Ecological hierarchies and self-organisation – Pattern analysis

Basic and Applied Ecology 11 (2010) 572–581
Ecological hierarchies and self-organisation – Pattern analysis, modelling
and process integration across scales夽
Hauke Reutera,∗ , Fred Joppb,c , José M. Blanco-Morenod , Christian Damgaarde ,
Yiannis Matsinosf , Donald L. DeAngelisg
a
Leibniz Center for Tropical Marine Ecology, Fahrenheitstr. 6, 28357 Bremen, Germany
Department of Biology, University of Miami, FL 33124, USA
c
Leibniz-Institute of Freshwater Ecology and Inland Fisheries, 12587 Berlin, Germany
d
Department of Plant Biology, University of Barcelona, Av. Diagonal 645, 08028 Barcelona, Spain
e
NERI-University of Aarhus, Denmark
f
Department of Environment, University of the Aegean, University Hill Xenia Building, Mytilene 81100, Greece
g
U.S. Geological Survey, Department of Biology, University of Miami, Coral Gables, FL 33124, USA
b
Received 6 May 2009; accepted 4 August 2010
Abstract
A continuing discussion in applied and theoretical ecology focuses on the relationship of different organisational levels and
on how ecological systems interact across scales. We address principal approaches to cope with complex across-level issues in
ecology by applying elements of hierarchy theory and the theory of complex adaptive systems. A top-down approach, often
characterised by the use of statistical techniques, can be applied to analyse large-scale dynamics and identify constraints exerted
on lower levels. Current developments are illustrated with examples from the analysis of within-community spatial patterns and
large-scale vegetation patterns. A bottom-up approach allows one to elucidate how interactions of individuals shape dynamics at
higher levels in a self-organisation process; e.g., population development and community composition. This may be facilitated
by various modelling tools, which provide the distinction between focal levels and resulting properties. For instance, resilience
in grassland communities has been analysed with a cellular automaton approach, and the driving forces in rodent population
oscillations have been identified with an agent-based model. Both modelling tools illustrate the principles of analysing higher
level processes by representing the interactions of basic components.
The focus of most ecological investigations on either top-down or bottom-up approaches may not be appropriate, if strong
cross-scale relationships predominate. Here, we propose an ‘across-scale-approach’, closely interweaving the inherent potentials
of both approaches. This combination of analytical and synthesising approaches will enable ecologists to establish a more
coherent access to cross-level interactions in ecological systems.
夽 This paper is based on a session held during the annual meeting of the German Ecological Society (GfÖ) and the European Ecological Federation
(EURECO) in 2008.
∗ Corresponding author. Tel.: +49 04212380058; fax: +49 04212380030.
E-mail address: [email protected] (H. Reuter).
1439-1791/$ – see front matter © 2010 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.
doi:10.1016/j.baae.2010.08.002
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
573
Zusammenfassung
In der theoretischen und angewandten Ökologie ist das Verhältnis der Organisationsebenen und wie ökologische Systeme über
unterschiedliche Skalen interagieren, weiterhin eine offene Diskussion. Unter Verwendung von Elementen der Hierarchitätstheorie und der Theorie der Komplexen Adaptiven Systeme werden methodische Ansätze erläutert, mit denen sich komplexe,
ebenenübergreifende Wechselwirkungen untersuchen lassen. ‘Top-Down’-Ansätze, die häufig durch die Benutzung statistischer Verfahren charakterisiert sind, werden bevorzugt zur Analyse großskaliger Dynamiken angewendet. Aktuelle methodische
Entwicklungen werden anhand von Beispielen bei der Analyse von großskaligen räumlichen Vegetationsmustern und räumlichen Mustern innerhalb von Gemeinschaften erläutert. Mit einem ‘Bottom-Up’ Ansatz wird gezeigt, wie die Interaktionen
von individuellen Organismen die Dynamiken auf höheren Organisationsebenen (z.B. Populationsentwicklungen und Zusammensetzungen von Gemeinschaften) in einem Selbstorganisationsprozess bestimmen. Dies wird durch Modellierungswerkzeuge
ermöglicht, die eine Unterscheidung zwischen der Fokalebene und den resultierenden Eigenschaften erlauben. Anhand der
Analyse des Resilienzverhaltens einer Grünlandgemeinschaft mit einem Zellulären Automaten und die Untersuchung der systembestimmenden Faktoren bei Populationszyklen von Kleinsäugern mit einem agenten-basierten Modell wird dieser Ansatz
erläutert. Beide Modellansätze illustrieren das Prinzip, höhere Organisationsebenen durch die Darstellung interagierender
Komponenten auf unteren Organisationsebenen zu analysieren.
Der Fokus vieler ökologischer Untersuchungen, der ausschließlich auf ‘Top-Down’ oder ‘Bottom-Up’-Ansätzen beruht,
erscheint bei Systemen mit starken Wechselwirkungen zwischen den Integrationsebenen nicht angemessen. Bei ebenenübergreifenden Fragestellungen schlagen wir daher die Verwendung eines ‘Across-Scale’-Ansatzes vor, der die Potentiale beider
Möglichkeiten einsetzt: Durch die Kombination von analytischen und synthetisierenden Ansätzen wird so ein konsistenter
Zugang zu skalenübergreifenden Wechselwirkungen in ökologischen Systemen ermöglicht.
© 2010 Gesellschaft für Ökologie. Published by Elsevier GmbH. All rights reserved.
Keywords: Across-scale integration; Hierarchy theory; Self-organisation processes; Spatial pattern; Vegetation pattern; Cellular automaton
models; Agent-based/individual-based models
Introduction: analysing connections
between ecological integration levels
A fundamental aim in ecology is to investigate how the
diverse elements at different organisational levels interrelate and how they interact with each other across scales.
Cross-scale interactions are central to several of the current
debates in ecology and conservation concerning biodiversity,
the establishment of invasive species, and long-term effects
of habitat change (Kerr, Kharouba, & Currie, 2007). Ecological systems are generally perceived as complex systems
organised at different nested hierarchical levels. The question
of how the dynamics at specific levels depend on components
and dynamics of neighbouring levels is still one of the most
open issues in ecology, as it addresses the central problem of
how level-specific processes can be conceptually connected
to provide a consistent assessment of processes acting at
different scales (Müller, 1992; Kolasa, 2005). A common
example for this open issue is the question of how the feedback between individual behaviour and population dynamics
influences the large-scale species pool (Meyer et al., 2011)
and how large-scale environmental factors in turn influence
species distributions and community composition.
Each of the organisational levels within an ecological system is defined by its respective components and dynamics,
which can be related to specific research perspectives and
theories.
As a result of different methodological approaches and
depending on the focal hierarchical levels, the data qual-
ity and richness vary markedly while crossing the scales.
A conspicuous lack of unifying concepts contrasts with
the strong need for practical approaches (e.g., related to
biodiversity issues, establishment of invasive species, long
term effects of habitat change and fragmentation) that can
connect levels and, therefore, allow one to investigate influences of biotic interactions on environmental constraints,
and vice versa. The abundance of different approaches that
have been applied within different disciplines of biology to
the different hierarchical levels reflects the methodological
problems and the need for interdisciplinary work from fields
such as physiology, behavioural ecology, population ecology and landscape ecology, and also shows the necessity
for the integration of adequate methods. Currently, however,
a perspective focused on single disciplines predominates in
ecology.
From the landscape- and macro-ecological points of view,
the focus of interest is on the emergence of large-scale properties (Storch & Gaston, 2004), and on how the different
patterns are connected to underlying processes (Schröder &
Seppelt, 2006). From the opposite perspective of bottomup causation, where modelling plays an important role in
establishing theories, the analysis of the relationships among
different hierarchical levels of organisation is thought to
be one of the greatest challenges (Ratzé, Gillet, Müller, &
Stoffel, 2007). Dieckmann, Law, and Metz (2000) express the
hope that this challenge would lead to manageable approximations that give a better understanding of the generic
properties of ecological dynamics. Ultimately, ecological
574
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
modelling should connect dynamics of different temporal,
spatial, and organisational scales, ranging from small-scale
variability to phenomena of large-scale aggregated quantities
(Pascual, 2005).
The theoretical background addressing methodological
and practical issues of cross-scale relationships is contained
in the ideas of hierarchy theory (Miller, 1978; Allen &
Starr, 1982) and the concept of ecological systems behaving as complex adaptive systems (for an overview see Levin,
1998). In complex adaptive systems (CAS) the overall system
behaviour emerges as a self-organising process of interacting
objects that have the ability to evolve and adapt to changing
environments. Viewing ecological systems as CAS allows
one to capture variability and heterogeneity at lower levels and focuses the analysis on the respective influences
of non-linear dynamics, contingency and how macroscopic
system properties feed back on subsequent developments
(Hartvigsen, Kinzig, & Peterson, 1998). Hierarchy theory
originated as a concept from General Systems Theory, with
the primary focus on dealing with complexity, different levels of organisational structures, and scaling issues. In this
sense, the theoretical concept of hierarchy theory helps to
analyse ecological complexity across scales. It structures
ecological systems in a set of hierarchically nested organisational levels and loosely coupled components (holons)
within each level. This goes along with decreasing process
rates (frequencies of processes) and increasing size of entities
towards higher levels. Thus each level shows an asymmetric relationship, exerting general constraints on the level
directly beneath it, while its components represent the mechanisms underlying the behaviour of the level directly above
it. Hence, fast processes mostly operate on small spatial
units, while slow processes have broad spatial extents. Due
to signal filtering procedures, low-frequency units on higher
levels have the potential to constrain processes of higher
frequencies (Müller, 1992). In this sense hierarchy theory
describes the relationships and connectivity of organisation
levels and helps to address complexity in ecology in the form
of modularity in structure and functionality (Wu & David,
2002).
This paper focuses on concepts that facilitate the integration of processes and patterns (Turner & Gardner, 1991)
over different organisation levels. In research questions
relating to across-scale analysis we emphasise the necessity of combining (1) a bottom-up causation approach on
how interactions of individuals shape dynamics at higher
levels with (2) a top-down analysis of how higher levels exert constraints and influence the specific lower level
processes. This is particularly relevant for addressing scientific issues, in which the results from the analysis of
different integration levels are relevant; e.g. invasion processes, management of species, changes in community
structure due to global change. We outline how these
approaches can be used for developing future research directions in analysing ecological process interactions across
levels.
Top-down analyses – identification of
constraints
Ecology has a long history of using pattern analysis and
sophisticated statistical tools to analyse aggregated information (e.g. biodiversity indices, biomass distribution, regional
productivity) at higher organisation levels and, thereby, to
determine the factors that constitute top-down constraints on
lower levels. After the relevant scale at which to examine a
pattern has been determined, a process-based scheme of the
global system may be devised, which is mostly characterised
by different scale-dependent domains (focal level). In a second step, the quality of this scheme can be confirmed by an
extrapolation process across scales, in which the pattern at
one scale is matched with processes at another scale.
Among these approaches to scale-explicit pattern exploration, we want to highlight two exemplary types of analyses
relating to the analysis of processes that structure communities and vegetation patterns.
Analysis and prediction of vegetation patterns
The principal explanatory factor constraining the distribution of vegetation types is the abiotic environment (Walter,
1985) at the regional and biome level. If the distributions of
temperatures and rainfall, as well as the soil type, are known,
then a trained plant ecologist will often be able to predict
the functional vegetation composition. With empirical filter
models (Keddy, 1992), the occurrence probability of specific
plant species may be predicted quantitatively using regression models in which abiotic environmental variables are
used as predictor variables (Pearce & Ferrier, 2000; Austin,
2002; Damgaard, 2006; Damgaard, 2008). A great advantage
is that the predictor variables are often easy to measure, readily available in databases and may be retrieved for large areas
(e.g. through analysing satellite imaging) thus making them
applicable as proxies for the focal variables which are not
available in the same extent.
However, many ecosystems are characterised by dynamic
properties, in which different species and abiotic factors interact in complicated ways, and these dynamic properties are
thought to play a critical role in determining the state of the
ecosystem (although: see Hubbell, 2001). Consequently, simple filter models can only be viewed as a first step towards
a trustworthy ecological prediction. If the model has limited reliability or discrimination capacity (Pearce & Ferrier,
2000), then the next logical step is to include the most important dynamical features of the ecosystem in the modelling
process. Here one of the main problems is that, even though
good working hypotheses exist regarding the principles governing the nature of the interactions, only a few terrestrial
ecosystems have been sufficiently investigated to allow quantitative predictive models to be formulated (Terry, Ashmore,
Power, Allchin, & Heil, 2004).
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
Analysis of within-community spatial patterns
As explained above, the main focus of pattern analysis
is the search for insights on the underlying mechanisms that
condition the structure of communities (Borcard & Legendre,
2002). However, all environments exhibit distinct internal
heterogeneity to some degree, showing rather aggregated
patterns than regular or random distributions of individuals (Wiegand, Gunatilleke, & Gunatilleke, 2007). At spatial
scales matching the activity range of the organisms, the environmental heterogeneity is usually less noticeable and the
structure may be determined by the nature of the interactions
among individuals (Uriarte, Condit, Canham, & Hubbell,
2004). In these cases, separating the top-down control
exerted by the environment (e.g. correlation of environmental variables with distribution or growth of individuals)
from bottom-up forces (interaction among individuals) can
improve our comprehension of the system. This is especially
true in species-rich ecosystems, where detailed knowledge
of the species’ biology is lacking, and theories explaining species coexistence and community structure need to
rely heavily on spatial pattern analysis. The development
of spatial models incorporating critical scales of clustering
(at landscape scale, at the group level or at the individual level) is needed for exploring the relationship between
the spatial structure of communities and of populations
and factors such as environmental constraints, competition,
dispersal and recruitment limitations, and Janzen–Connell
effects (Wiegand, Gunatilleke, Gunatilleke, & Okuda, 2007).
However, in spite of recent developments on spatial analysis to reveal the underlying processes, there is at least one
important area that will benefit from further research: the
extension of spatial analysis to dynamic changes in pattern
formation (Fortin & Dale, 2005).
Bottom-up organisation – emergence of
higher level structures and dynamics
The complementary approach to the top-down analysis
of aggregated entities consists in assessing ecological complexity by employing a bottom-up causation perspective that
considers ecosystems as self-organised complex adaptive
systems. An isolated perception of the system components
is not sufficient to explain the properties of higher levels,
which emerge as the result of lower level interactions (e.g.:
Nielsen & Müller, 2000; Müller & Nielsen, 2000).
Besides empirical approaches (Meyer et al., 2011), different modelling approaches with the potential to process
changing interaction structures between a large number of
heterogeneous components have been developed. In the
following, we describe two of these approaches (cellular
automata and agent-based models) that have strongly facilitated inclusion of cross scale processes, and self-organisation
over several integration levels, and are applicable to rep-
575
resenting a wide range of ecological processes (Breckling,
2002; Reuter et al., 2008).
Cellular automata
One approach for dealing with the inherent complexity of
ecological processes is to use cellular automaton (CA) models, which were introduced by Ulam and von Neumann in
the late 1940s, early 1950s (Ulam, 1950 in Beyer, Sellers, &
Waterman, 1985; von Neuman, 1951 in Burks, 1969). CA
can reproduce the emergence of complex behaviour from
simple rule sets iteratively applied to interacting cells on a lattice (Wolfram, 1984). They provide a simulation framework
allowing for simultaneous occurrence of concurrent ecological processes at the local (i.e. neighbourhood) scale, and are
often used to deal with spatial competition and colonisation
processes.
Due to their inherent ease of implementation and replication of spatial patterns, CA models have been widely
applied to ecological problems related to spatial processes,
such as epidemic propagation (Sirakoulis, Karafyllidis, &
Thanaikakis, 2000), plant population dynamics (e.g. Pascual,
Roy, Guichard, & Flier, 2002), coral interactions (Langmead
& Sheppard, 2004) and post-disturbance resilience (Matsinos
& Troumbis, 2002).
The CA model by Matsinos and Troumbis (2002) of grassland community dynamics illustrates how interactions at the
level of small patches (dispersal and competition) determine overall community structure. The model focuses on
the effect of community resilience in relation to gap-creating
disturbances (i.e. fire) that are imposed at different spatial
scales. Simulations are based on data from an experimental community with five grassland species with dispersal and
competition parameters being derived at both the individual and species level from experimental plots designed to
test biodiversity effects on ecosystem functioning in Lesvos,
Greece. Results at the level of the whole simulated area
showed that plants dispersing farther had an overall competitive advantage, due to their success in colonising open
areas, as compared to better competitors, especially in the
cases of disturbance-mediated creation of gaps in coverage.
An increase in species number led to more resilient communities and a higher percent cover. A further model adaptation
incorporated (i) a scale-related neighbourhood structure, (ii)
asymmetrical hierarchy in competition and (iii) invasion processes, which altogether linked processes being effective at
different spatio-temporal scales (e.g. direct spatial competition versus dispersal ability).
The neighbourhood structure was based on dispersal
attributes of species and showed significant change in final
assemblage patterns, resulting in decreased abundance for
the short-dispersers. Asymmetrical hierarchy, in terms of
modelling competition as a stochastic process, altered the
composition of end-state communities significantly, favouring assemblages with low overall diversity. Invasion as a
576
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
spatial process proved also to alter the overall pattern of abundance. However, a great amount of information is necessary
for parametrising the model; in particular, an experimental determination of pairwise and higher order competition
strength is needed. The approach highlights the emergence
of complex spatial community patterns from simple local
interactions.
tions and predator–prey interaction may lead to the observed
dynamics, thus offering an explanation for the controversial discussions about the mechanisms underlying rodent
dynamics, and integrating some of the previously existing
opposing explanations. The ABM approach thus allows one
to represent the emergence of empirically determined patterns (Grimm et al., 2005) and to analyse the processes which
determine the basic system functioning.
Agent-based modelling concepts
Agent-based models (ABM, synonymous with
‘individual-based model’) simulate the global consequences of locally interacting units (Breckling, 2002). They
represent basic entities (often individual organisms) with
their properties and with rules determining how to change
these properties depending on the internal state, the state of
the environment and interactions with other entities. In the
case of individual organisms the rules may refer, for example,
to behaviour, physiological processes or resource utilisation.
Higher level properties then emerge in a self-organisation
process explicitly as a result of the interactions of lower
level components (Breckling, Müller, Reuter, Hölker, &
Fränzle, 2005; Reuter et al., 2005). Due to the potential to
easily simulate spatial processes, and incorporate a variable
interaction structure of the components and stochastic
processes, ABM have led to a paradigm shift in ecological
modelling (Grimm & Railsback, 2005) and have been
applied to simulate a large number of topics in applied as
well as theoretical ecology (DeAngelis & Mooij, 2005).
The example of rodent population cycles as higher level
phenomena may illustrate the potential of the approach. Population cycles of rodents have given rise to decade-long
controversies concerning the driving processes that lead to
these regular multi-annual dynamics (e.g. Stenseth, 1999).
Many explanatory hypotheses have been put forward relating, for example, to intrinsic rodent population processes
or different consumer–resource interactions. With an ABM
approach Reuter (2005) was able to integrate many of the
relevant factors by considering three trophic levels, spatial
interactions and the life history traits of the involved species,
thus extending previous modelling approaches that focused
on a small set of these factors (Turchin & Hanski, 2001).
The ABM allowed for the parametrisation of the life history
and inter-/intra-specific interactions for different species by
applying parameters e.g. for reproduction (litter size, seasonality), habitat use, resource exploitation, dispersal and
energetics. The model dynamics at the community level
resulted from the repeated interactions among the represented
rodent and predator individuals, which responded to internal
and external stimuli according to their inherent rule system
and parameter configuration (Fig. 1).
Overall system dynamics thus emerged in an, a priori,
non-predictable manner, as a result of the interactions of
defined processes and components. Model results indicated
that, under the same global conditions, both resource limita-
Combining potentials in an ‘across-scale
approach’
Ecological studies predominantly apply either an aggregated top-down approach or a mechanism-based bottom-up
approach for studying patterns. For most of the prevailing
scientific questions in ecology, which focus on investigating
specific organisation levels, this seems to be adequate. However, studies that analyse processes on different integration
levels and their interactions across their respective scales usually cannot be covered by either approach alone. A top-down
approach might ignore central aspects, as it is predominantly
descriptive, and information is lost through an aggregation
process (Table 1). As causal analyses are not the primary
focus, resulting explanations have a high potential of overlooking ecological complexity. The examples of analysing
large-scale patterns (see “Top-down analyses – identification of constraints”) illustrate a correlational linkage between
abiotic pattern and predictor variables applied to access the
underlying processes and mechanisms.
On the other side, bottom-up approaches, which are
focussed on the analysis of the causal mechanisms, often
originate from single cases. Generalization of findings from
such approaches is difficult and analytical solutions are nearly
impossible. However, a transfer of the results to similar situations is often possible, as basic mechanisms, e.g. ontology,
behaviour of organisms, are included, which lead to the synthesis of higher level dynamics. A problem with this approach
is the difficulty in deciding which processes are relevant for
the investigated dynamics, and should therefore be included
either as internal system-determining factors contributing
to the self-organisation or as external constraints. Here, a
decisive risk exists in losing the explanation power in the
overall ecological complexity of the investigated system and,
instead, constructing a representation which is as complex as
the system under study (Pascual, 2005). In the rodent cycle
model (Reuter, 2005) many of the relevant hypotheses have
been included in the ABM approach, thus clearly extending
the equation-based representations (e.g. Turchin & Hanski,
2001). However, as the model representation has to restrict
itself to a limited number of processes and components, there
is a certain possibility that the claim of including the relevant
processes and components will not be met. A thorough testing
of simulation processes and results on all included organisational levels may ensure high congruence of model operation
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
577
Fig. 1. In the agent-based model on population cycles the dynamics of higher organisation levels (populations and community) arise as a
result of repeated interactions of individuals of several species. The model dynamics are mostly defined on the individual level, which thus
constitutes the definition level of the model (environmental drivers not shown).
with the studied system, thus constituting a first step in validating models that include self-organisation. Nevertheless,
the careful delimitation of considered model processes and
the conditions for which these are valid remains a constant
challenge.
To provide approaches that overcome the outlined restrictions we propose to further develop a combination of
both methodological approaches within an ‘across-scaleapproach’ (Fig. 2). This constitutes a way of avoiding the
limitations and drawbacks that normally occur with either
method alone. A more comprehensive approach would thus
consist in the combined application of analytical methods
and single models that originally were constructed for specific focal levels. In such a coupling procedure, the model
output from lower levels serves as the mechanistic description of emergence for higher level models and, vice versa,
the upper level models and analyses define the constraints
to the lower level models. This coupling procedure constitutes a valuable tool for testing the overall consistency of the
representation of the investigated ecological systems.
Table 1. Overview of the general properties of bottom-up and top-down-approaches in analysis and modelling.
Properties of the approach
Aggregated top-down-analysis
Process-based bottom-up causation
Primary focus
Large scale pattern at higher integration levels
(usually snapshot analysis)
Description and analysis of overall ecological
pattern in relation to environmental conditions.
Causality is derived from statistical relationships
Aggregated general perspective
Risk of ignoring ecological complexity
Processes and interactions between components
focused on lower levels (dynamic modelling)
Causal relationships are directly applied to
describe interactions of subsystems and to derive
emergent properties and self-organised dynamics
Often, starting points are specific cases
Risk of loosing the explanation power in the
overall ecological complexity
Causality
Specificity
Complexity
578
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
Fig. 2. Steps of the proposed across-scale approach with respective examples on each step.
Because scale still is and will continue to be a central working issue in ecology (Urban, 2005), we propose the following
procedure on scaling methods when approaching hierarchically nested phenomena in the ecosystem context:
1. Characterise the nature of constraints of higher level properties in a top-down analytic approach by breaking down
the known system into modules a priori. Here, a distinction of the hierarchical levels, with defining the holons
and their relative positions, is started for the top levels.
This comprises also the development of large-scale predictor variables as are used within statistical approaches
(see “Top-down analyses – identification of constraints”).
Models that use species distributions or distributions of
suitable habitats (e.g. SDM) to make predictions are examples for this step of the procedure.
2. Represent the nature of lower level properties (dynamics, components) in a synthesising bottom-up approach
by piecing together the known sub-systems to the global
system and analyse the emerging properties. Modelling
examples for this step are integrating approaches like
agent-based models (ABM) that synthesise higher level
dynamics by representing large numbers of components
and their variable interaction structure (see “Bottom-up
organisation – emergence of higher level structures and
dynamics”).
3. Describe the connectivity between the scale-bounded
phenomena in relation to the top-down and bottom-up
approaches. This includes a further aspect to the analysis
by determining which top-down constraints are decisive
for the investigated phenomena and how they relate to
mechanisms and factors leading to the self-organisation
of the system. Here, the regulation principles between the
hierarchical phenomena are analysed and heterogeneities
between the scales get clearer. For example, extending the example rodent model described in “Bottom-up
organisation – emergence of higher level structures and
dynamics” to a cross-scale approach would imply adding
an analysis of factors that operate on large scales or
higher integration levels (e.g. habitat distribution, climate
and seasonality, community composition). These analyses could be used to derive different scenario conditions
for model simulations, which in turn would allow one
to broaden the research questions and facilitate investigation of how the intrinsic dynamics of the modelled
communities are constrained (or influenced) by top down
processes. A posteriori, discuss the synthesising and the
analytical approaches again, and try to fill gaps of knowledge or inconsistencies. If necessary, repeat the preceding
steps until a consistent picture appears, which conforms
to real-world patterns.
As a consequence, these hybrid analyses and modelling
approaches will enable the establishment of a more coherent
analysis of cross-level ecological interactions. The application of these analyses will also have relevance for specific
levels, as they help to define the range of model validity with
respect to the integrated system features. A few additional
studies can be found that utilise a methodological concept
similar to the one already described. A first example involves
the analyses of the environmental implications of genetically
modified oilseed rape in Northern Germany. In this study,
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
Breckling et al. (2010) applied an approach consisting of
(1) a combination of large scale analysis of landscape structure and environmental constraints (using satellite imagery,
phenomenological and weather data), (2) a simulation of
dispersal and persistence of oilseed rape for representative
locations (agent-based modelling) and (3) a transition back
to the landscape level to analyse large-scale impact (statistical
extrapolation of model results to landscape parts with matching input conditions). Another example is the ATLSS-project
in the Everglades, Florida. A combination of modelling methods allows one to capture dynamics at specific organisational
and trophic levels with environmental models and models of
lower trophic levels (the lowest level being physical models) constituting the input conditions for models on higher
trophic levels (for example, spatially explicit models of the
small fish community, seaside sparrows, and wading birds;
see DeAngelis et al., 1998). In the context of simulating
ecological dynamics in the marine environment, the discussion of applying model coupling (hybrid approaches) is
much more advanced. One example is the coupling of a biological model representing individual larval ontology with
large scale oceanographic modelling of physical factors to
represent spatio-temporal population recruitment of teleosts
(Rose et al., 2007; Gallego, in press). Also the discussion on
developing integrated model systems to represent community dynamics (Cury et al., 2008) is another example for the
perspectives of model coupling. However, in our opinion this
discussion would clearly benefit from a distinction between
processes relating to different organisational levels and a
conceptual separation between analytical and synthesising
approaches.
Conclusions
In this paper we have presented examples of current
synthesising approaches and techniques that allow one to
connect processes over different organisational levels. These
approaches are based on the concept of ecological hierarchies as an abstraction, which integrates numerous empirical
and theoretical studies. As these abstractions are used to
understand ecological dynamics, the validity of the operating
definitions for the scales, functions and parameters that are
used is decisive for the evaluation process.
We showed how these and comparable approaches have
been implemented in current research, and how they can be
used to integrate structural and functional subsystems, and,
hence, to test the consistency of ecological knowledge representation. With accelerating global change in environmental
conditions, the combination of complementary approaches
gains increasing importance. By an explicit consideration of
both top-down analysis and bottom-up synthesis, the outlined
across-scale approach facilitates the application to new conditions, making model-based evaluations and extrapolations
more powerful. However, we are aware that the approach
we have described is usually accompanied with greater com-
579
plexity and often with increased requirements with respect to
empirical knowledge, parameters and model development.
Thus it will often be restricted to situations that allow the
involvement of scientists from different disciplines.
The analysis of pattern and process, and modelling across
scales remains one of the biggest challenges in theoretical and applied ecology, but it needs to be tackled to
meet the challenge of the increasing speed of environmental
change.
Acknowledgements
We thank Bram Van Moorter, Mathieu Basille and two
anonymous reviewers for very valuable comments on previous versions of this manuscript. We are very grateful to the
organisers of the Annual Meeting of the Ecological Society
of Germany, Austria and Switzerland (GfÖ) and of the XI
conference of the European Ecological Federation EURECO
in Leipzig, 2008, for hosting the session which inspired the
results this paper.
References
Allen, T. F. H., & Starr, T. B. (1982). Hierarchy: Perspectives for
ecological complexity. Chicago: Chicago University Press.
Austin, M. P. (2002). Spatial prediction of species distribution: An
interface between ecological theory and statistical modelling.
Ecological Modelling, 157, 101–118.
Beyer, W. A., Sellers, P. H., & Waterman, M. S. (1985). Stanislaw
M. Ulam’s contribution to theoretical theory. Letters in Mathematical Physics, 10, 231–241.
Borcard, D., & Legendre, P. (2002). All-scale spatial analysis of
ecological data by means of principal coordinates of neighbour
matrices. Ecological Modelling, 153, 51–68.
Breckling, B. (2002). Individual based modelling: Potentials and
limitations. Scientific World Journal, 2, 1044–1062.
Breckling, B., Müller, F., Reuter, H., Hölker, F., & Fränzle, O.
(2005). Emergent properties in individual-based ecological models – Introducing case studies in the ecosystem research context.
Ecological Modelling, 186, 376–388.
Breckling, B., Reuter, H., Middelhoff, U., Glemnitz, M., Wurbs,
A., Schmidt, G., et al. (2010). Risk indication of Genetically
Modified Organisms (GMO): Modelling environmental exposure and dispersal across different scales oilseed rape in Northern
Germany as an integrated case study. Ecological Indicators,
doi:10.1016/j.ecolind. 2009.03.002
Burks, A. W. (1969). Von Neumann’s self-reproducing automata.
Technical Report, University of Michigan.
Cury, P. M., Shin, Y. J., Planque, B., Durant, J. M., Fromentin, J.,
Kramer-Schadt, M., et al. (2008). Ecosystem oceanography for
global change in fisheries. Trends in Ecology & Evolution, 23,
338–346.
Damgaard, C. (2006). Modelling ecological absence–presence
data along an environmental gradient: Threshold levels of
the environment. Environmental and Ecological Statistics, 13,
229–236.
580
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
Damgaard, C. (2008). Modelling pin-point plant cover data
along an environmental gradient. Ecological Modelling, 214,
404–410.
DeAngelis, D. L., & Mooij, W. M. (2005). Individualbased modeling of ecological and evolutionary processes.
Annual Reviews in Ecology, Evolution and Systematics, 36,
147–168.
DeAngelis, D. L., Gross, L. J., Huston, M. A., Wolff, W. F.,
Fleming, D. M., Comiskey, E. J., et al. (1998). Landscape
modeling for everglades ecosystem restoration. Ecosystems, 1,
64–75.
Dieckmann, U., Law, R., & Metz, J. A. J. (2000). The geometry of ecological interactions – Simplifying spatial complexity.
Cambridge: Cambridge University Press.
Fortin, M. J., & Dale, M. R. T. (2005). Spatial analysis: A guide for
ecologists (1st ed.). Cambridge: Cambridge University Press.
Gallego, A. (in press). Bio-physical models: An evolving tool in
marine ecological research. In F. Jopp, H. Reuter, & B. Breckling (Eds.), Modelling complex ecological dynamics. Chapter 20,
Heidelberg: Springer (ISBN 978-3-642r-r05028-2).
Grimm, V., & Railsback, S. (2005). Individual-based modeling and
ecology. Princeton: Princeton University Press.
Grimm, V., Revilla, E., Berger, U., Jeltsch, F., Mooij, W. M.,
Railsback, S. F., et al. (2005). Pattern-oriented modeling of
agent-based complex systems: Lessons from ecology. Science,
310, 987–991.
Hartvigsen, G., Kinzig, A., & Peterson, P. (1998). Use and analysis
of complex adaptive systems in ecosystem science: Overview of
special section. Ecosystems, 1, 427–443.
Hubbell, S. P. (2001). The unified neutral theory of biodiversity and
biogeography. Princeton: Princeton University Press.
Keddy, P. A. (1992). Assembly and response rules: Two goals for
predictive community ecology. Journal of Vegetation Science, 3,
157–164.
Kerr, J. T., Kharouba, H. M., & Currie, D. J. (2007). The macroecological contribution to global change solutions. Science, 316,
1581–1584.
Kolasa, J. (2005). Complexity, system integration, and susceptibility
to change: Biodiversity connection. Ecological Complexity, 2,
431–442.
Langmead, O., & Sheppard, C. (2004). Coral reef community
dynamics and disturbance: A simulation model. Ecological Modelling, 175, 271–290.
Levin, S. A. (1998). Ecosystems and the biosphere as complex
adaptive systems. Ecosystems, 1, 431–436.
Matsinos, Y. G., & Troumbis, A. Y. (2002). Modeling competition,
dispersal and effects of disturbance in the dynamics of a grassland community using a cellular automaton model. Ecological
Modelling, 149, 71–83.
Meyer, K., Münkemüller, T., Schiffers, K., Schädler, M., Calabrese,
J., Basset, A., et al. (2011). Crossing scales from biotic interactions to community patterns. Basic and Applied Ecology.
Miller, J. G. (1978). Living systems. New York: McGraw-Hill.
Müller, F. (1992). Hierarchical approaches to ecosystem theory.
Ecological Modelling, 63, 215–242.
Müller, F., & Nielsen, S. N. (2000). Ecosystems as subjects of
self-organized processes. In S. E. Joergensen, & F. Müller
(Eds.), Handbook of ecosystem theories and management (pp.
177–194). New York: CRC Publishers.
Nielsen, S. N., & Müller, F. (2000). Emergent properties of ecosystems. In S. E. Joergensen, & F. Müller (Eds.), Handbook of
ecosystem theories and management (pp. 195–216). New York:
CRC Publishers.
Pascual, M. (2005). Computational ecology: From the complex to
the simple and back. PLoS Computational Biology, 1, e18.
Pascual, M., Roy, M., Guichard, F., & Flier, G. (2002). Cluster size
distributions: Signatures of self-organization in spatial ecologies. Philosophical Transactions of the Royal Society B, 357,
657–666.
Pearce, J., & Ferrier, S. (2000). Evaluating the predictive performance of habitat models developed using logistic regression.
Ecological Modelling, 133, 225–245.
Ratzé, C., Gillet, F., Müller, J. P., & Stoffel, K. (2007). Simulation
modelling of ecological hierarchies in constructive dynamical
systems. Ecological Complexity, 4, 13–25.
Reuter, H. (2005). Community processes as emergent properties:
Modelling multilevel interaction in small mammals communities. Ecological Modelling, 186, 427–446.
Reuter, H., Hölker, F., Middelhoff, U., Jopp, F., Eschenbach, C.,
& Breckling, B. (2005). The concepts of emergent and collective properties in individual based models – Summary and
outlook of the Bornhöved case studies. Ecological Modelling,
186, 489–501.
Reuter, H., Jopp, F., Hölker, F., Eschenbach, C., Middelhoff, U., &
Breckling, B. (2008). The ecological effect of phenotypic plasticity – Analysing complex interaction networks with agent based
models. Ecological Informatics, 3, 35–45.
Rose, K. A., Werner, F. E., Megrey, B. A., Aita, M. N., Yamanaka,
Y., Hay, D. E., et al. (2007). Simulated herring growth responses
in the Northeastern Pacific to historic temperature and zooplankton conditions generated by the 3-dimensional NEMURO
nutrient–phytoplankton–zooplankton model. Ecological Modelling, 202, 184–195.
Schröder, B., & Seppelt, R. (2006). Analysis of pattern–process
interactions based on landscape models – Overview, general concepts, and methodological issues. Ecological Modelling, 199,
505–516.
Sirakoulis, G. C. H., Karafyllidis, I., & Thanaikakis, A. (2000). A
cellular automaton model for the effects of population movement
and vaccination on epidemic propagation. Ecological Modelling,
133, 209–229.
Stenseth, N. C. (1999). Population cycles in voles and lemmings:
Density dependence and phase dependence in a stochastic world.
Oikos, 87, 427–461.
Storch, D., & Gaston, K. J. (2004). Untangling ecological complexity on different scales of space and time. Basic and Applied
Ecology, 5, 389–400.
Terry, A. C., Ashmore, M. R., Power, S. A., Allchin, E. A., & Heil,
G. W. (2004). Modelling the impacts of atmospheric nitrogen
deposition on Calluna-dominated ecosystems in the UK. Journal
of Applied Ecology, 41, 897–909.
Turchin, P., & Hanski, I. (2001). Contrasting alternative hypotheses about rodent cycles by translating them into parameterized
models. Ecology Letters, 4, 267–276.
Turner, M. G., & Gardner, R. H. (Eds.). (1991). Quantitative methods in landscape ecology. New York: Springer Verlag.
Urban, D. L. (2005). Modeling ecological processes across scales.
Ecology, 86, 1996–2006.
Uriarte, M., Condit, R., Canham, C. D., & Hubbell, S. P. (2004).
A spatially explicit model of sapling growth in a tropical forest:
Does the identity of neighbours matter? Journal of Ecology, 92,
348–360.
H. Reuter et al. / Basic and Applied Ecology 11 (2010) 572–581
Walter, H. (1985). Vegetation of the earth and ecological systems of
the geo-biosphere (3rd ed.). Berlin: Springer.
Wiegand, T., Gunatilleke, S., & Gunatilleke, N. (2007). Species
associations in a heterogeneous Sri Lankan dipterocarp forest.
American Naturalist, 170, E77–E95.
Wiegand, T., Gunatilleke, S., Gunatilleke, N., & Okuda, T.
(2007). Analyzing the spatial structure of a Sri Lankan
581
tree species with multiple scales of clustering. Ecology, 88,
3088–3102.
Wolfram, S. (1984). Universality and complexity in cellular
automata. Physica D, 10(1), e35.
Wu, J., & David, J. (2002). A spatially explicit hierarchical approach
to modeling complex ecological systems: Theory and applications. Ecological Modelling, 153, 7–26.