5 hours ago
13
The Amazon rainforest, according to a 2021 study, is losing its capacity as a carbon sink and now emits more than it absorbs. In the tropics, marine scientists are reporting that coral reefs are in decline, threatening fish stocks. Equally concerning is research into the Atlantic Meridional Overturning Circulation (Amoc), a vast system of ocean currents that helps regulate the climate and is at risk of collapsing this century. The entire global ecosystem appears to be losing its ability to function.
We find this view in newspapers, magazines, technical reports and the journals of learned societies. But thinking about the environment in terms of its functions is also how many of us tend to understand the world. We may think that forests exist to produce oxygen, wetlands to filter water and bees to pollinate our crops.
There is a problem with this way of thinking: ecosystems don’t exist to perform goals. The Amazon absorbs carbon, but it doesn’t “aim” to do so. It simply exists. Any standards of operation we find in nature have come directly from our own desires for things such as climate stability, abundant fisheries, beauty or cultural meaning.
So why do we keep thinking ecosystems have functions they could fail to perform?
I came to this puzzle as a graduate student in the late 1990s, a time when research into biodiversity and ecosystem function was rapidly increasing. Initially, I thought I would write my dissertation on a conventional ecological topic: whether species richness drives productivity. Instead, I fell in with the philosophy of science crowd, attended their seminars and eventually earned a master’s degree in philosophy alongside my work in ecology. There I encountered a rich debate over the concept of function – what it means, when it applies, what work it does. But no one seemed to be connecting that debate to the way ecologists were using the same word, unreflectively, to describe what ecosystems do. This essay is an attempt to bring those conversations together.
My concern with ecosystems and function was never just academic. I am an environmentalist, unsettled by the loss of natural places. And as a father, I am concerned that my generation will leave to our children a planet depleted in both richness and resilience. These commitments also drive my interest in debates about function. If the way we think about ecological crisis is conceptually shaky, we risk obscuring what’s really at stake.
I worry that the ways we often conceive of the problems before us are inadequate. For if ecosystems have no intrinsic ends and cannot truly “break down”, how do we repair them? How do we respond to environmental crises in a world of aimless ecosystems?
Approaches to conservation have long been shaped by debates about whether nature has a purpose or whether we are projecting our own aims on to it. Behind every attempt to justify new protections lies an implicit answer to the question: what is the environment for?
In the United States and the United Kingdom during the 19th century, these answers were rooted in game laws and hunting traditions that sought to maintain populations of species valued for sport or resource use. By the mid-20th century, the American forester and early conservationist Aldo Leopold offered a more expanded answer by proposing that our moral community should include “the land” itself: soils, waters, plants and animals.
In the 1970s and 80s, the answers of conservationists were increasingly grounded in the intrinsic value of specific species, reflected in legislation such as the US Endangered Species Act. But a decade later, the species-focused approach of “conservation biology” was seen by many as lacking. It targeted only rare organisms that contributed little to the circulation of their ecosystems – species such as the spotted owl and the snail darter fish. Some researchers worried that the species approach might have overlooked more consequential concerns, including the major “services” provided by ecosystems, such as food production, clean water, drought mitigation, storm protection, timber and fibre.
In the late 1990s, this crisis led to a new research agenda, which crystallised around “biodiversity and ecosystem function” (BEF). This approach presented itself as a scientifically rigorous framework while serving as a rhetorically powerful justification for conservation. In contrast to a hyper-focus on individual populations of rare species, BEF embraced all biodiversity.
In the early decades of the 21st century, this logic scaled up, underpinning UN projects and intergovernmental science policy. National governments began commissioning natural-capital accounts, attempting to assign monetary value to pollination, flood regulation, carbon storage and other ecological phenomena. The answer to the question “What is nature for?” had become this: nature is for the services it provides to people. The language of ecosystem function was the conceptual bridge that made this answer sound scientific rather than merely political.
As a result, the idea of function now pervades how ecosystems are described and understood. Consider for a moment how you think about the ecosystems around you. If you have ever described a forest as a carbon sink or a wetland as a natural filter, you have inherited the ethic of BEF. If you’ve ever thought of a rainforest as something that provides oxygen for humans, or a reef as something that helps furnish us with protein (in the form of fish), you’ve inherited the logic of “ecosystem services”.
What do we mean when we use the word “function”? Sometimes, it refers to designed purposes. For example, when we say that the function of a clock is to tell the time, or the function of a carburettor is to mix air and fuel for combustion. In these cases, the object (or one of its parts) was intentionally made for a specific end. The same logic applies up a hierarchy of wholes and parts: the carburettor is part of the engine, the engine part of the car, the car part of a transport system.
Other kinds of functions arise through co-option rather than design. Writing at a picnic table, I might use a book or a rock to keep my papers from blowing away. The rock was not designed and the book was intended for another purpose, yet both can serve the goal I have in mind. I give them their function by using them in a certain way.
Still other functions emerge without any intention, particularly in nature. The philosopher Karen Neander offers a striking example: penguins are myopic on land. Their eyes are not defective but optimised for underwater focus, where penguins feed. Land myopia is a byproduct of a visual system shaped for a different environment.
Though there are several ways that “function” is used, there are two main theories that guide (and justify) the ways scientists typically think about it: causal role theory and selected effects theory.
Robert Cummins developed causal role theory in response to Ernest Nagel’s argument in The Structure of Science (1961) about how science should avoid teleological language. Nagel proposed that scientists should not explain things in a way that suggests the influence of specific goals or purposes.
For example, rather than saying: “The function of the lungs is to oxygenate the blood”, Nagel might say: “Given the structure of lung tissue, the properties of gases and the pressure differences during breathing, oxygen diffuses into the bloodstream and carbon dioxide diffuses out.” This becomes a scientific explanation based on laws and initial conditions.
Cummins, however, thought this missed how scientists actually think about function. He saw that references to function could be a useful explanatory shortcut when talking about how things work, and so proposed a different approach. According to Cummins, ascribing function to anything is simply a way of identifying a component’s contribution to the “capacity” of the system that contains it. Functional language, from this view, is fine. For example, the carburettor in a car enables the engine to convert chemical energy to mechanical energy; the engine enables the car to transport passengers; and so on.
It is easy to see why this theory would be attractive to ecologists who are typically interested in tracing causal chains. The function of bacteria and other decomposers, in their view, is to break down dead organisms into smaller particles and transform their chemical composition; the function of green plants is to convert carbon dioxide into bioavailable carbon for herbivores. From this perspective, everything exists for something else.
However, Cummins’s causal role theory has some serious limitations. First, it provides no real way of determining which processes count as genuine capacities. The capacities we select depend on what phenomena scientists happen to be interested in, rather than those that are objectively important to the system. The philosopher Ruth Millikan illustrates the difficulty this way: the heart pumps blood, but it also makes a thumping noise. Doctors may use this noise diagnostically, yet they do not treat it as a function of the heart. Why not? In the causal role theory, there is no way to distinguish genuine functions from incidental effects.
Another limitation is that causal role theory cannot account for how something could malfunction. As the philosopher Ema Sullivan-Bissett explores in her 2016 essay Malfunction Defended, any adequate theory of function must be able to explain how biological items can fail to do what they are supposed to do. Though the causal role theory can explain that a heart with a defective valve is still doing something (moving blood, albeit inefficiently), it cannot say that the heart is doing its job badly. It offers no way of describing what the standard for doing a good job is supposed to be.
The alternative to causal role theory, and probably the dominant theory among philosophers of biology today, is the selected effects theory, developed by Larry Wright along with Neander and Millikan. In this view, to say that a trait has a function is to give an account of its history, identifying the cause for which it exists and persists. According to this theory, any biological function is the effect for which the trait was selected in the process of natural selection. It’s likely that you have understood the world in this way, too. You may understand that the function of the heart is to pump blood because pumping blood was the reason proto-hearts were favoured by animals in the evolutionary past. This historical anchoring distinguishes selected effects explanations from causal role accounts, which focus only on present-day contributions and not on how the trait came to be.
This theory matters because it provides scientists with a standard against which something can succeed or fail. If a trait has a function grounded in evolutionary history, then it can malfunction when it fails to do what that history selected it to do. The question is whether ecosystems can also have this kind of standard.
As we’ve seen, “function” doesn’t mean the same thing in all cases. We can distinguish between two broad uses of thw word. The first sense is descriptive: explaining how a system works. The other is goal-directed (or teleological): it specifies what a system is for (and how it can fail). This distinction becomes particularly important when we turn to rainforests, coral reefs and other systems that have effects we can describe but no ends that we can point to – and without ends they’re meant to achieve, the idea that an ecosystem can “malfunction” begins to unravel.
In the early 20th century, the ecologist Frederic Clements proposed that ecosystems develop through predictable stages of succession toward a stable “climax” community, much as an organism grows and matures. Other ecologists even used the metaphor of a “superorganism”, implying that ecosystems had an intrinsic trajectory and a kind of unified purpose. While influential for decades, this view has long since been abandoned.
Nowadays, ecologists think that ecosystems, for the most part, are not like organisms at all. Ecosystems are not shaped by selection, they do not reproduce, and it’s debatable whether they are even identifiable biological entities (in contrast to hearts, say, or cell receptors). Instead, they are open, dynamic systems composed of countless interactions among organisms and their local microenvironments – contingent combinations of organisms that we identify and name primarily for the purposes of our understanding. If you haphazardly throw together a bunch of organisms in a place, you have an ecosystem.
And yet, ecologists continue to borrow the language of function to describe ecosystem-level processes. Wetlands “function” to filter surface water; forests “function” as carbon sinks.
The establishment of the journal Functional Ecology in the 1980s marked one moment in this conceptual evolution. Articles in this journal began investigating how individual species within ecosystems used their “functional traits” to influence major ecological processes. Consider the ways that vultures scavenge animal carcasses. For the vulture, scavenging provides sustenance. At the level of the ecosystem, however, this same behaviour can be described differently: thinking in terms of “trait-based ecology”, scavenging becomes just one process of many by which organic matter is broken down. That is, it contributes to the large-scale processes usually defined by ecologists as “ecosystem functions”, including nutrient cycling, primary production and decomposition. By describing the behaviour of vultures this way, ecologists turn a goal-directed function in the organism into a contribution to the ecosystem.
Once species are assigned roles in this way, they begin to resemble carburettors in an engine or organs in a body. This is where the language becomes unstable.
From the perspective of function, descriptions of how biodiversity shapes ecological processes can start to merge with judgments about what those processes are for, and whether they are being sustained or lost. For example, a decline in insect populations can be described as a change in pollination rates but also recast as a loss of the ecosystem’s “ability” to support crops. Likewise, reduced microbial activity in soils can be described as leading to slower decomposition but also framed as a failure of the system to maintain soil fertility.
The distinction between describing how something happens and making normative judgments about what the resulting processes are for is one that matters if we want to think clearly about what’s taking place when ecosystems change. When these two are not kept apart, the idea of “ecosystem function” begins to carry more weight than it can support.
What about the standard justifications for using functional language? For ecosystem processes, the account offered by the selected effects theory will not work. First, ecosystems are not shaped by natural selection as cohesive units. A forest such as the Amazon may often be described as “the lungs of our planet”, but it has nothing in common with human organs, or any other cohesive unit shaped by natural selection. Rainforests, like all ecosystems, don’t have selected effects. They do not reproduce. Their boundaries are often impermanent. It is debatable whether they are even identifiable biological entities.
Plants fix carbon, microbes decompose organic matter and forest animals redistribute nutrients. These processes can be described straightforwardly. But it is so easy to take the further step and say the rainforest is for storing carbon or maintaining stability. At that point, a description of what happens begins to look like a claim about what the system is meant to do. Any such claims are necessarily anthropocentric.
And so, if we say an ecosystem is malfunctioning, we must also ask: malfunctioning for whom, and for what purpose? These questions reveal the assumptions embedded in our language and highlight the risks of conflating ecological processes with human-centred goals.
Were ecologists aware of the deeper connotations of the language they were using to describe ecosystems? The answer is yes. I asked Peter Calow, the founding co-editor of Functional Ecology, how the journal got its name and whether he had misgivings about applying “function” to ecosystems. He told me he was “comfortable with the notion of function applying to adaptation within species through natural selection” – but “less comfortable with it being applied to ecosystems”. The British Ecological Society’s publications committee, which oversees the journal, debated the matter at length before, in Calow’s words, “getting tired of discussing it” and adopting the title. (He recalled that “functional” terminology was not an unthinking carry-over: it was chosen despite conceptual unease and largely because the kinds of papers the journal was seeking to publish connected ecology with physiological research, where functional concepts were well entrenched and largely understood through the selected-effects account.)
Another place to look is the landmark book Biodiversity and Ecosystem Function (1993), based on a 1991 symposium in Germany and supported in part by Unesco’s Man and the Biosphere programme – tellingly gendered, and unabashedly anthropocentric. Both the sponsorship and the volume itself bear out this orientation. In the foreword to the book, the late ecologist Paul Ehrlich justifies its intellectual premise:
Of special interest to humanity is the relationship of biodiversity to the variety of services provided by ecosystems and, in particular, to the stability of the flow of those services, such as the maintenance of the gaseous composition of the atmosphere, preservation of soils, recycling of nutrients and provision of food from the sea.
He then revisits the “rivet popper” analogy, which he had previously introduced in the environmental classic Extinction (1981), co-authored with Anne Ehrlich. They described each species in an ecosystem as a rivet in an aeroplane wing: remove one rivet and the plane will fly on, but remove enough rivets and the plane will fail, typically catastrophically. The presupposition is that “failure” matters because the aeroplane’s value lies in safely transporting people. The metaphor is rhetorically powerful but imperfect. Rivets are static, fully interchangeable and single-purpose; species are dynamic, unique and exhibit a vast diversity of behaviours that shift across contexts. Importantly, rivets were placed by design engineers. Ehrlich’s analogy smuggles in the idea that ecosystems, like machines, have a proper configuration, and that deviation constitutes malfunction.
During the past few decades, this kind of metaphorical scaffolding has done important political work. Framing biodiversity loss as akin to losing rivets from an aeroplane wing makes the stakes vivid for policymakers and the public. It can also harmonise neatly with the “ecosystem services” agenda, which links ecological science directly to human welfare. In this policy context, “ecosystem function” becomes a conceptual hinge: it can be presented as a purely scientific measure of ecological processes, while simultaneously serving as a proxy for the benefits those processes deliver to people. That duality made the term powerful but also ensured that the teleological and value-laden connotations scientists worried about in private would persist in public discourse.
What shall we do with the notion of ecological function? From my perspective, ecosystems can only malfunction when they are appropriated or co-opted. Just as I might select a stone to serve as a paperweight, a wetland may be designated as a water filtration system, in which case a disruption in its ability to filter water is correctly seen as a malfunction. Similarly, if a forest is managed for carbon sequestration, a decline in its carbon storage capacity should be considered a failure. In these cases, the notion of malfunction arises not from the ecosystem’s intrinsic properties but from its role in meeting human-defined goals.
“Malfunctions” reflect human values and priorities by framing nature’s worth in terms of utility, aesthetics or cultural and spiritual value. Examples of undesirable ecological events such as algal blooms, coral bleaching and deforestation illustrate the complexity of these judgments. An algal bloom caused by fertiliser flowing into the ocean from rivers might disrupt aquatic ecosystems, yet whether that disruption counts as a “malfunction” or a “natural” response to nutrient inputs depends on the standard we apply. Coral bleaching may be described as a failure of reefs to support marine life, but this framing reflects human concerns about biodiversity or fisheries production rather than intrinsic purpose. These cases underscore that our reasons for repairing ecosystems rest on human ideas – such as duties, norms and objectives – that are external to the ecosystems themselves. So how can we think about ecosystems, and our obligations to them, more clearly?
To more fully move beyond teleology in their descriptions of the world, ecologists could focus simply on characterising the interactions in an ecosystem and quantifying changes of state, without any reference to purposes or goals. Such an approach respects the autonomy of the nonhuman world without imposing human values and priorities. But conceptually moving beyond teleology doesn’t stop us viewing ecosystems through the lens of our duties, norms and objectives. Even when scientists engage in apparently objective research, human values always come along for the ride.
This point can be sharpened by turning to the philosophy of science. In The Empirical Stance (2002), Bas van Fraassen argues that empiricism – the view that the world is known through observation and experience – is not a doctrine about what exists, but a stance. It is a set of attitudes and commitments about how to conduct inquiry. The same is true of what is sometimes called “value-free science” – the ideal of describing the world independently of the inquirer’s perspective. To adopt that ideal is itself a choice, shaped by values about what counts as knowledge and what is worth knowing. It is a commitment, not a discovery. When ecologists study ecosystems, they cannot escape the values that guide their attention.
I am not saying we should purge those values. Understanding the ways that we are bound to our values is an invitation to examine honestly how they enter into scientific practice. Likewise, recognising that value-free science is a myth does not weaken the case for environmental action. It clarifies that the task of thinking about ecosystems, and our obligations to them, is both descriptive and normative.
When we say that natural systems exist to provide services for us – oxygen, protein, climate stability – we appropriate certain processes for our own purpose. In doing so, we are actively privileging one ecological process over others. We are not merely observing a function. We may value pollination, for example, for its role in sustaining crop yields while ignoring or even suppressing other equally “natural” processes, such as the consumption of plant matter by pests. When we then perpetuate that chosen process by intervening in an environment, through conservation or technological design, its continued existence is no longer solely the product of natural conditions but also of our deliberate selection. These functions become selected effects: they persist because they are chosen by us in the present, not because they were favoured by natural selection in the past.
Ecosystems cannot malfunction on their own. They may change, reorganise or even collapse. But these should be understood as natural processes, not failures. Teleological framings can be deployed, but only if we are explicit about whose needs are being served, and to what ends. Used in this way, appeals to “function” can make the value of ecosystems legible to human concerns while avoiding the pretence that such purposes belong to nature itself.
What is at stake here is a question of intellectual honesty. Environmental arguments often present these purposes as if they were natural facts, rather than human commitments. When we say an ecosystem is “breaking down”, we risk disguising our own values as properties of the world. That move can be rhetorically effective, but it is conceptually misleading.
By reframing our understanding of ecological functions and malfunctions, we can advance a more rigorous and reflective ecology. We can directly state those reasons when we recognise that our reasons for caring about ecosystems come from us (our needs, our ethics, our futures). In doing so, we arrive at an ecology that joins scientific description with explicit moral responsibility, rather than blurring the two.
The work ahead is not to repair nature’s purposes, but to take responsibility for our own – and for the world they shape.
