from type, the various approaches to ecol Koestler A (1967) The Ghost in the Machine. Chicago: Gateway.
ONeill RV, DeAngelis D, Waide J, and Allen TFH (1986) Monographs in
ogy achieve a sharper depth of focus, offering clear Population Biology 23: A Hierarchical Concept of Ecosystems.
relief between types of investigation. The subdisciplines Princeton: Princeton University Press.
of ecology are not scalar levels. If they are levels at all, Overton WS (1975) Decomposability: A unifying concept? In: Levin S
(ed.) Proceedings of the SIAM SIMS Conference on Ecosystems
they are type based levels of organization, with the Analysis and Prediction, pp. 297 299. Philadelphia: Society for
different types related to one another by asymmetric Industrial and Applied Mathematics.
relationships made explicit in the definitions. As a Patten BC (1978) Systems approach to the concept of environment.
Ohio Journal of Science 78: 206 222.
separate issue, a typed level of organization itself con Pattee HH (ed.) (1973) Hierarchy Theory: The Challenge of Complex
tains scale based hierarchies, as in fractal landscapes. In Systems. New York: Braziller.
that scaled universe, the ecosystem modeling strategy Salthe SN (1985) Evolving Hierarchical Systems. New York: Columbia
University Press.
may apply across a range of sizes, where local processes Simon HA (1962) The architecture of complexity. Proceedings of the
are part of more global processes. Communities too American Philosophical Society 106: 467 482.
120 Goal Functions and Orientors

Waltner Toews D, Kay JJ, Neudoerffer C, and Gitau T (2003) Relevant Websites
Perspective changes everything: Managing ecosystems from
the inside out. Frontiers in Ecology and the Environment
1(1): 23 30. James Kay Web Page, Network for
Webster JR (1979) Hiearchical organization of ecosytems. In: Halfon E Ecosystem Sustainability and Health (NESH).
(ed.) Theoretic Systems Ecology, pp. 119 131. New York: Academic Stanley N. Salthe Web Page, Center for
Whyte LL, Wilson AG, and Wilson D (1969) Hierarchical Structures. the Philosophy of Nature and Science Studies (CPNSS),
New York: Elsevier. Niels Bohr Institute.

Goal Functions and Orientors

H Bossel, University of Kassel (retd.), Zierenberg, Germany
2008 Elsevier B.V. All rights reserved.

Introduction Simulation of the Evolution of System Orientation

System Concepts Further Reading
System Orientation in a Complex Environment

Introduction When we talk about a viable system, we mean that this

system is able to survive, be healthy, and develop in its
The global ecosystem is made up of an ensemble of inter particular environment. In other words, system viability has
acting local and regional ecosystems, each composed of something to do with both the system and its properties, and
biotic and abiotic subsystems. The evolution of these with the environment and its properties. And since a system
systems is constrained by physical and system laws and by usually adapts to its environment in a process of coevolu
the basic properties of their environment, including the tion, we can expect that the properties of the systems
constraints of exergy (energy that can be usefully trans environment will be reflected in the properties of the sys
formed into work), material, and information flows. tem; for example, the form of a fish and its mode of motion
Sustainability (persistence) of a system in its environment reflect the laws of fluid dynamics of its aquatic environment.
therefore requires respecting these constraints. Conversely, Systems are termed complex if they have an internal
the very fact of its persistence demonstrates that a system structure of many qualitatively different processes,
has successfully adapted to its operating conditions. subsystems, interconnections, and interactions. Besides
Evolution has forced it to respect physical and system laws assuring their own viability, the individual systems that
and the basic properties of its environment. To an observer, are part of a complex total system specialize in certain
the systems behavior appears to be guided by a particular functions that contribute to the viability of the total sys
attractor state, or by attention to a number of orientors. tem. Viability of subsystems and the total system requires
that subsystem functions and interactions are organized
efficiently (or at least effectively). In the evolution of
complex systems, two organizing principles in particular
System Concepts have established themselves: hierarchy and subsidiarity.
System Organization They can be found in all successful complex systems:
biological, ecological, social, political, technological.
System is anything that is composed of system elements
Hierarchical organization means a nesting of subsys
connected in a characteristic system structure (Figure 1).
tems and responsibilities within the total system. Each
This configuration of system elements allows it to per
subsystem has a certain degree of autonomy for specific
form specific functions in its environment. These
actions, and is responsible for performing certain tasks
functions can be interpreted as serving a distinct system
contributing to the viability of the total system. For
purpose. The system boundary is permeable for inputs
example, body cells are relatively autonomous subsys
from, and outputs to, the environment. It defines the
systems identity and autonomy. tems, but contribute specific functions to the operation
Goal Functions and Orientors 121

System environment input (mostly solar energy) and a finite stock of materials.
The global ecosystem is therefore forced to recycle all of
Feedbacks its essential material resources. The development of local
S1 ecosystems is constrained by the local rate of exergy flux
System output (solar radiation input) and by the local rate of material
S2 recycling (weathering rate, absorption rate, decomposi
System tion rate, etc.) that it produces.
elements S4
Evolution favors those species or (biotic) subsystems of
System the ecosystem that have learned to use available resources
more efficiently and effectively than their competitors.
System inputs System structure This learning is embedded in their genetic code, and it is
System boundary manifest in the dissipative structures they construct. Both
Figure 1 System notation. will increase in complexity as a species evolves. At the
ecosystem level, species evolution will cause increasingly
of particular body organs, which in turn contribute to the better use of (exergy and material) resources. Species as
viability of an organism. well as ecosystems as a whole therefore tend to progress
Subsidiarity means that each subsystem is given the toward more complex dissipative structure producing
responsibility and the means for keeping its own house in more complex behavior.
order, within the range of its own abilities and potential. Interacting species in a common ecosystem coevolve
Only if conditions occur that cannot be handled by the in the direction of increasing fitness of each individual
subsystem would the suprasystem step in and help. The species. Evolution of ecosystems therefore proceeds in the
principles of hierarchical organization and subsidiarity
direction (arrow of time) of specialization, speciation,
require that each subsystem has a certain measure of
synergy, complexification, diversity, maximum through
autonomy. In its particular environment, each subsystem
flow of exergy, and more efficient use of material
must be viable. The total system can only be viable if each
resources. This development becomes manifest in the
of the subsystems supporting it is viable. Each subsystem
corresponding emergent properties: exergy degradation,
reflects the properties of its individual environment; its
recycling, minimization of output, efficiency of internal
behavior is informed (oriented) by that environment.
flows, homeostasis and adaptation, diversity, hetero
Note that this way of looking at complex systems is
geneousness, hierarchy and selectivity, organization,
recursive. If necessary, we can apply the same system/
minimization of maintenance costs, storage of available
subsystem dichotomy of viable systems again at other
organizational levels. For example, a person is a subsys resources. These properties can be viewed as orientors,
tem of a family; a family is a subsystem of a community; a propensities, or attractors guiding system evolution and
community is a subsystem of a state; a state is a subsystem development. They are not limited to ecosystems; they
of a nation, etc. are a general feature of living systems, including human
It is not enough to be concerned with the viability of organizations. When quantified and used in models, we
individual systems. There are no isolated systems in the refer to them as goal functions.
real world; all systems depend in one way or another on In particular, ecosystems will therefore build up in the
other systems. Hence their viability, and ultimately the course of their development as much dissipative structure as
viability of the total system are also preconditions for can be supported by the available exergy gradient. Available
sustainable development. This means that a holistic sys opportunities will eventually be found out by the processes
tem view must be adopted. of evolution, and will then be utilized. The ability to
respond successfully to environmental challenges can be
interpreted as intelligent behavior, although it is strictly
Evolution of Systems, and Emergence of the result of nonteleological evolutionary development.
Orientors and Goal Functions
The adaptation of a system to its environment is reflected
in its structure, including its nonmaterial, cognitive struc
ture. This system structure determines its behavior, and System Orientation in a Complex
hence the adaptive response to its particular environment. Environment
System structures of material systems are dissipative; they
require exergy and material flows for their construction, Basic concepts can be introduced by visualizing a simple
maintenance, renewal, and reproduction. animal with limited vision in a simple environment. The
The dissipative structures of the global ecosystem are animal requires exergy for self organization, motion, har
constructed and maintained by a finite rate of exergy vesting food, and maintenance. The environment provides
122 Goal Functions and Orientors

food in certain locations, usually associated with obstacles (explicit or implicit) normative concepts that direct beha
that must be avoided since they have an exergy cost. vior and development of systems in general. In the social
In a stable environment where sufficient (regenerat context, values and norms, objectives and goals are impor
ing) food is distributed in a completely regular pattern, tant orientors. Ecosystems and organisms tend toward
evolutionary adaptation would eventually lead to optimi certain attractor states whose specific characteristics can
zation of an animals movements in a regular grazing be viewed as orientors. Orientors exist at different levels
pattern, with a single objective, optimum exergy uptake of concreteness within an orientor hierarchy. The most
and use. The regular grazing pattern reflects the complete fundamental orientors, the basic orientors, are identical
certainty of the next step, which the animal learns by for all complex adaptive systems. Orientors are dimen
accumulating and internalizing experience in a cognitive sions of concern; they are not specific goals. Their
structure aiding its limited vision. satisfaction can be determined by observation of corre
In more complex and diverse environments the ani sponding indicators, which can also be used to define goal
mal, because of its limited vision, may not know for functions for model studies.
several steps which situation it will encounter next. It In addition to the physical constraints of exergy and
will therefore have to develop decision rules that have material flows, ecosystem and species development
greater generality and are applicable to (and will be is determined by the general properties of the environment:
reinforced by) different motion sequences with different
1. Normal environmental state . The actual environmental
outcomes. In addition to the requirement of harvesting
state can vary around this state in a certain range.
and using exergy resources effectively and efficiently,
2. Scarce resources. Resources (exergy, matter, informa
another objective is now implicitly added, to secure food
tion) required for a systems survival are not
under the constraint of incomplete information, that is, a
immediately available when and where needed.
security objective. Note that this is an emergent property
3. Variety. Many qualitatively very different processes
that is not explicit in the reward system (which still
and patterns occur in the environment constantly or
rewards only food uptake). Failure to heed this implicit
security objective will reduce food uptake and may
4. Reliability. The normal environmental state fluctuates
endanger survival. On the other hand, the pressure to
in random ways, and the fluctuations may occasionally
play it safe will occasionally mean giving up relatively
take it far from the normal state.
certain reward. With other words, efficiency is traded for
5. Change. In the course of time, the normal environmen
more security, and both are now prominent normative
tal state may gradually or abruptly change to a
orientations (goals, values, interests) incorporated in the
permanently different normal environmental state.
cognitive structure.
6. Other systems. The behavior of other systems changes
Orientation theory deals in a more general way with
the environment of a given system.
the emergence of behavioral objectives (orientors) in self
organizing systems in general environments. The propo
sition is that if a system is to survive in a given Basic Orientors
environment characterized by a specific normal envi
If evolution enforces fitness of (natural) systems, then
ronmental state, sparse resources, variety, unreliability,
persistent systems must reflect the properties of their
change, and the presence of other systems it must be
environment in their structure. More generally, the
able to physically exist in (be compatible with) this envi
basic properties of the environment require correspond
ronment, effectively harvest necessary resources, freely
ing basic system features. Since the basic environmental
respond to environmental variety, protect itself from
properties are independent of each other, a similar set of
unpredictable threats, adapt to changes in the environ
independent system features must exist, and it must find
ment, and interact productively with other systems.
expression in the concrete features of the system
These essential orientations emerge in the course of the
systems evolution in its environment.
There is a one to one relationship between the prop
erties of the environment and the basic orientors of
Properties of Environments systems (Figure 2):
There is obviously an immense variety of system envi 1. Existence. Attention to existential conditions is necessary
ronments, just as there is an immense variety of systems. to insure the basic compatibility and immediate survival
But all of these environments have some common general of the system in the normal environmental state.
properties. These properties will be reflected in systems. 2. Effectiveness. In its efforts to secure scarce resources
These reflections, or basic orientors, orient not just struc (exergy, matter, information) from, and to exert influ
ture and function of systems, but also their behavior in the ence on its environment, the system should on balance
environment. The term orientor is used to denote be effective.
Goal Functions and Orientors 123

In the assessment and orientation of system behavior,

environmental we deal with a two phase assessment process where each
state phase is different from the other.
Phase 1. First, a certain minimum satisfaction must be
Scarce guaranteed separately for each of the basic orientors.
reliability resources A deficit in even one of the basic orientors threatens
Existence long term survival. The system will have to focus its
attention on this deficit.
Security Efficiency
Phase 2. Only if the required minimum satisfaction of
all basic orientors is guaranteed is it permissible to try to
raise system satisfaction by improving satisfaction of indi
Adaptability Freedom vidual orientors further.
Adequate satisfaction of each of the basic orientors
Coexistence requires, on a lower level, system and environment
Environmental Environmental
change variety specific satisfaction of thermodynamic, structural, func
tional, ecophysiological, and system orientors. Network
analysis suggests complementarity of different formula
Other tions of extremal principles as orientors describing
systems ecosystem development.
Characteristic differences in the behavior of otherwise
Figure 2 A tentative typology of emergent properties.
very similar systems (animals, humans, political, or cultural
groups) can often be explained by differences in the rela
3. Freedom of action. The system must have the ability to tive importance attached to different basic orientors
cope in various ways with the challenges posed by (i.e., emphasis on freedom, or security, or effectiveness, or
environmental variety. adaptability) in phase 2 (i.e., after minimum requirements
4. Security. The system must have the ability to protect for all basic orientors have been satisfied in phase 1).
itself from the detrimental effects of variable, fluctu The basic orientor proposition has three important
ating, unpredictable, and unreliable environmental implications:
1. If a system evolves in a normal environment, then that
5. Adaptability. The system should be able to change its
environment forces it to implicitly or explicitly ensure
parameters and/or structure in order to generate more
minimum and balanced satisfaction of each of the basic
appropriate responses to challenges posed by changing
orientors (and of lower level orientors contributing to
environmental conditions. this satisfaction).
6. Coexistence. The system must modify its behavior to 2. If a system has successfully evolved in a normal envi
account for behavior and interests (orientors) of other ronment, its behavior will exhibit balanced satisfaction
systems. of each of the basic orientors.
Obviously, the system equipped to secure better overall 3. If a system is to be designed for a normal environment,
orientor satisfaction will have better fitness, and will proper and balanced attention must be paid to satisfac
therefore have a better chance for long term survival tion of each of the basic orientors.
and sustainability. In persistent systems or species, these The third implication has particular relevance for the crea
orientors will be found as emergent objectives (or system tion of programs, institutions, and organizations in the
interests). sociopolitical sphere, among other things. Note that for a
specific system in a specific environment, each orientor will
have a specific meaning. For example, security of a nation is
Properties of Orientors a multifacetted objective set with very different content
from the security of an individual particular organism.
Each of the basic orientors stands for a unique require However, the systems theoretical background for satisfac
ment. Attention (conscious or unconscious) must tion of the security orientor is the same in both cases.
therefore be paid to each of them, and the compensation
of deficits of one orientor by over fulfillment of other
orientors is not possible. Fitness forces a multicriteria Orientors as Implicit Attractors
response, and comprehensive (conscious or unconscious) Better orientor satisfaction (better fitness) for more parti
assessments of system behavior and development must cipants in a system requires more dissipative structure,
also be multicriteria assessments. which requires more exergy throughput as well as exergy
124 Goal Functions and Orientors

accumulation. Since the exergy flow of ecosystems is goal (where the final state is specified). These attractors
limited (capture of solar radiation by photoproduction), do not determine the exact future states of the system at
increasingly better utilization is to be expected in the all; they only pose constraints on choices (or evolutionary
course of system development. This saturates at maxi selection). The process and its rules are known, the
mum exergy flow utilization for the ecosystem as a whole. product is unknown. The spectrum of (qualitatively dif
Ecosystems as a whole therefore move in the direction of ferent) possible future development paths and sustainable
using all available exergy gradients. For organisms in the states remains enormous. The shape of the future, and of
ecosystem, this implies development tendencies (orien the systems that shape it, cannot be predicted this way. All
tors, propensities, attractors) toward specialization (using one can say with certainty, however, is that (1) all possible
previously unused gradients), more complex structure futures must be continuous developments from the past,
(greater use efficiency), larger individuals (less mainte and (2) paths with better orientor satisfaction are more
nance exergy required per biomass unit), mutualism, etc. likely to succeed in the long run (if options to change
For species development, this translates into a principle of paths have not been foreclosed).
maximum exergy use efficiency. On the basis of these In many systems, in particular ecosystems, specific
principles, prediction of development trends in ecosys attractors or functional orientors are often more imme
tems is possible. diately obvious than the basic orientors that cause the
The selection for better fitness in evolutionary pro emergence of these orientors in the first place. These
cesses favors systems (organisms) with better coping orientors can be viewed as appearing on a level below
ability. Aspects of the behavioral spectrum of a system the basic orientors in the hierarchical orientation system
that improve coping ability (basic orientors) can be (see Table 1). They translate the fundamental system
understood as implicit goals or attractors: existence, needs expressed in the basic orientors into concrete
security, effectiveness, freedom, adaptability, coexistence. attractor states linking system response to environmen
In the developmental stage of ecosystems, emphasis is on tal properties. In models and ecosystem analyses,
measures of ecosystem integrity can be based on corre
the basic orientors: existence, effectiveness, and freedom;
sponding ecosystem goal functions. Ecosystem attractor
in the mature stage it shifts to security, adaptability, and
states emerge as general ecosystem properties in the
coexistence (see Table 1, where orientor concepts have
coevolution of ecosystem and environment. They can
been linked to E. P. Odums classical model of ecological
be viewed as ecosystem specific responses to the need
to satisfy the basic orientors. Major ecosystem orientors
The existence of these implicit goals does not imply
are optimization of use of solar radiation, material, and
teleologic or teleonomic development toward a given
energy flow intensities (networks); matter and energy
cycling (cycling index); storage capacity (biomass accu
Table 1 Orientor concepts in the context of ecological
mulation); nutrient conservation, respiration, and
transpiration; diversity (organization); hierarchy (signal
Developmental Mature stage filtering).
stage The emergence of basic orientors in response to the
Basic orientor emphasis general properties of environments can be deduced from
Existence Coexistence general systems theory, but supporting empirical evidence
Freedom Security and related theoretical concepts can also be found in
Effectiveness Adaptability such fields as psychology, sociology, and the study of
Ecosystem orientor Orientor emphasis (goal function) artificial life.
Growth and change High Low
Life cycle Short, simple Long, complex
Biomass Low High
Energy conservation Low High Orientor Guidance in System Development,
Nutrient conservation Low High Control, Adaptation, and Evolution
Nutrient recycling Low High
Specialization Low High Environmental influences partially determine system
Diversity Low High behavior. The magnitude of their effect on behavior
Organization Low High
Symbiosis Low High
depends on the influence structure of the system.
Stability; feedback Low High Sometimes systems can be controlled by controlling
control the inputs from their environment. However, the feed
Structure Linear, simple Network, backs in the system itself are usually more important for
complex system control and adaptation of behavior to environ
Information Low High
Entropy High Low
mental conditions. Feedback means that the system state
influences itself. Behavior changing internal feedbacks
Goal Functions and Orientors 125

are possible on several hierarchical levels in complex gradual lowering of the groundwater level by growing
systems with different typical response characteristics its roots to greater depth. This constitutes a parameter
and time constants (typical response times). These pos change (root length and root surface). The fundamental
sibilities are also shown in Figure 3. system structure of a tree, in particular, the function of the
roots, has not changed in this case.
On the next higher level we find processes of self
Response time Level Response organization in response to environmental challenges.
This means structural change in the system. Processes
Immediate Process Causeeffect of this kind have a longer response time and can only be
Short Feedback Control
Medium Adaptation Parameter change
conducted by systems having the capability for self
Long Self-organization Structural change organization. Adult organisms or technical systems
Very long Evolution Change of identity rarely or never belong to this category; on the other
Always Basic orientors Maintaining integrity hand, this characteristic is often found in the develop
ment of organisms, social systems, organizations, and
The simplest type of system response is the cause A system may also change its identity in the course of
effect relationship. It occurs at once as in, for example, an evolutionary process. This means that its functional
stimulusresponse reflex. It is the only type of system characteristics, and hence its system purpose, change with
behavior which can legitimately be described by relating time. Adaptations of this kind take place as a result of
the output directly to the input. Unfortunately, it is often reproduction and evolution of living organisms. It is char
assumed that the same simple relationship is also applic acteristic of this process that the system change coincides
able to other types of system response (such as the with a possibly drastic shift in system identity (change of
following), and this erroneous assumption often leads to goal function and of system purpose). An evolutionary
fundamental mistakes. example is the development of flying animals (birds) from
On the next higher level we find responses which are water dwelling reptiles.
generated by feedback in the system, involving at least All of these system responses to challenges from the
one state variable or delay such as an empty stomach environment in essence constitute attempts to maintain
causing hunger and the search for food. Control processes system integrity (possibly over many generations and
belong to this category. The response time is short, and over a long time period) even if it means changing system
influence structure and system parameters remain identity, that is, system purpose. From this observation it
invariant. can be deduced that a system must orient its development
On the next higher level we find processes of adapta with respect to certain basic criteria (basic orientors) to
tion. In this case the system maintains its basic influence assure its long term existence and development in an
structure, but parameters are adjusted to adapt to the often hostile environment. This orientation may be impli
situation, possibly changing the response characteristics cit (forced upon the system) or explicit (actively pursued
in the process. For example, a tree may adapt to the by the system). It does not require conscious decision or

Integrity Orientors

Evolution Purpose

Self-organization Structure

Adaptation Parameters

System input Discrepancy System state System output


Stimulus response
Figure 3 System response can be caused by different processes with very different time constants: stimulusresponse, feedback
control, adaptation, self-organization, evolution, maintaining system integrity.
126 Goal Functions and Orientors

even cognitive ability, although resulting action may Random rules are created in unknown situations. The pro
appear to an observer as intelligent or even goal or cess is repeated for a large number of steps (typically 10 000).
value oriented behavior. Eventually, a set of behavioral rules develops which allows
optimal behavior under the given set of conditions.
Note that this optimal behavior has not been defined in
Simulation of the Evolution of System terms of an objective function guiding the evolution of the
Orientation set of behavioral rules. The rule set develops solely from
the reinforcement of rules which lead to food or avoid
Animats and Genetic Algorithms for Orientation
collisions. An explicit exergy balance accounts for all
Orientation theory is not just a conceptual framework for exergy losses associated with movement, collisions with
understanding system evolution and behavior under the obstacles, and rule generation, and exergy gains due to
exergy availability constraint. It also allows quantitative uptake of food. The development of the rule set is then
and comparative analysis of system performance under driven by the requirement to optimize exergy pickup in
different environmental conditions. the given environment (with specific resource availabil
Genetic algorithms are models of biological adaptive ity), while allowing for environmental variety, variability,
processes that are being widely and successfully applied and change specific for that environment. Neglect of these
to a wide spectrum of adaptation and optimization pro properties is penalized by lack of fitness, and threat to
blems. In particular, these algorithms have been used to survival, and causes disappearance of deficient rules.
simulate learning and adaptation of artificial animals (ani Other criteria besides efficiency will therefore be
mats) in simulated environments containing food and reflected in the set of behavioral rules. Since these were
obstacles. They can be used to demonstrate the emer not expressly introduced, we must recognize them as
gence of basic orientors in self organizing systems having emergent value orientations or objective functions.
to cope with complex environments. The animat experiment contains all components
The animat model incorporates essential features of a necessary for a study in the basic orientor framework.
simple animal in a diverse environment. Being an open Animat fitness depends on the ability to maintain a posi
system, an animal depends on a flow of exergy from the tive exergy balance in the long term. This exergy balance
environment. In the course of its (species) evolution, it has to is therefore at the core of the orientor satisfaction assess
learn to associate certain signals from the environment with ment. At each step, exergy uptake (by food consumption)
reward or pain and to either seek or avoid their respective and exergy losses (by collisions with obstacles, motion,
sources (exergy gain or exergy loss). This learning phase (of and learning of rules) are recorded and used to compute
populations) will eventually lead to the establishment of the momentary exergy balance. Attention to all orientors
cognitive structure and behavioral rules which are approxi is mandatory to ensure a positive exergy balance even
mately optimal in the particular environment (with respect under adverse environmental conditions.
to maximization of reward, minimization of pain, and secur Quantitative measures must be defined for character
ing survival). These behavioral rules incorporate knowledge izing the different properties of the environments used in
which enables intelligent behavior. the animat experiments. Animat performance in different
The animat is designed to simulate this process. It can environments is compared by using measures of orientor
pick up sensory signals from its environment (containing satisfaction. These have to be defined using relevant
food and obstacles), and classify them with available rules parameters of animat performance.
to determine an appropriate action (direction of move
ment). After a successful move, the strength of rules
Emergence of Basic Value Orientations,
leading up to it is increased by sharing in the reward
Anticipation, and Individual Differences
(i.e., exergy gain). New rules are occasionally generated
by either random creation, or by genetic operations Since the animats training depends on a number of ran
(crossing over and recombination). They are added to dom factors, each animat develops a different cognitive
the existing rule set, and compete with the other rules system (classifier set and decision rules), even though
for reward. Unsuccessful rules are not reinforced and lose final performance may be similar. In order to show gen
strength and influence in the rule set. eral tendencies despite these individual differences, mean
The training process consists of placing the animat at a values over large populations were obtained. These dealt
random empty location in an environment with specific with (1) results of the training process in two (otherwise
environmental properties, and allowing it to move around identical) environments having different variety and
searching for food. A collision with an obstacle causes a loss variability, and with (2) performance of animats after
of exergy and throws the animat back to its previous posi transfer from their training environment to environments
tion. Rules leading to success are rewarded. A genetic event challenging them with more variety, or variability, or
of rule generation may occur with a prescribed probability. change.
Goal Functions and Orientors 127

One remarkable result from these experiments is that system viability and survival they would not have
individuals achieve comparable performance in a given emerged unless important for the viability of the system.
training environment with very different cognitive sys Balanced attention still leaves room for individual
tems, and in particular with different orientor emphasis. differences in the relative emphasis given to the differ
While this may not provide any particular advantage in ent orientors. Individuals belonging to the populations
the training environment, it may provide distinct fitness used in the animat experiments evolve significant dif
advantages if the animat is moved to a different environ ferences in value emphasis (e.g., specialist, generalist,
ment. Three particular types of individuals stand out: cautious type). These individual variations, while not
generalists (type F) stressing freedom of action, specialists significantly reducing performance in the standard
(type E) focusing on effectiveness, and cautious type training environment, provide comparative advantage
(type S) emphasizing security. Figure 4 shows the and enhanced fitness when resource availability, vari
different orientor stars for these three types. ety, or reliability of the environment change. They also
The ability to develop a cognitive system reflecting its result in distinctly different behavioral styles. However,
environment makes the animat a suitable vehicle for inves pathological behavior will follow if orientor attention
tigating goal function emergence and value orientation. becomes unbalanced (e.g., dominant emphasis on one
Genetic algorithms are very effective processes that seem orientor).
to capture the essentials of real processes found in the Training of animats in different environments, the per
evolution of organisms and ecosystems. In the animat, formance of animat individuals in environments that differ
they very effectively build up a cognitive model (or goal from their training environments, and the simulation of
function) that enables anticipatory behavior; since rewards adaptive learning in a changing environment, lead to
flow back to earlier rules leading to later pay off, the some general conclusions that are in full agreement with
activation of the initial rules in a pay off chain means everyday observations and general systems knowledge:
that the system suspects possible pay off and anticipates
the near future, that is, it has an internal model of the Generalists have a better survival chance than others if
moved to an environment of greater variety.
results of its actions under the given circumstances.
In the animat experiments (and similarly, in real life), Cautious types have a better survival chance than
others if moved to a less reliable environment.
implicit and (more or less) balanced multidimensional
attention to the basic orientors emerges from the simple Training in more unreliable and/or more diverse
environments increases satisfaction of the security
one dimensional mechanism of rewarding success in the
and/or freedom of action orientors at the cost of the
given environment. Thus, in the course of its evolutionary
effectiveness orientor.
development in interaction with its environment, the
system evolves a complex multidimensional behavioral Training in an uncertain environment teaches caution
and improves fitness in a different environment.
objective function from the very unspecific requirement
of fitness. Conversely, this also means that balanced atten Learning caution (better satisfaction of the security
orientor) takes time and decreases effectiveness, but
tion to the emergent basic orientors is necessary for increases overall fitness.

Investment in learning (exergy cost of learning in the
animat) pays off in better fitness; the learning invest
ment is (usually) much smaller than the pay off gain.
2 Animat individuals not only develop behavior that can be
interpreted as intelligent, they also develop a complex
1 goal function (balanced attention to basic orientors), or
value orientation. Serious attention to basic values (basic
Security 0 Freedom orientors: existence, effectiveness, freedom, security,
adaptability, coexistence) is therefore an objective
requirement emerging in, and characterizing self organi
zing systems. These basic values are not subjective human
inventions; they are objective consequences of the process
of self organization in response to normal environmental

Generalist (F) Specialist (E) Cautious (S)

Figure 4 In an identical training environment, different lifestyles See also: Ecological Network Analysis, Ascendency;
may evolve. Generalists stress freedom of action, specialists Ecological Network Analysis, Environ Analysis; Exergy;
focus on effectiveness, while cautious types emphasize security. Fundamental Laws in Ecology.
128 Exergy

Further Reading Jrgensen SE (2001) A tentative fourth law of thermodynamics.

In: Jorgensen SE (ed.) Thermodynamics and Ecological Modelling,
Ashby WR (1962) Principles of the self organizing system. In: von pp. 305 347. Boca Raton, FL: Lewis.
Foerster H and Zopf GW (eds.) Principles of Self Organization, Krebs F and Bossel H (1997) Emergent value orientation in self
pp. 255 278. New York: Pergamon. organization of an animat. Ecological Modelling 96: 143 164.
Bossel H (1977) Orientors of nonroutine behavior. In: Bossel H (ed.) Mayr E (1974) Teleological and teleonomic: A new analysis. Boston
Concepts and Tools of Computer Assisted Policy Analysis, Studies in the Philosophy of Science 14: 91 117.
pp. 227 265. Basel: Birkhauser. Mayr E (2001) What Evolution Is. New York: Basic Books.
Bossel H (1999) Indicators for Sustainable Development: Theory, Miller JG (1978) Living Systems. New York: McGraw Hill.
Method, Applications. Winnipeg: IISD International Institute for Odum EP (1969) The strategy of ecosystem development. Science
Sustainable Development. 164: 262 270.
Bossel H (2001) Exergy and the emergence of multidimensional system Muller F and Leupelt M (eds.) (1998) Eco Targets, Goal Functions, and
orientation. In: Jrgensen SE (ed.) Thermodynamics and Ecological Orientors. Berlin/Heidelberg/New York: Springer.
Modelling, pp. 193 209. Boca Raton, FL: Lewis. Wilson SW (1985) Knowledge growth in an artificial animal.
Fath BD, Patten BC, and Choi JS (2001) Complementarity of ecological In: Grefenstette JJ (ed.) Proceedings of the First International
goal functions. Journal of Theoretical Biology 208(4): 493 506. Conference on Genetic Algorithms and Their Applications,
Holland JH (1992) Adaptation in Natural and Artificial Systems. pp. 16 23. Pittsburgh PA and San Mateo: Lawrence Earlbaum and:
Cambridge, MA: MIT Press. Morgan Kaufmann.
Jantsch E (1980) Self Organizing Universe: Scientific and Human
Implications of the Emerging Paradigm of Evolution. New York:

S E Jrgensen, Copenhagen University, Copenhagen, Denmark
2008 Elsevier B.V. All rights reserved.

Definition: Exergy Losses and Gains of Eco-Exergy by Human Activities

Definition: Eco-Exergy Included Pollution
Exergy and Information Formulation of a Thermodynamic Hypothesis for
Exergy and the Dissipative Structure Ecosystems
How to Calculate Eco-Exergy of Organic Matter and Support to the Maximum Eco-Exergy Hypothesis
Organisms? Further Reading
Why Do Living Systems Have Such a High Level
of Eco-Exergy?

Definition: Exergy
S, U, V, N1, N2, N3,...
Energy is defined as the amount of work ( entropy free T, p, c , c , c ,...
1 2 3

energy) a system can perform when it is brought into

thermodynamic equilibrium with its environment Toward thermodynamic
equilibrium with the
(Figure 1). The considered system is characterized by the environment
extensive state variables S, U, V, N1, N2, N3, . . . , where S is
the entropy, U is the energy, V is the volume, and N1, N2, N3, S, Uo, V, N1, N2, N3,...
. . . are moles of various chemical compounds, and by the
To, po, oc , oc , oc ,...
intensive state variables, T, p, c1, c2, c3, . . . , where T is the 1 2 3

temperature, p the pressure, and  symbolizes the chemical

potential of the components 1,2,3,. . .. The system is Figure 1 Definition of exergy is shown. The work is symbolized
by the gain in potential energy of the weight.
coupled to a reservoir or reference state by a shaft, together
forming a closed system. The reservoir (the environment) is
characterized by the intensive state variables To, po, oc1,
oc2, oc3, . . . , and as the system is small compared with the reservoir and is simultaneously able to release entropy free
reservoir the intensive state variables of the reservoir will energy to the reservoir. During this process the volume of
not be changed by interactions between the system and the the system is constant as the entropy free energy (i.e., work
reservoir. The system develops toward equilibrium with the energy) must be transferred through the shaft only.
Exergy 129

According to the definition of exergy, Ex, we have The efficiency of concern is the ratio of useful energy
(work) to total energy which always is less than 100% for
Ex U U Uo 1 real processes, which always are irreversible. This effi
As ciency expresses that a part of the energy cannot be
X utilized as work and that all processes are irreversible
U TS pV c Ni 2 because exergy is lost by all energy transfer processes as
heat to the environment.
(when we only consider the three energy forms: Exergy efficiency, defined as work performed divided
heat, spatial energy (displacement work), and chemical by the total exergy available, is also of interest, particu
energy; see any textbook in thermodynamics), and larly in technology. It expresses how much of the work
correspondingly capacity we are able to utilize.
All transfers of energy imply that exergy is lost because
Uo To S po V co Ni energy is transformed to heat at the temperature of the
c environment. It has therefore been of interest to set up for
all environmental systems an exergy balance in addition
we get the following expression for exergy, when in this
to an energy balance. Our concern is exergy loss because
case kinetic energy, potential energy, electrical energy,
it means that first class energy which can do work is
radiation energy, and magnetic energy are excluded we
converted to second class energy (heat at the tempera
ture of the environment) which cannot do work. So, the
X particular properties of heat and temperature are a meas
Ex S T To V p po c co Ni 4 ure of the movement of molecules, given limitations in
our possibilities to utilize energy to do work. Due to these
limitations, we have to distinguish between exergy which
The total transfer of entropy free energy in this case is
can do work and anergy which cannot do work, and all
the exergy of the system. It is seen from this definition
real processes imply inevitably a loss of exergy as anergy
that exergy is dependent on the state of the total system
(see also the next section).
( system reservoir) and not dependent entirely on the
Exergy or rather the loss of exergy as heat, which
state of the system. Exergy is therefore not a state variable.
means production of entropy, seems more useful to
This definition of exergy is used in engineering to
apply than entropy to describe the irreversibility of real
express the efficiency of power plants. The energy effi
processes. It has the same unit as energy and is an energy
ciency of power plants is of course 100%, according to the
form, while the definition of entropy is more difficult to
first law of thermodynamics, while the interesting effi
relate to concepts associated to our usual description of
ciency is the exergy efficiency: how much of the chemical
reality. In addition entropy is not clearly defined for
energy (exergy) in the applied fossil fuel if fossil fuel is the
far from thermodynamic equilibrium systems,
energy source is converted to useful work (exergy)? What particularly for living systems. Moreover, it should be
is not converted to exergy in form of electricity is lost as mentioned that the self organizing abilities of systems
heat to the environment at the temperature of the envir depend strongly on the temperature. Exergy takes the
onment it contains therefore no work potential. temperature into consideration as the definition shows,
Notice that the exergy of the system is dependent on while entropy does not. It implies that exergy at 0 K is 0
the intensive state variables of the reservoir. Notice that and at minimum. Negative entropy is not expressing
exergy is not conserved only if entropy free energy is the ability of the system to do work (we may call it
transferred, which implies that the transfer is reversible. the creativity of the system as creativity requires
All processes in reality are, however, irreversible, which work), but exergy becomes a good measure of the
means that exergy is lost (and entropy is produced). Loss creativity, which is increasing proportional with the
of exergy and production of entropy are two different temperature. Furthermore, exergy facilitates the differen
descriptions of the same reality, namely, that all processes tiation between low entropy energy and high entropy
are irreversible, and we unfortunately always have some energy, as exergy is entropy free energy.
loss of energy forms which can do work to energy forms Information contains exergy. Boltzmann showed that
which cannot do work (heat at the temperature of the the free energy of information (it means exergy) that we
environment). So, the formulation of the second law of actually possess (in contrast to the information we need to
thermodynamic by use of exergy is all real processes are describe the system) is kT ln I, where I is the information
irreversible which implies that exergy inevitably is lost. we have about the state of the system, for instance,
Exergy is not conserved, while energy of course is con that the configuration is 1 out of W possible ones and k
served by all processes according to the first law of is Boltzmanns constant 1.3803  10 23 J/(molecules
thermodynamics. deg). It implies that one bit of information has the exergy
130 Exergy

equal to kT ln2. Transformation of information from one

system to another is often almost an entropy free energy
System at T, p
transfer. If the two systems have different temperatures,
then the entropy lost by one system is not equal to the
entropy gained by the other system, while the exergy lost
by the first system is equal to the exergy transferred and
equal to the exergy gained by the other system, provided work, not useful Work (exergy)
that the transformation is not accompanied by any loss of
exergy. Also, in this case, it is obviously more convenient
to apply exergy than entropy.

Reference environment
at T, p
Definition: Eco-Exergy
Figure 2 The exergy content of the system is calculated in the
text for the system relatively to a reference environment of the
In ecology, technological exergy is not so useful same system at the same temperature and pressure, as an
because the reference state, the environment, would inorganic soup with no life, biological structure, information, or
be the adjacent ecosystem and we would like to find organic molecules.
an expression that can measure how developed an
ecosystem is, that is, how far it is from thermodynamic
the system, there seem to be no reason to assume a
equilibrium. For a reservoir or reference state, it is
(minor) temperature and pressure difference between
therefore advantageous in ecology to select the same
the system and the reference environment. Under these
system but at thermodynamic equilibrium, that is, that
circumstances we can calculate the exergy content of the
all components are inorganic and at the highest oxida
system as coming entirely from the chemical energy:
tion state, if sufficient oxygen is present (nitrogen as
nitrate, sulfur as sulfate, etc.). The reference state will Ex c co Ni 5
in this case correspond to the ecosystem without life c
forms and with all chemical energy utilized or as an This represent the nonflow chemical eco exergy. The
inorganic soup. Usually, it implies that we also con difference in chemical potential (c co) between the eco
sider T To, and p po, which means that the exergy system and the same system at thermodynamic equilibrium
becomes equal to the difference of Gibbs free energy determines the eco exergy. This difference is determined by
of the system and the same system at thermodynamic the concentrations of the considered components in the
equilibrium, or the chemical energy content included system and in the reference state (thermodynamic equili
the thermodynamic information (see below) of the brium), as it is the case for all chemical processes.
system. Gibbs free energy is defined according to We can measure the concentrations in the eco
the following equation: system, but the concentrations in the reference state
dG dE p dV S dT (thermodynamic equilibrium) can only be based on the
usual use of chemical equilibrium constants. If we have
where dV is the change in volume and dS is the change the process
in entropy. T and p are the temperature and pressure,
respectively. The exergy becomes by this definition Component A $ inorganic decomposition products
clearly a measure of how far the ecosystem is from thermo
dynamic equilibrium, that is, how much (complex) It has a chemical equilibrium constant, K:
organization the ecosystem has build up in the form of K inorganic decomposition products=Component A 6
organisms, complex biochemical compounds, and com
plex ecological network. Here, we use the available work, The concentration of component A at thermodynamic
that is, the exergy, as a measure of the distance from equilibrium is difficult to find, but we can find the
thermodynamic equilibrium. concentration of component A at thermodynamic equili
This description of exergy development in an ecosys brium from the probability of forming A from the
tem makes it pertinent to assess the exergy of ecosystems. inorganic components.
It is not possible to measure exergy directly but it is We find by these calculations the exergy of the system
possible to compute it by eqn [4]. Figure 2 illustrates the compared with the same system at the same temperature
definition of eco exergy. As the chemical energy embo and pressure but in form of an inorganic soup without any
died in the organic components and the biological life, biological structure, information, or organic mol
structure contributes far most to the exergy content of ecules. As (c co) can be found from the definition of
Exergy 131

the chemical potential replacing activities by concentra instance, with a biochemical function that is rare else
tions, we get the following expressions for eco exergy: where, carries exergy and information. On more complex
levels, information may still be strongly related to exergy
i n but in more indirect ways. Information is also a conveni
Ex RT ci ln ci =ci;o 7
ent measure of physical structure. A certain structure is
i 0
chosen out of all possible structures and defined within
where R is the gas constant (8.314 J K 1 mol 1 certain tolerance margins.
0.08207 l atm. K 1 mol 1), T is the temperature of the It is possible to distinguish between the exergy of
environment (and the system; see Figure 2), while ci is information and the exergy of biomass. pi defined as
the concentration of the ith component expressed in a ci/A, where
suitable unit, for example, for phytoplankton in a lake X
ci could be expressed as mg l 1 or as mg l 1 of a focal A ci 8
nutrient. ci,o is the concentration of the ith component at i 1

thermodynamic equilibrium and n is the number of com is the total amount of matter in the system, is introduced
ponents. ci,o is of course a very small concentration (except as new variable in eqn [7]:
for i 0, which is considered to cover the inorganic com
pounds), corresponding to a very low probability of X
Ex A RT pi ln pi =pio A ln A=Ao 9
forming complex organic compounds spontaneously in an i 1
inorganic soup at thermodynamic equilibrium. ci,o is even
lower for the various organisms because the probability of As A  Ao, exergy becomes a product of the total biomass
forming the organisms is very low with their embodied A (multiplied by RT) and Kullbacks measure:
information, here represented by the genetic code. X

By using this particular exergy based on the same K pi ln pi =pio 10

i 1
system at the thermodynamic, chemical equilibrium as
reference, the eco exergy depends only on the chemical where pi and pio are probability distributions, a posteriori
potential of the numerous biochemical components that and a priori to an observation of the molecular detail of the
are characteristic for life. It is consistent with Boltzmanns system. It means that K expresses the amount of informa
statement that life is a struggle for free energy. Eco tion that is gained as a result of the observations.
exergy has a definition close to the free energy, but unlike If we observe a system, which consists of two connected
free energy, eco exergy is not a state variable. It will chambers (see Figure 3), then we expect the molecules to
depend on the reference state that will vary from be equally distributed in the two chambers, that is, p1 p2
ecosystem to ecosystem. Furthermore, it is difficult to is equal to 1/2. If, on the other hand, we observe that all
apply the classic state variables in thermodynamics far the molecules are in one chamber, then we get p1 1 and
from thermodynamic chemical equilibrium. Classic p2 0. Let us presume that the chamber to the left con
thermodynamics presumes that the system is close to tains 1 mol of a pure ideal gas, while the chamber to the
equilibrium, which makes it possible to show that for right is empty. If we open the valve between the two
instance free energy is a state variable that gives the chambers, the loss of eco exergy (and also of technological
same result independent on the pathway. We want to exergy) for the system would be RT ln2 in accordance
use eco exergy far from thermodynamic equilibrium with eqns [7], [9], and [10]. We could utilize at least a part
and can therefore not use free energy in this context. of the exergy by installation of a small propeller in the
As we know that ecosystems due to the throughflow valve. The system will by this process increase its entropy
of energy have the tendency to move away from R ln2. This is in accordance with eqn [8].
thermodynamic equilibrium losing entropy or gaining
exergy and information, we can at this stage formulate

... .. .
the following proposition of relevance for ecosystems:
ecosystems attempt to develop toward a higher level of

... . . .. . .

. ...
Exergy and Information
Figure 3 The left chamber contains 1 mole of a pure ideal gas,
while the right chamber is empty. If we open the valve, the
Information means acquired knowledge. The thermody system will loose eco-exergy (or technological exergy) RT ln2,
namic concept of exergy is closely related to information. which we could utilize by installation of a propeller in the valve.
A high local concentration of a chemical compound, for The entropy of the system will simultaneously increase by R ln2.
132 Exergy

Exergy and the Dissipative Structure the number of components. cio is very low for living
component because the probability that living compo
As an ecosystem is nonisolated, the entropy changes dur nents are formed at thermodynamic equilibrium is very
ing a time interval, dt, can be decomposed into the low. It implies that living components get a high eco
entropy flux due to exchanges with the environment, exergy. cio is not zero for organisms, but will correspond to
and the entropy production due to the irreversible pro a very low probability of forming complex organic com
cesses inside the system such as diffusion, heat pounds spontaneously in an inorganic soup at
conduction, and chemical reactions. It can also be thermodynamic equilibrium. cio on the other hand is
expressed by use of exergy: high for inorganic components, and although cio still is
low for detritus, it is much higher than for living
Ex=dt de Ex=dt di Ex=dt 11 component.
where d Ex/dt represents the exergy input to the system The exergy of structurally complicated material can
and di Ex/dt is the exergy consumed (is negative) by the be estimated based on the elementary composition. This
system for maintenance, etc. e is used to indicate an has, however, the disadvantage that a higher organism and
external source and i to indicate the internal exergy a microorganism with the same elementary composition
change. will get the same exergy which is in complete disagree
Equation [11] shows among other things that sys ment with the lower probability to form a more complex
tems can only maintain a nonequilibrium steady state organism, that is, the lower concentration of cio in the
by compensating the internal exergy consumption equation. The composition will not account for the con
with a positive exergy influx (de Ex/dt > 0). Such an tribution of Kullbachs measure of information, which is
influx induces order into the system. In ecosystems, often the major part of the eco exergy, as it is shown
the ultimate exergy influx comes from solar radiation, below.
and the order induced is, for example, biochemical The problem related to the assessment of cio has been
molecular order. If de Ex > di Ex (the exergy con discussed and a possible solution proposed. For dead
sumption in the system), the system has surplus organic matter, detritus, which is given the index 1, it
exergy input, which may be utilized to construct can be found from classical thermodynamics.
further order in the system, or as Prigogine calls it, For the biological components, 2,3,4, . . . ,N, the prob
dissipative structure. The system will thereby move ability, pio, consists at least of the probability of producing
further away from thermodynamic equilibrium. the organic matter (detritus), that is, p1o, and the prob
Evolution shows that this situation has been valid for ability, pi,a, to find the correct composition of the enzymes
the ecosphere on a long term basis. In spring and determining the biochemical processes in the organisms.
summer, ecosystems are in the typical situation that Living organisms use 20 different amino acids and each
de Ex exceeds di Ex. If de Ex <di Ex, the system gene determines in average the sequence of about 700
cannot maintain the order already achieved, but will amino acids. pi,a, can be found from the number of permu
move closer to the thermodynamic equilibrium, that tations among which the characteristic amino acid
is, it will lose order. This may be the situation for sequence for the considered organism has been selected.
ecosystems during fall and winter or due to environ It means that
mental disturbances.
pi;a a Ngi 13

How to Calculate Eco-Exergy of Organic where a is the number of possible amino acids 20, N is
Matter and Organisms? the number of amino acids determined by one gene
700, and gi is the number of non nonsense genes. The
The following expression for what we could call the following two equations are available:
ecological exergy per unit of volume has been presented;
see eqn [7]: pio p1o pi;a p1o a Ng  p1o :20 700g 14

The exergy contribution of the ith component can be

i n  found by combining eqns [12] and [14]:
Ex RT ci lnci =cio ML 1 T 2 12
i 0
Ex RT ci ln ci =p1o a Ngc0o 1 1o ci ci ln pi;a
1 1o ci ci ln a Ng  i
where R is the gas constant, T is the temperature of the 18:7ci 700ln20 ci gi ML 1 T 2 15
environment, ci is the concentration of the ith component
expressed in a suitable unit, and cio, is the concentration of The total eco exergy can be found by summing up the
the ith component at thermodynamic equilibrium and n is contributions originated from all components. The
Exergy 133

contribution by inorganic matter can be neglected as the knowledge of the genome size and the complexity of
contributions by detritus and even to a higher extent from different organisms. A  value of 2.0 means that the
the biological components are much higher due to an extre eco exergy embodied in the organic matter and the infor
mely low concentration of these components in the mation are equal. As the  values in Table 1 are much
reference system (thermodynamic equilibrium for the sys bigger than 2.0 (except for virus, where the  value is
tem). The contribution by detritus, dead organic matter, is 1.01) the information eco exergy is the most significant
18.7 kJ g 1 times the concentration (in gram per unit of part of the eco exergy of organisms.
volume) corresponding to the composition of detritus, The eco exergy due to the fuel value of organic
namely lipids, carbohydrates, and proteins mainly, while matter (chemical energy) is about 18.7 kJ g 1 (compare
the eco exergy of living organisms with approximations with coal: about 30 kJ g 1 and crude oil: 42 kJ g 1). It can
consists of be transferred to other energy forms for instance mechan
ical work directly, and be measured by bomb calorimetry,
Ex1chem 18:7 kJ g 1 times the concentration ci which requires destruction of the sample (organism),
gram per unit of volume however. The information eco exergy (  1) c is
taken care of by the control and function of the many
and biochemical processes. The ability of the living system to
Exibio RT 700 ln20 ci gi RT 2100 gi ci 16 do work is contingent upon its functioning as a living
dissipative system. Without information eco exergy, the
R 8.314 J mole 1 and if we presume a molecular weight organic matter could only be used as fuel similar to fossil
of an average 105 for the enzymes, we obtain the follow fuel. But due to the information eco exergy, organisms
ing equation for Exibio at 300 K: are able to make a network of the sophisticated biochem
ical processes that characterize life. The eco exergy (of
Exibio 0:0529 gi ci 17 which the major part is embodied in the information) is a
measure of the organization. This is the intimate relation
where the concentration now is expressed in g per unit ship between energy and organization that Schrdinger
of volume and the exergy in kilojoules per unit of was struggling to find.
volume. As calculated here, eco exergy is a result of evolution
For the entire system the eco exergy, Ex total and of what Elsasser calls re creativity to emphasize that
exergy chemical exergy biological can be found as the information is copied and copied again and again in a
long chain of copies where only minor changes are intro
N N 
Ex total 18:7 ci 0:0529 ci gi ML 1 T 2 18 duced for each new copy. The energy required for the
i 1 i 1 copying process is very small, but it has of course required
a lot of energy to come to the mother copy through the
where g for detritus (i 1) of course is 0. Table 1 shows
evolution for instance from prokaryotes to human cells.
the weighting factor, , which is introduced to be able to
Kullbacks measure of information covers the gain in
cover the exergy for various organisms in the unit detritus
information when the distribution is changed from pion to
equivalent or chemical exergy equivalent:
pI. Note that K is a specific measure (per unit of matter).
Expressed by the Kullbachs measure of information, we
Ex total i ci as detritus equivalent 19 get the following equation for eco exergy:
i 1
Ex organism cRTK 20
The calculation of eco exergy accounts for the chemi  is therefore RTK.
cal energy in the organic matter as well as for the The total eco exergy of an ecosystem cannot be cal
(minimum) genetic information embodied in the living culated exactly, as we cannot measure the concentrations
organisms. The latter contribution is measured by the of all the components or determine all possible contribu
extremely small probability to form the living compo tions to exergy in an ecosystem. If we calculate the exergy
nents, for instance algae, zooplankton, fish, mammals, of a fox for instance, then the above shown calculations
etc., spontaneously from inorganic matter. Weighting fac will only give the contributions coming from the biomass
tors defined as the exergy content relatively to detritus and the information embodied in the genes, but what is
(see Table 1) may be considered quality factors reflecting the contribution from blood pressure, sexual hormones,
how developed the various groups of organisms are and to network interactions, etc.? These properties are at least
what extent they contribute to the exergy due to their partially covered by the genes but is that the entire
content of information which is reflected in the computa story? We can calculate the contributions from the domi
tion. The  values in Table 1 are found on basis of latest nant components, for instance by the use of a model or
134 Exergy

Table 1 Eco-exergy of living organisms

Early organisms Plants Animals

Detritus 1.00
Virus 1.01
Minimal cell 5.8
Bacteria 8.5
Archaea 13.8
Protists (Algae) 20
Yeast 17.8
33 Mesozoa, Placozoa
39 Protozoa, amoebe
43 Phasmida (stick insects)
Fungi, molds 61
76 Nemertina
91 Cnidaria (corals,sea anemones, jelly fish)
Rhodophyta 92
97 Gastroticha
Prolifera,sponges 98
109 Brachiopoda
120 Plathyhalminthes (flatworms)
133 Nematoda (round worms)
133 Annelida (leeches)
143 Gnathostomulida
Mustard weed 143
165 Kinorhyncha
Seedless vascular plants 158
163 Rotifera (wheel animals)
164 Entoprocta
Moss 174
167 Insecta (beetles, flies, bees, wasps, bugs, ants)
191 Coleodiea (Sea squirt)
221 Lepidoptera (butterflies)
232 Crustaceans
246 Chordata
Rice 275
Gymnosperms (incl. Pinus) 314
310 Molluska, bivalvia, gastropodea
322 Mosquito
Flowering plants 393
499 Fish
688 Amphibia
833 Reptilia
980 Aves (birds)
2127 Mammalia
2138 Monkeys
2145 Anthropoid apes
2173 Homo sapiens

 values exergy content relatively to the exergy of detritus (Jrgensen et al.).

measurements that cover the most essential components 2. We do not know the genes in all details for all
for a focal problem. The difference in eco exergy by organisms.
comparing two different possible structures (species 3. We calculate only in principle the eco exergy embo
composition) is decisive here. Moreover, eco exergy died in the proteins (enzymes), while there are other
computations give always only relative values, as the components of importance for the life processes. These
eco exergy is calculated relatively to the reference components are contributing less to the exergy than the
system. enzymes and the information embodied in the enzymes
Eco exergy calculated using the above equations has control the formation of these other components, for
some clear shortcomings: instance hormones. It can however not be excluded that
these components will contribute to the total exergy of the
1. We have made some although minor approxima system. The life processes are of course considered indir
tions in the equations presented above. ectly as the enzymes determine the life processes.
Exergy 135

4. We do not include the eco exergy of the ecological Why Do Living Systems Have Such a High
network. If we calculate the exergy of models, the net Level of Eco-Exergy?
work will always be relatively simple and the contribution
coming from the information content of the network is A frog of 20 g will have an eco exergy content of
considerably less than the exergy contribution from the 20  18.7  688 kJ  257 MJ, while a dead frog will have
organisms. The real ecological network may contribute only an exergy content of 374 kJ, although they have the
much more to the total exergy. When network models are same chemical composition, at least a few seconds after
compared it may also be relevant to compare exergy of the frog has died. The difference is rooted in the informa
different networks. tion or rather the difference in the useful information.
5. We will always use a simplification of the ecosys The dead frog has the information a few seconds after its
tem, for instance by a model or a diagram or similar. This death (the amino acid composition of the proteins has not
implies that we only calculate the exergy contributions of yet been decomposed), but the difference between a live
the components included in the simplified image of the frog and a dead one is the ability to utilize the enormous
ecosystem. The real ecosystem will inevitably contain information stored in the genes and the proteomes of the
more components which are not included in our frog.
calculations. The amount of information stored in a frog is really
surprisingly high. The number of amino acids placed in
It is therefore proposed to consider the eco exergy found the right sequence is about 200 000 000 and for each of
by these calculations as a relative minimum eco exergy these 200 000 000 amino acids there are 20 possibilities.
index to indicate that there are other contributions to the This information is again repeated in billions of cells that
total exergy of an ecosystem, although they may be of are cooperating to make up the frog. This enormous
minor importance. In most cases, however, a relative amount of information is able to allow reproduction and
index is sufficient to understand the reactions of ecosys is transferred from generation to generation which
tems because the absolute exergy content is irrelevant for implies that the evolution can continue because what is
the reactions. In most cases, the change in eco exergy is of already a favorable combination of properties is con
importance to understand ecological responses. served through the genes.
The weighting factors presented in Table 1 have been The information in living organisms applies nano
applied successfully to calculate eco exergy applied as an technology in the sense that the weight of 200 000 000
indicator to assess ecosystem health and in several struc amino acids is for an average amino acid molecular weight
turally dynamic models to express the model goal of 125 g moles 1 2.5  1010 g/A 4  10 14 g, where A is
function, and furthermore in many illustrations of the Avogadros number (A 6.2  1023). A book with the
maximum eco exergy principle, that is presented below. same amount of information would weigh several hun
Structural dynamic models are able to take a shift of dreds of kilograms.
species composition into account: which combinations of Because of the very high number of amino acids, about
properties are able to offer most survival? Further infor 200 000 000, it is not surprising that there will always be a
mation about structural dynamic models is given in minor difference from frog to frog in the amino acid
Structural Dynamic Models. The relatively good results sequence. It may be a result of mutations or of a minor
in applying the weighting factors in this context, in spite mistake in the copying process. These variations are
of the uncertainty of their assessment, seems only to be important because they give possibilities to test which
explicable by the robustness of the application of the amino acid sequence gives the best result with respect to
factors in modeling and other quantifications. The differ survival and growth. The best representing the most
ences between the factors of the microorganism, the favorable combination of properties will offer the high
vertebrates, and invertebrates are so clear that it does est probability of survival and give the most growth and
not matter if the uncertainty of the factors is very high the corresponding genes will therefore prevail. Survival
the results are influenced slightly. and growth mean more exergy, resulting in a bigger
On the other hand, from a theoretical point of view it distance from thermodynamic equilibrium. Exergy could
would be an important progress to get better weighting therefore be used as a thermodynamic function which
factors but also because it would enable us to model the could be used to quantify Darwins theory. In this context,
competition between species which are closely related. it is interesting that it has been demonstrated that eco
The key to find better  values maybe the proteomes exergy also represents the amount of energy needed to
(the total compositions of the proteins that as enzymes tear down the system. It means that the more exergy the
determine the life processes). Our knowledge about the system possesses the more difficult it becomes to degrade
composition of proteomes in various organisms is, how the system and the higher is therefore the probability of
ever, more limited than for the number of the genes. survival. Consequently, eco exergy can be applied as a
136 Exergy

measure of sustainability. The crucial question is there also ash, but let us not consider it in our calculations), the
fore: do we hand over the Earth to our children and exergy loss due to the dispersion can be determined by
grandchildren with the same distance from the thermo the following calculations:
dynamic equilibrium, that is, the same exergy, as we
received it from our ancestors? 0:018:314  0:300=32 ln 0:01=50  10 9
0:998:314  0:300=12 ln 0:99=4  10 4
1617 J  1:6 kJ

Losses and Gains of Eco-Exergy by where 50  10 9 and 4  10 4 represents concentrations

Human Activities Included Pollution (expressed as ratios, i.e., no units) of sulfur dioxide S and
carbon dioxide C in a typical town atmosphere. The
When contaminants, for instance heavy metals, are chemical exergy content of 1 g coal is about 32 kJ. The
widely dispersed, eco exergy is lost. When leaded gaso loss of exergy by dispersion is therefore only 5% of the
line was used to obtain a higher octane number, on the loss directly of chemical exergy by burning coal. As all
order of 400 000 t of lead were dispersed annually around our calculations will have a higher uncertainty than 5%,
the globe. Lead was even found in the ice pack of and the quality of coal may vary more than 5%, it seems
Greenland! A typical concentration in lead ore is about acceptable not to include the dispersion exergy loss by
5% or 0.05 kg kg 1ore, while a typical concentration in use of fossil fuel or as alternative to multiply all exergy
the environment after the dispersion is 1 mg kg 1 soil. If losses due to our consumption of fossil fuel by a factor
we presume 300 K, the annual eco exergy lost can be of 1.05 to compensate approximately for the exergy
found by eqn [7] as loss due to the dispersion of the formed gases in the
Ex lost 8:314  0:300  4  1011 =207 The deterioration of ecosystems. By the use of eqn [20] it is
ln 0:05=10 9  85 000 GJ yr 1 possible to find the eco exergy of an ecosystem or rather
of the ecosystem corresponding to our model of the
where 207 is the atomic weight of lead. The consumption
of lead has decreased due to shifts to other additives in the ecosystem. Consequently, the loss of eco exergy due to
gasoline in most countries. This loss of eco exergy by the deterioration and pollution of ecosystems can be found by
use of lead as additive to gasoline is therefore today only calculation of the eco exergy before and after deteriora
estimated to be around 40 000 GJ yr 1. tion. The difference will directly yield the loss.
The loss of exergy due to dispersion of resources in The use of renewable resources. The formation of renew
general can be calculated parallel to the application of able resources are found separately by multiplication of
eqn [7] as shown in eqn [21]. In addition to lead (only the the annual consumption of the various resources by the
dispersion of lead as gasoline additive is considered in this exergy content of each renewable resource. If, for
context), the loss of exergy by dispersion of other non instance, the annual fishery in the North Sea has the last
renewable resources is shown in Table 2. many years been in the order of 100 000 t, which implies
The loss of eco exergy due to the consumption of that the eco exergy of the North Sea has been
fossil fuel is found by addition of the chemical free reduced 1011  499  18.7 kJ 9.3  1017 J, then 499 is
energy (the work capacity) of the fossil fuel and the loss the  value for fish (Table 1). Sustainability requires
of eco exergy due to the dispersion of the gases resulting that the growth of fish biomass compensate for this loss
from the chemical processes. The exergy loss due to of eco exergy. It has, unfortunately, for a couple of dec
dispersion of the components of fossil fuel is found by ades in many marine ecosystems, including the North
the following calculations: if we consider 1 g of coal that Sea, not been the case due to over fishing.
contains 1% of sulfur and 99% of carbon (coal contains Dispersion of waste. This to a certain extent can be
calculated parallel to eqn [21]. This is often named the
external costs of our activities including the industrial
Table 2 Loss of eco-exergy due to dispersion of nonrenewable
and agricultural activities, but it is actually as the other
mineral resources
four points just a question about the loss of eco exergy.
Element GJ yr It is not surprising that the cost of treating waste is
Chromium 32 000
increasing as the environmental agencies require a
Nickel 15 000 more and more complete elimination of these exergy
Zinc 80 000 losses. Or expressed differently: we are coming closer
Copper 18 000 and closer to the carrying capacity of the Earth for
Mercury (included fossil fuel) 27 000 man made production. In this context it should not be
Lead (today) 40 000
forgotten that also the treatment of waste costs eco
Calculations based upon principles shown in eqn [21]. exergy.
Exergy 137

Consumption of nonrenewable fuel, including fossil fuel and development of ecosystems and the reactions of ecosys
nuclear fuel. The annual loss of eco exergy is found by tems to perturbations: If a system receives an input of
multiplication of the exergy content and the annual exergy, it will utilize this exergy after the maintenance of
consumption. the system far from thermodynamic has been covered to
We calculated above the loss of eco exergy by disper move the system further from thermodynamic equili
sion of waste due to consumption of nonrenewable brium If more than one pathway to depart from
resources. This is, however, not the entire story, as energy equilibrium is offered, the one yielding the most gradi
is required in producing various materials from ore and ents, and most exergy storage (dEx/dt is maximum) under
raw source materials. The energy requirement when the the prevailing conditions, to achieve the most ordered
material is produced from scrape is included. As it can be structure furthest from equilibrium, will tend to be
seen, the energy requirement is less when scrap if used for selected.
the production. Reuse and recycling gives therefore dou Just as it is not possible to prove the first three laws of
ble benefits: we save to draw on the limited resources and thermodynamics by deductive methods, so can the above
we save energy. Notice, particularly for aluminum, that hypothesis only be proven inductively. In the next sec
the energy requirement by the use of scrap instead of ore tion we do examine a number of concrete cases which
is considerable. As energy consumption explains one of contribute in a general way to the weight of evidence in
our major losses of eco exergy, the latter benefit is of favor.
great importance. This tentative law may be considered a translation
of Darwins theory into thermodynamics. Exergy mea
sures survival: the biomass and the network,
Formulation of a Thermodynamic information, and organization that imply that the
Hypothesis for Ecosystems resources are used in the best possible way to gain
most survival. The question is which of the possible
If an (open, nonequilibrium) ecosystem receives a bound combinations of properties by the entire spectrum of
ary flow of energy from its environment, it will use what it organisms in an ecosystem will be able to store most
can of this energy, the free energy or the exergy content, exergy (obtain most survival)? The organisms with the
to do work. The work will generate internal flows, leading properties that make it possible to gain most survival
to storage and cycling of matter, energy, and information, (exergy) will win in accordance to Darwin and in
which move the system further from equilibrium. Self accordance to the tentative fourth law of thermody
organizing processes get started. This is reflected in namics. Notice that the resources are always limited
decreased internal entropy and increased internal relatively to the number of possible offsprings.
organization. Therefore there will always be a competition about
The open question of this section is which of many the resources and this competition explains together
possible pathways will an ecosystem take in realizing its with the huge variation of properties even by the same