Published as Chapter 52 of The Way: An Ecological Worldview, originally published in 1992. This text is taken from the revised and enlarged edition, University of Georgia Press, Athens, Georgia, 1998.
“The stability of complex continental ecosystems was no armour against the Japanese beetle, the European gypsy moth or the Oriental chestnut blight Edothia Parasitica in North America. It is trivial but not irrelevant to observe that stability was hardly enhanced by the extra links added to the trophic web in these instances.”
Robert M. May
“The term ‘complex’ (as used by modern ecologists) need imply no more than a haphazard conglomeration, whereas in the living system, we find distinctive orderliness of the complexes.”
Until recently ecologists have tended, along with the famous British ecologist Charles Elton, to assume that as a system becomes more complex, it also becomes more stable – though Eugene Odum considers that complexity tapers off some time before a climax, the most stable state, is achieved. Today, however, this link between complexity and stability is denied by most mainstream ecologists.
One of the books that has played an important role in changing this attitude, is R. M. May’s Stability and Complexity in Model Ecosystems.  Underlying it is the assumption that a system’s complexity can be measured by the number of its constituent parts without considering the way they are organised, let alone their wider role within the Gaian hierarchy. In other words, it does not distinguish between a system’s natural or homeotelic parts and its unnatural or heterotelic parts.
This means that because a highly destructive ecological invasion, such as the introduction of the rabbit into Australia or the walking catfish into Florida, increases the number of the constituent parts of ecosystems in these countries, it is seen as increasing their complexity. The fact that it also tends to reduce their health, integrity and stability can be construed as confirming May’s preposterous thesis that complexity reduces stability.
In reality, of course, all it does is confirm the indubitable principle that to introduce into an ecosystem a living thing that has been designed by its evolution to fulfil a very different role as a constituent of a very different ecosystem, can only increase the former’s randomness and reduce its organised or real ‘complexity’.
Prigogine also insists that increased ‘complexity’ is associated with growing instability. This for him is confirmed by the fact that the world is becoming ever more complex, yet, at the same time, increasingly unstable. It is this instability that is reflected in the ever-growing fluctuations – floods, droughts, epidemics and wars – that are everywhere on the increase. In his particular scheme of things, however, such fluctuations are highly desirable because they give rise to economic development or progress. 
The ‘complexity’ that Prigogine refers to, is again of the random kind. It offers no basis for distinguishing between the biospheric complexity required to maintain the integrity and stability of the real world and the technospheric complexity created by economic development, which is heterotelic to the biosphere, necessarily disrupting its critical order and stability.
If neither Prigogine nor May can handle organised complexity it is because, among other things, it is extremely difficult to quantify and hence to model. Not surprisingly, the model that May builds to prove that complexity reduces stability bears little relationship to the real world. Among other things, the underlying assumptions are totally unrealistic. Thus May admits that his model only applies to systems with an even number of species, a “disquieting” thought he agrees, but not one that leads him to question its intrinsic value because, for his purposes,
“whether or not the Lotka-Volterra equations (on which the model is based) are applicable to real-world situations is beside the point being made here, which is that simple mathematical models are, in general, less stable than the corresponding simple mathematical models with few species.” 
In other words, he is not concerned with the relationships between complexity and stability in the real world, but only in his mathematical model. In the real world, May admits (though only as an afterthought) that things may be different:
“Natural ecosystems, whether structurally complex or simple, are the product of a long history of co-evolution of their constituent plants and animals. It is at least plausible that such intricate evolutionary processes have, in effect, brought about those relatively tiny and mathematically atypical regions of parameter space which endow the system with long term stability.” 
However, as May himself states, such an ecosystem is “mathematically atypical” and hence, he intimates, of little relevance to a mathematical model; and to him, it is this that is important. 
In the real world, complexity must first be distinguished from diversity. A natural system can increase the number of its constituent parts so as better to achieve two different and indeed rival strategies. The first is to seek to fulfil adaptive functions with greater accuracy, thereby increasing its homeostasis, within a given environment. The second is to increase the number of environmental challenges it can deal with adaptively.
The first strategy is the most adaptive in an orderly and predictable environment; the second, in a disorderly and unpredictable environment. The first strategy has been referred to by the German biologist B. Rensch as “anagenesis”, the second as “cladogenesis” – terms that have been adopted by a number of theoretical biologists such as Julian Huxley and Waddington.
Natural systems that organise their differentiated parts so as best to achieve the first strategy, we can refer to as displaying complexity. Those that organise their differentiated parts so as best to achieve the second strategy, we can refer to as displaying diversity.
Let us first consider the former. Complexity is organised and purposive. The different parts of a complex system do not come into being by accretion but by differentiation. Differentiation implies integration. For the constituent parts of a natural system to be mobilised to act in conjunction with each other so as to achieve a particular adaptive function, they must be differentiated and hence integrated.
Thus in a simple ecosystem, for instance, different species of herbivores are relatively undifferentiated eaters. A mountain goat must be able to eat practically anything if it is to survive in its inhospitable habitat, so much so that a slight increase in the population of other species is sufficient to cause a shortage of the basic foodstuffs they have in common.
Complex ecosystems, however, are far less vulnerable to such discontinuities. Impala and eland in the African savannah, for instance, have a much more specific diet. They will not only eat different plants but often different parts of the same plants. This means that they make the best use of their environment, which will thereby support a greater diversity of herbivores without being degraded. An increase in the number of predator species must also further differentiate, and hence further refine, the quantitative and qualitative controls applied on prey populations. In these and a host of other ways, a system becomes more stable as it becomes more complex.
There is a cost however: the greater the specialisation and the commitment to specific environmental conditions, the smaller the range of possible responses that a system is capable of mediating and hence the fewer the environmental challenges to which it can respond adaptively (unless cybernismic complexity is increased). This means that the more highly integrated a complex ecosystem, the less capable it is of tolerating improbable internal or external challenges.
Thus a tropical rainforest is a highly integrated ecosystem judged by the standard of other ecosystems (though clearly it displays nothing like the degree of integration of a biological organism). For this reason it does not withstand improbable external disturbances. Cut down its trees, for instance, and it does not readily recover, whereas a simpler and less integrated system such as a savannah can recover from similar treatment much more readily. The human organism, being still more highly integrated, is correspondingly more vulnerable to a disturbance. Deprive it of the organs that ensure essential metabolic functions, such as the liver or the kidneys, and it too will fail to recover.
Natural systems can only function adaptively (maintain their homeostasis) within specific conditions. In the case of complex integrated systems, these conditions are often very specialised. For this reason, certain ecologists have maintained that such systems are non-persistent. But this is not so, for on the basis of a very long experience such living things can predict, usually with justification, that these specialised conditions will be maintained.
What is more, why such conditions have been maintained can very easily be explained. A system, together with its environment, constitutes a larger system. This larger system, by maintaining its own stability, assures the orderliness of the environment to which its sub-systems are submitted and hence the stability of their relationship with their environment. Thus an embryo will tolerate only minor changes to the highly ordered environment or field it requires.
However it cannot, on this account, be regarded as unstable. The requisite orderliness of its environment is assured by virtue of being an integral part of the hierarchy of natural systems whose goal it is to maintain overall stability, which must include, of course, that of its internal environment. Similarly, a child can only flourish in an environment or field displaying a certain measure of order – that of the family.
This does not mean that it is unstable, or non-persistent, as this orderly environment is preserved by the normal behaviour of the various members of the family unit. The family itself is designed to live in an environment or field that also displays a certain measure of order, though less so than that required by its individual members and this is provided by the community and so on, all the way up the hierarchy of the biosphere.
In other words, though a complex integrated system may only be able to adapt to a limited range of environmental conditions, changes that are outside this range are unlikely to occur because of the orderly and predictable environmental conditions provided by the hierarchy of natural systems of which it is part. In other words, it displays high resistance stability.
A less complex and less integrated system may be able to adapt to a wider range of environmental conditions, but it is not insulated in the same way from these conditions, which are correspondingly more likely to occur. For this reason it must be more capable of dealing with them; displaying the requisite level of resilience stability. In both cases, the system’s capacity to deal with change in the environment to which it has been adapted by its evolution, seems commensurate with the probability of the occurrence of such change. It clearly must be if the system is to maintain its stability and hence survive.
|1.||Robert M. May, Stability and Complexity in Model Ecosystems; p.39. Princeton University Press, Princeton NJ, 1974.|
|2.||Ilya Prigogine and Isabelle Stengers, La Nouvelle Alliance; p.178. Gallimard, Paris, 1979.|
|3.||May, ibid.; p.52.|
|4.||May, ibid.; p.49.|
|5.||May, ibid.; p.173.|