The idea of complexity in ecology

Ecology as a Study of Complexity

It is the living world that first confronts us with real complexity. The physical world seems at first glance to be either simple or chaotic, but the patterns in the living world are evident everywhere - from coral reefs to rainforests -and they reveal deeper patterns on even the most cursory examination. Their intricacy and beauty have always astounded us, and they have always seemed qualitatively distinct from the patterns of the physical world.

Our first response to this complexity is taxonomy. We attempt to order and classify the phenomena so we may see the pattern. It is not surprising that taxonomy (first an art; now, perhaps a science) grew out of biology. Indeed, taxonomy may be the first stage in the development of any science: the emergence of the idea that there is a 'something', an object of study.

Thus we may say that the founders of systematic biology, such as ARISTOTLE and LINNAEUS, in erecting their taxonomies, were the first to study complexity in its own right.

As ecology emerged as a distinct discipline at the turn of this century, we can see the workings of taxonomists in their early attempts to come to grips with the complexity of ecological phenomena. These resulted in a lively taxonomic enterprise concerned with classifying assemblages of species, and they created an edifice of which ARISTOTLE or LINNAEUS would be proud. Ecologists such as SCHRÖTER and FLAHAULT of the Zurich-Montpellier school and TANSLEY of the British Vegetation Committee classified vegetation into ecological units in this period (MCINTOSH 1985). This taxonomy - phytogeography - closely mimicked the traditional taxonomy of systematic biology in its focus on the generation of a static typology. It coexisted rather uneasily, however, with the first efforts to classify dynamic ecological phenomena as the early proponents of ecology uncovered the inherent dynamism of ecological processes.

These attempts to classify and understand ecological processes, as distinct from ecological associations, are synonymous with the work of CLEMENTS (1905), who developed the first logical ecological system that could be described as theory. It sought to explain the complexity of ecological phenomena in terms of a holistic superorganism, developing to a climax state controlled by the climate.

The idea of complexity can therefore be found even as ecology metamorphosed from natural history, and has continued as an evident theme in ecological thought since. We see three more or less distinct traditions in the development of this thought. The first is a tradition of 'selfconscious' ecology (ALLEE et al. 1949) in which the main work is describing and defining phenomena that are considered to be distinctly ecological, and thus building directly on the taxonomic work of describing the 'object of study'. The second we call the 'heroic-tradition', where the grand questions of the field were first posed and work begun on, answering them, and where great confidence was espoused that the questions would succumb to a rigorous analytical attack. The last we tentatively name the 'evolutionary' tradition, which does not measure its progress against heroic questions, but rather with the beginnings of a consonance with other complex systems. It uses principles gleaned from the study of all evolved, adaptive systems, with the hope that a new understanding of all such systems may emerge.

These three traditions emerged one after the other during the past hundred years of development of the science, but now coexist as vigorous contributors to ecology as a whole, as well as to each other.

However, we need to understand a little more of the structure of ecology if we are to understand how the idea of complexity informs and, in turn, is informed by ecology. We need to understand that the body of ecology is riven by a major fault line. It affects all parts of the discipline and has been part of its structure since ecology first crystallised as a discipline more than a century ago. It is the dichotomy between autecology and synecology. This distinction was first made by Schröter in 1896 (CHAPMAN 1931) and concerns the proper subject matter of ecology: autecology is the study of how a population of a single species adapts to its environment, while synecology is the study of the interactions among communities of species and their environment. Autecology generally takes a reductionist approach, while synecology takes a holistic approach.

Given the strength and persistence of this fault line it may seem remarkable that the discipline has not cleaved into parts, each with its own view of ecological complexity, but ecologists themselves provide cohesion as they cross and recross the line during their professional lives. It is rare for an ecologist to remain purely on one side of this continuing argument for her entire life.

Complexity means different things on either side of the fault line, because the purpose of studying complexity differs across this divide. The synecological Clementsian approach to complexity, embodied in the superorganism, is a platonic approach, directly descended from the nineteenth century German idealism of KANT and GOETHE; the autecological phytogeography of Tansley and the Zurich-Montpellier school is more in the naturalist tradition of, say, HUMBOLDT. Thus the holist tradition stresses the search for the simplicity (the platonic ideal) underlying the observed complexity (the epiphenomena), while the reductionist tradition stresses the search for the simpler constituents or components that make up the observed complexity. Each tradition studies the complexity of ecosystems, not for itself but to reveal a putative simplicity - either the simplicity of an holistic ideal or the simplicity of reductionist components.

We who seek to study complexity in its own right face two risks in ecology: the first is that we risk being grossly misunderstood, because complexity has rarely been so studied in ecology, and the second is that we risk being 'captured' by either the reductionists or the holists as they use our insights to further their own programs. These programs are identical: to capture the soul of ecology.

Complexity and the Soul of Ecology

In reviewing GELL-MANN'S (1994) The quark and the jaguar, MARSCHALL (1994) neatly states the dilemma of those who would study complexity:

'Complexity is what financial markets, mammalian immune systems and ecological communities have in common. The ability to interact with the environment, to recognize patterns in the world and to apply acquired knowledge to the modification of future behaviour is easily detected, yet the definition of complexity remains elusive.' (MARSCHALL 1994: 45).

In terms of a complexity agenda, this description is not far removed from SEMPER'S (1881) description of a nascent ecology:

'which regards the species of animals as actualities and investigates the reciprocal relations which adjust the balance between the existence of any species and the natural, external conditions of its existence, in the widest sense of the term'. (SEMPER 1881: 33)

In the century or so between SEMPER and GELL-MANN, the idea of ecology as a study in complexity has not changed much. Nor has the issue of how to do ecology. 'Ecology as complexity' is the battleground for the soul of ecology at the end of the twentieth century, just as the 'ecosystem as microcosm' was at the end of the nineteenth. We can see this in the startlingly modern thesis of FORBES (1887) in his famous article 'The lake as a microcosm':

'The lake is an old and relatively primitive system, isolated from its surroundings. Within it matter circulates, and controls operate to produce an equilibrium comparable with that in a similar area of land. In this microcosm nothing can be fully understood until its relationship to the whole is clearly seen.'

Forbes's clear statement of a holistic approach to ecology would still encourage community ecologists and irritate population ecologists today.

The battle for the soul of ecology is the continuing struggle within the discipline between the reductionist and holist approaches to understanding ecological phenomena. It is rarely joined directly, but skirmishes have continued on shifting fields for all of the past century. The skirmishes include the controversy over CLEMENTS'S theory of the community as a superorganism at the turn of the century, the issue of the role of competition in the so-called 'golden age of theoretical ecology' in the 1920s and ‘30s (SCUDO & ZIEGLER 1978), the issue of density dependent regulation of population in the middle years of the century (ANDREWARTHA & BIRCH 1954), and the multispecies modelling of the sixties and seventies (CODY & DIAMOND 1975). The common thread in each of these debates is the nature of the proper object of study in ecology.

Throughout this century, one approach or other has had the ascendancy and claimed ecology's soul. By the middle years of the century, a fiercely empirical single species ecology made a reductionist autecology the norm. It decreed that the interesting ecological questions could be reduced to five (ANDREWARTHA & BIRCH 1954): how was the distribution and abundance of animals affected by the weather, other animals of the same kind, other organisms of different kinds, food, and a place in which to live? This was challenged by a synecological view, which first emerged in a coherent way at the famous Cold Spring Harbor Symposium on Quantitative Biology in 1957 (HUTCHINSON 1957). In this view, the most interesting and fundamental ecological questions had more to do with the dynamics of species interactions than with the physical determinants of a species' distribution. By the time of the Brookhaven Symposium on Diversity and Stability in Ecological Systems (WOODWELL & SMITH 1969), the autecological view was routed, and a synecological view came to dominate ecological thinking. This view was championed first by MACARTHUR & WILSON (1967) and more recently by ROBERT MAY (1974, 1976).

If this synecological view has a credo, it is that the manifest complexity of ecosystems is the result of a small set of ecological interactions between species - predation, competition and symbiosis - and that the dynamics of these interactions can be described by relatively simple mathematical models.

It is fair to say that this view is the dominant view in ecology today, with the only significant criticism coming from intensely empirical autecologists such as PETERS (1980), and from evolutionary ecologists such as LEVINS & LEWONTIN (1980) who maintain that both 'Cartesian reductionism' and 'obscurantist holism' should be replaced in ecology by dialectical materialism that views 'the whole as a contingent structure in reciprocal interaction with its own parts and with the greater whole of which it is a part'.

A Taxonomy of Complexity - the Tradition of 'Self-conscious' Ecology

As ecology began to separate itself from other parts of biology at the turn of the century, it did so through a process of describing and defining ecological phenomena - a taxonomic process. The major part of this work, which continues vigorously today, concerned the description, dissection and definition of ecological complexity. We can see a continuous tradition from the early phytogeographers, through the first major synthetic work in ecology - CHARLES ELTON'S (1927) Animal Ecology - to more modern work using multivariate techniques to classify biological communities (PIELOU 1984).

While early ecologists such as CLEMENTS AND TANSLEY had concentrated on describing the broad form of bio logical communities, ELTON, in his 'scientific natural history' (as he described it), was the first ecologist to dissect out the components of ecological complexity. He identified the food chain, the food web (which he called the food cycle), the ecological niche, and the pyramid of numbers. His work laid the foundation for much of the later work. His empirical work on the distribution and abundance of animals such as voles and lemmings, whose populations often fluctuate violently, established a tradition of patient, long-term field work and did much to dispel then-current ideas of a 'balance of nature'. It was thus a precursor to the complexity/stability debate which, as we will argue, was to provide the heroic tradition with a purpose for complexity. These studies also provided a rich hunting ground for theoretical ecologists keen to test their population dynamics models.

ELTON'S focus on what eats what in the community also paved the way for the trophodynamic approach (LINDEMAN 1942) which, as we will see, competed with the population dynamics approach Of MACARTHUR and MAY for many years for the intellectual leadership of the heroic tradition.

Later work in the self-conscious ecology tradition concentrated on deriving more sophisticated measures of ecological complexity by establishing the statistical functions that most parsimoniously described the distribution of commonness and rarity of species in natural communities. FISHER et al. (1943) proposed the log-series distribution while PRESTON (1948, 1962) proposed the log-normal distribution. This led to the idea of biological diversity as a measurable trait of ecosystems. Parameters from these distributions were used as diversity indices - compact measures of the degree of complexity of the communities being described. Work in this vein continued with the development of information-theoretic measures of diversity. These were first used in ecology by MARGALEF (1957) and rapidly evolved into a sophisticated suite of measures of community structure (PIELOU 1975) which dissected out the components of diversity and related them to measures of niche width and overlap.

In more recent times, the focus of this ecology has returned to the concerns of the fin de siècle phytogeographers - the objective, empirical description of biological communities - using multivariate techniques of classification and ordination (PIELOU 1984). This work was pioneered by W. T. WILLIAMS (WILLIAMS & LAMBERT 1959) who was the first to use the power of modern computers to analyse and describe the complexity of biological communities.

A Purpose for Complexity - the 'Heroic' Tradition in Ecology

The 1920s saw not only the publication of the integrative but essentially empirical work of ELTON, but also the first formulation of a mathematical population biology that was to underpin the heroic tradition. In this golden age of theoretical ecology, LOTKA and VOLTERRA arrived independently at equations for two-species interactions, paving the way for a more analytical consideration of the role of ecological interactions in the generation of biological complexity. Indeed, both LOTKA and VOLTERRA were after bigger game than populations, and they explicitly couched their results in terms of the dynamics of whole communities (MCINTOSH 1985).

This consideration of the larger entity, the biological community, brought with it a consideration of the purpose of the evident complexity of that entity. Perhaps loosely analogising from the dynamics of single species populations, where genetic diversity seems to confer evolutionary success, there has been a continuing effort to find some evolutionary reason at the level of the community for ecosystem complexity. If the work of LOTKA and VOLTERRA gave the heroic tradition a tool-box with which to analyse communities, it was ELTON (1955) who finally gave it a purpose: stability.

ELTON (1955) argued that communities that were more diverse seemed to be more stable. He based his arguments on his observations of the dramatic oscillations in the populations of the simple communities of the Arctic, the commonness of pest outbreaks in simplified agricultural systems, the apparent absence of pest outbreaks in complex tropical forests, and the ease with which species can invade species-poor islands.

The idea that ecosystem complexity begets ecosystem stability was quickly taken up as the central dogma of the heroic tradition (MACARTHUR 1955; HUTCHINSON 1959) and led to a body of theoretical and analytical work that broadly supported the idea (MACARTHUR & WILSON 1967; WOODWELL & SMITH 1969; CODY & DIAMOND 1975). Even if later work frequently came to the opposite conclusion (MAY 1974), it did little to reduce the power of the idea as an inclusive force in ecological thinking. The idea provided an optimising principle for the simulation modelling work of so-called 'systems ecology' (VAN DYNE 1969; WATT 1973; ULANOWICZ 1979); it provided a link to trophodynamics through a major body of analytical work on food webs (COHEN 1978; PIMM 1982); and it extended itself into a qualitative analysis of the conditions for stability through LEVIN'S (1970) elegant loop analysis. It also provided a springboard for explorations of these dynamics in geometries beyond Euclidean and Riemannian (ANTONELLI et al. 1992, 1993).

Indeed, the observation that complexity seemed to beget instability provided the heroic tradition with perhaps the first look at chaos in any complex system (MAY 1972). This served mainly to deepen the work on the idea, creating the view that the problem itself was complex and needed to be teased out more carefully. PIMM (1984) argued that the components of both complexity and stability needed to be more clearly understood, and that the links between the various components deserved more study. He identified species richness, connectance, interaction strength and evenness as components of complexity; and local stability, global stability, resilience, resistance, and variability as the components of stability.

This work on complexity and stability in ecosystems was advanced mainly through population dynamics rather than trophodynamics; that is, through considering the different abundances of species rather than the energy residing in different species or trophic levels of a community. The study of trophodynamics had begun well with ELTON (1927) and LINDEMAN (1942). It had provided some of the impetus for the first systems ecology (ODUM & ODUM 1955), and it had seemed well placed to provide the ecological salient for a wider attack on the problem of biological complexity using thermodynamics (SCHRÖDINGER 1944; MOROWITZ 1968). However, trophodynamics succumbed to some major theoretical and empirical objections (SLOBODKIN 1969, 1972) and yielded any major role in the heroic tradition to population dynamics. The fundamental objection to trophodynamics, according to SLOBODKIN (1969) in a review Of MOROWITZ'S (1968) book, is that it leads US to commit SCHRÖDINGER's Fundamental Biological Error:

'Essentially, SCHRÖDINGER said that animals consume food because they are thermodynamically open systems which are producing positive entropy and must consume an equal amount of negative entropy to maintain their steady state. In a sense this is equivalent to saying that animals must eat to avoid starving to death. The fallacy arises from the fact that for, say, a normal rabbit population, being eaten is a much more important source of death than starvation.

Part of the food eaten by rabbits is required to maintain the rabbits against thermodynamic decay, but a large part is to keep one jump ahead of the foxes and dogs. The energetic requirements of a population are therefore partially predictable from the thermodynamic properties of the animals but the residual part is predictable only if we know the precise ecological situation of the rabbit population. The actual energy flow patterns in nature are neither simple and direct consequences of thermodynamic theory, nor are they directly predictable from any extant evolutionary theory.'

SLOBODKIN thus encapsulates a key concern of the heroic tradition - the role of evolution in understanding ecological processes (HUTCHINSON 1965) - but tacitly acknowledges that the tradition is really incapable of meeting it. For that, we must turn to our last tradition.

Contingent Complexity - the 'Evolutionary' Tradition

The evolutionary tradition in ecology is the smallest, the most distinctive and the hardest to describe succinctly. Like self-conscious ecology, it accepts the evident complexity of ecosystems as a contingent fact of evolution: it does not agonise over the issue of 'emergent' properties of ecosystems that has so distracted the heroic tradition for many years (SMITH 1975; SALT 1979; HARPER 1982). Like heroic ecology, it looks to other disciplines for inspiration, but unlike that tradition it looks to those disciplines for more than analytical tools: it seeks a broader conceptual framework for itself.

It is tempting to imagine that, because of this quest for a larger conceptual framework, we might be able to trace the origins of the evolutionary tradition to the development of systems ecology during the 1960s. Certainly, this ecology embraced other disciplines, built mathematical models of ecosystems, and predicated itself on an holistic approach. However, the purpose of embracing other disciplines such as chemistry, physics and meteorology, was to include abiotic components in this 'ecology', rather than to find a common conceptual framework; the mathematical models were almost universally sets of differential equations in the empirical tradition of operations research; and the approach was based on the grandly named but essentially trivial 'general systems' philosophy of VON BERTALANFFY (1968).

Instead of springing from the hubris of systems ecology and its transformation into the 'big biology' of the International Biological Program, the evolutionary tradition emerged in a quieter setting: the gathering of a range of concerns about the nature of the theory needed to explain biological complexity in all its forms. Was the theory of evolution a necessary and sufficient theory for all of this evident, evolved biological complexity? At the centre of this work was geneticist C. H. WADDINGTON, who organised a series of symposia at the Villa Serbelloni in Bellagio beginning in 1966 (WADDINGTON 1968, 1969, 1970, 1972). The Serbelloni symposia brought molecular biologists, biochemists, developmental biologists, geneticists, ecologists and evolutionary biologists together with mathematicians, physicists and philosophers. They shared a common view that the complexity of all biological systems was evolved and probably adaptive. They also shared a view that the analytical tools should somehow acknowledge that process of evolution.

These symposia marked a turning point in the consideration of biological complexity: away from a view of living things as being like physico-chemical systems and toward a view of living things as being more like information processing systems; and away from a view of living things existing in some sort of equilibrium or balance and toward a view of a more muddled dynamics. The choice of analytical tools reflected these changes. Although there was much discussion on the use of tools to model the dynamics of populations near equilibrium, other approaches to the analysis of complexity were unveiled there. LEVIN’S (1970) loop analysis, a rigorous but qualitative approach to the dynamics of biological complexity, was first discussed at Serbelloni. Biologists also had their first introduction to THOM’S (1968, 1970) astounding catastrophe theory which, whatever else it is, remains a mathematical tour de force - perhaps the greatest contribution to topology this century - and a compellingly complete biological construct that has tantalised biologists and mathematicians ever since (BRADBURY & ANTONELLI 1985).

The idea that biological systems could be most profitably analysed as if they were information systems, first discussed by MARGALEF (1957), has led to the development of a biological cybernetics where biological systems were treated as if they were automata. The theoretical foundations of this work were laid at Serbelloni using the simplest sorts of examples from genetics. Here the biological system is essentially linear - the genes strung along the chromosome. These one-dimensional systems are such good analogues of a Turing machine processing information along a tape, that it is little wonder the information approach took hold.

The appeal of these analogies was strengthened by work on the linkages between the behaviour of grammars and the behaviour of automata (HOPCROFT & ULLMAN 1969; FU 1974): complexity in language, biology and automata seemed to be alike. Many areas of biology yielded to analyses that were based on the automata analogy: morphology (HOGEWEG & HESPER 1974); psychology (KUMUR 1977); ecology (BRADBURY & LOYA 1978); molecular biology (STEFIK 1978); ethology (HOGEWEG & HESPER 1984). It was further strengthened by the subsequent development of cellular automata (WOLFRAM 1984; TOFFOLI & MARGOLUS 1987), which bring a spatial dimension to the automata formalism. Spatial pattern is an intrinsic part of all biological complexity, but especially so of ecology (SOUTHWOOD 1982): cellular automata provide a way of dealing directly with the role of spatial pattern in ecology (VAN DER LAAN & BRADBURY 1990; VAN DER LAAN & HOGEWEG 1992).

We can see two clear, but perhaps not fully realised, consequences of the work begun at Serbelloni: the creation of a body of theory and tools that, while it is distinctly ecological, sees itself as part of an emerging discipline of complex systems (BRADBURY 1977; ULANOWICZ 1979; BRADBURY & GREEN & REICHELT 1986); and the steady prosecution of the analysis of ecosystems of increasing size and complexity (GREEN & BRADBURY & REICHELT 1987; VAN DER LAAN & BRADBURY 1990). Together these give substance to the argument for a distinct evolutionary tradition in the study of ecology.

Conclusion - a Question of Some Complexity

We argue that, while ecologists see ecosystems as complex systems, they may not be complex systems in the way in which all ecologists wish them to be. Ecologists will easily recognise in ecosystems many of the characteristics of a 'canonical' complex system, as described by KELLY (1992):

• perpetual novelty: the system continues to surprise the observer no matter how long it has been studied;

• resilience: the capacity for self-repair, graceful degradation and limping along instead of dying;

• emergence of the aggregate: a recognition that there exists a 'something' - an object of study - without necessarily recognising that the 'whole is greater than the sum of the parts';

• formation of individuality: each instance of a complex system is recognisable as an example of the system but also has a distinct individuality, which may also lead to an internal identity of self through the generation of internal models;

• internal models: through its architecture, the system can represent a level of abstraction internally, allowing it in a limited way to anticipate the future;

• non-zero sum game: gains somewhere in the system are not necessarily offset by losses elsewhere, nor is the system necessarily in equilibrium;

• exploration tradeoff: processes exist for optimally balancing new ways of surviving (exploration and learning behaviours) against ways of maximising what already exists (exploitation and efficiency behaviours).

However, how far down this list an individual ecologist is prepared to go is likely to serve as a useful diagnosis of the tradition to which that ecologist belongs. Selfconscious ecology is likely to recognise only the first three attributes, while perhaps strenuously denying the later ones. Heroic ecology would be likely to recognise more - perhaps the next two - but it would have difficulty with the last couple. The evolutionary tradition would recognise them all without too much difficulty.

And the evolutionary tradition, while still the smallest, is becoming steadily more assertive. Early works, such as DUNBAR (1972) and WILSON (1980) were tentative heresies from within the heroic canon, while later works used a far broader canvas than ecology (Buss 1987) or even biology (KAUFFMAN 1993). This progressive reaching out' to other branches of science is mirrored in those other branches themselves. ANDERSON (1991), for example, observes a similar process within physics, a process that seeks understanding rather than integration:

'A movement is underway toward joining together into a general subject all the various ideas about ways new properties emerge. We call this subject the science of complexity. Within this topic, ideas equal in depth and interest to those in physics come from some of the other sciences. This movement is overdue and healthy. On the other hand, one may well be apprehensive - or at least I am - that such an enterprise may go the way of General Semantics, General Systems Theory and other well-meant but premature and intellectually lightweight attempts at building an overall structure. We complexity enthusiasts (perish the thought that we be called complexity scientists!) are talking, at least for the most part, about specific, testable schemes and specific mechanisms and concepts. Occasionally we find that these schemes and concepts bridge subjects, but if we value our integrity, we do not attempt to force the integration.' (ANDERSON 1991: 11)

We feel that this intellectual fountain, rather than heroic ecology, will quench the thirst of the evolutionary tradition from now on. We also believe that evolutionary ecology, comfortable with the operational view that ecosystems are like complex systems, will be the tradition that will discover within ecosystems a richer understanding of all complex systems.

References