My bridge to studying ecosystems started once I shifted to combine the functional and numerical response equations with others concerning other processes in order to make a population model, of interacting predator and prey. That is when, suddenly and unexpectedly, multi-stable states appeared. Lovely indeed. Great fun and a big surprise to me! A new landscape for exploration opened.
Non-linear forms of the functional responses (e.g. the Type 3 S-shaped response) and of reproduction responses (e.g. the Allee effect) interacted to create two stable equilibria for interacting populations, with an enclosed stability domain around one of them. It was the responses at low densities that were critical- that is where vertebrate predators have yet to learn to locate the prey easily, and where mates are too scarce to find each other easily. Once discovered, it seemed obvious that conditions for multi-stable states were inevitable. And that, being inevitable, there were huge consequences for theory and for practice.
Up to that time, a concentration on a single equilibrium and assumptions of global stability had made ecology, as well as economics, focus on near equilibrium behavior, and on fixed carrying capacity with a goal of minimizing variability. Command and control was the policy for managing fish, fowl, trees, herds, and freedom was unlimited to provide opportunity for people.
The multi-stable state reality, in contrast, opened an entirely different direction that focused on behavior far from equilibrium and on stability boundaries. High variability, not low variability, became an attribute necessary to maintain existence and learning. Surprise and inherent unpredictability was the inevitable consequence for ecological systems. Data and understanding at low densities, rare because they are all the more difficult to obtain, were more important than those at high-density. I used the word resilience to represent this latter kind of stability
Hence the useful measure of resilience was the size of stability domains, or, more meaningfully, the amount of disturbance a system can take before its controls shift to another set of variables and relationships that dominate another stability region. And the relevant focus is not on constancy but on variability. Not on statistically easy collection and analysis of data but statistically difficult and unfamiliar ones. That needs a different eye to see and a different theory to perceive consequences.
About that time, I was invited to write a 1973 review article for the Annual Review of Ecology and Systematics. I therefore decided to turn it into a review of the two different ways of perceiving stability and in so doing highlight the significance for theory and for practice. That required finding additional rare field data in the literature that demonstrated flips of populations from one level or state to another, as well as describing the recently discovered known non-linearities in the processes that caused or inhibited the phenomenon. That was a big job and I recall days when I thought it was all bunk, and days when I believed it was all real. I finished the paper on a “good” day, when all seemed pretty clear. By then I guess I was convinced. The causal, process evidence was excellent, though the field evidence concerning population flips, was only suggestive. Nevertheless the consequences for theory and management were enormous. It implied that uncertainty was inevitable. And that ecosystems, in an evolutionary time span, were momentary entities pausing in a flip to different states. As I’ll describe, it took about 30 years to confirm those conclusions for others.
This paper began to influence fields outside population/community ecology a bit – anthropology, political science, systems science first, then, later, ecosystem science. It became the theoretical foundation for active adaptive ecosystem management. But it was largely ignored or opposed by practitioners in the central body of ecology. What followed was the typical and necessary skepticism released by new ideas, that I’ll describe briefly here because it is such a common foundation for developing science.
One early ecological response to the paper was by Sousa and Connell (1985). They asked the good question “was there empirical evidence for multi-stable states?”. They attempted to answer by analyzing published data on time series of population changes of organisms to see if the variance suggested multi-stable behavior. They found no such evidence. This so reinforced the dominant population ecology single equilibrium paradigm, that the resilience concept was stopped dead, in that area of science.
It seemed to be an example of evidence that refuted this new theory. But their evidence was inappropriate and the theory was not! In fact, their evidence, as is often the case, was really a model, incomplete because the collators unconsciously used an inappropriate model for choosing data that were incomplete.
There are two problems with their analysis:
- They did not ask any process question (are there common non-linear mechanisms that can produce the behavior?). That is where the good new hard evidence that I had discovered lay.
- They rightly saw the need for long time series data on populations that had high resolution. As population/community ecologists of tradition, however, their view of time was a human view- decades were seen as being long. That view is reinforced by a “quadrat” mentality. Not only small in time, but small in spatial scale; and a theory limited to linear interactions between individuals in single species populations or between two species populations, all functioning at the same speed (e.g. predator/prey, competitors). It represents the dangers caused by inferring that “microcosm” thought and experiments have anything to contribute to the multiscale functioning of ecosystems. Steve Carpenter has a perceptive critique of that tendency (Carpenter, 1996).
The multi-stable behavior can only be interpreted within the context of at least three but, as suggested in the Panarchy paper/chapter, probably not more than five variables. These variables need to differ qualitatively in speed from each other. It is therefore inherently ecosystemic. It is the slow variables that determine how many years of data are needed for their kind of test. None of their examples had anywhere near the duration of temporal data needed.
As an example: The available 45 years of budworm population changes they analyzed seemed long to Sousa and Connell and to all those conditioned by single variable behavior and linear thinking of the times. But the relevant time scale for the multi-equilibrium behavior of budworm is set by their hosts, the trees or the slow variable. What is needed for their tests was yearly budworm data (the fast variable) over several generations of trees (the slow variable), i.e. perhaps one and a half centuries – not 45 years. The normal boom and bust cycle is 40-60 years
It has since taken 25 years of study of different ecosystems to develop data for appropriate tests. Examples include those using paleo-ecological data covering centuries at high resolution, the deep and shallow lake studies and experiments of Carpenter (Carpenter 2000) in the United States and of Marten Scheffer, in Europe (Scheffer et al. 1993), the experimental manipulations of mammalian predator and prey systems in Australia and Africa by Tony Sinclair (Sinclair et al. 1990), and a variety of studies of specific ecosystems- sea urchin, coral reef etc. Terry Hughes and his colleagues’ works on coral reefs stand out as examples. Carpenter’s important summary paper makes the point (Carpenter, 2000). Multi-stable states are real and of great importance, although they are difficult to demonstrate. Surprise, uncertainty and unpredictability are the inevitable result. Command and control management temporarily hides the costs, but the ultimate cost of surprises produced by managing systems that ignore multi-stable properties is too great. Active adaptive management is the only alternative management response possible. Steve Carpenter and Buz (W.A.) Brock – a great ecosystems scientist together with a wonderful ”non-linear” economist- show why in a classic paper where a minimal model of a watershed, farming styles, of regional monitoring and regional decision regarding phosphate control, encounter the surprises created as a consequence of a multi-stable state (Carpenter, Brock, and Hanson, 1999).
References:
Carpenter, Stephen R. 1996. Microcosm experiments have limited relevance for community and ecosystem ecology. Ecology 77 (3) : 677-690.
Carpenter, S.R. 2000. Alternate states of ecosystems. Evidence and its implications for environmental decisions. In, M.C.Press, N.Huntley and S. Levin. (eds). Ecology: Achievement and Challenge, Blackwell, London.
Carpenter, S.R., Brock, W.A., Hanson, P.C., 1999. Ecological and social dynamics in simple models of ecosystem management. Conservation Ecology 3(2), 4. URL: http://www.consecol.org/vol3/iss2/art4
Scheffer, M., S.H. Hopsper, M-L. Meijer, B.Moss and E. Jeppesen. 1993. Alternative equilibria in shallow lakes. Trends in Ecol. & Evol. 8 (8): 275- 279.
Sinclair, A.R.E. , P.D. Olsen, and T.D. Redhead. Can predators regulate small mammal populations? Evidence from mouse outbreaks in Australia. Oikos 59: 382-392.
Sousa, W.P. and J.H. Connell. 1985. Further comments on the evidence for multiple stable points in natural communities. American Naturalist 125, 612-615.
