Ecology and Climate: Feedbacks Decide the Outcome

Two ideas do most of the work in understanding environmental systems, and neither is complicated.

The first is that these systems are networks of dependencies, so a change in one place propagates in ways that are difficult to predict from the parts alone.

The second is that they contain feedbacks — responses that either amplify the original change or absorb it. Which type dominates matters more than the size of the initial push, and it is the main reason small changes sometimes produce large shifts while large ones sometimes produce almost nothing.

Both are examples of feedback loops and emergence operating in a physical system, which is why the same reasoning transfers to economies and organisations.

Why removing one thing rarely stays local

The clearest demonstration in ecology is what happens when a predator is removed from a system.

The immediate effect is obvious: prey numbers rise. The subsequent effects are not. More grazing animals means heavier pressure on vegetation. Certain plants decline. The animals that depended on those plants decline. Soil held by the roots erodes differently. In documented cases, river channels have shifted as a consequence, because vegetation along the banks changed.

None of that is predictable from a list of the species involved. It emerges from the network of dependencies between them.

This produces a practical rule that applies well beyond ecology: interventions in networked systems have effects at a distance from where they were applied. The immediate effect is usually the intended one; the important effects are often two or three steps away, and arrive later.

It also produces the concept of a keystone species — one whose influence on the system is disproportionate to its abundance. These are usually identified only after removal, which is an uncomfortable way to learn about them.

The general caution follows directly from Gall's Law and Chesterton's fence: in a system that evolved into its current arrangement, components frequently serve functions nobody has documented.

What is well established about climate, and what is not

This is an area where the state of knowledge is often misrepresented in both directions, so it is worth being precise about which claims sit where.

Very well established. Certain gases in the atmosphere absorb and re-emit infrared radiation — measurable in a laboratory and known since the nineteenth century. Concentrations of these gases have risen substantially since industrialisation, and the isotopic signature identifies the source as fossil carbon. Global average temperatures have risen over the same period. The basic physical mechanism connecting these is not seriously disputed among people who work on it.

Well established but with real ranges. How much warming results from a given increase in concentration — climate sensitivity — is known within a range rather than precisely, because it depends on feedbacks. Water vapour amplifies warming; clouds do complicated things that remain the largest single source of uncertainty.

Genuinely uncertain. Regional effects, timing of specific changes, and the behaviour of potential tipping points. Whether particular thresholds exist, and where, is an active research question rather than a settled fact. Confident specific predictions about a given region in a given decade are going beyond what the models support.

Being clear about that middle and lower tier matters. Overstating certainty is a poor foundation for policy and gets corrected in public, which damages the credibility of the well-established parts.

The honest summary is that the mechanism and direction are solid, the magnitude has a real range, and the local detail is uncertain — which is enough to act on, since decisions under uncertainty are normal rather than exceptional.

Why some environmental problems got solved

A useful exercise is comparing environmental problems that were substantially addressed with those that were not, because the difference is structural rather than moral.

Problems that got solved — ozone-depleting chemicals, leaded fuel, acid rain, several river and air pollution issues — shared features. A small number of identifiable emitters. Substitutes that already existed or could be developed quickly. Costs falling on industries rather than on everyone's daily behaviour. And usually a visible, attributable harm.

Problems that remain hard have the opposite shape. Emissions come from billions of sources including ordinary heating, transport, and food production. Substitutes exist for some uses and not others. The harm is diffuse, delayed, and attributable to no one in particular.

That last feature makes it a tragedy of the commons at global scale: each emitter captures the full benefit of their activity and bears a negligible share of the cost, and unilateral restraint is genuinely costly to the restrainer while barely affecting the outcome.

Understanding this reframes what interventions can work. Appeals to individual behaviour struggle against that arithmetic — not because people are selfish, but because the structure punishes unilateral action. What has historically worked is changing the arithmetic: making the cost fall on whoever creates it, or making the substitute cheaper than the thing being replaced.

The second of those has been doing more than expected. Where clean alternatives became cheaper than the incumbent on plain cost grounds, adoption followed without requiring anyone to accept a sacrifice — which is a considerably more reliable mechanism than persuasion, and the main reason for cautious optimism in an otherwise difficult problem.