Biodiversity: A Concept Without a Referent

A portmanteau that cannot do the work we ask of it

“Sub-Saharan Africa has a current estimated biodiversity intactness of 76% (±14%).”¹

The grammar of loss here is familiar and powerful. In this case a quantity called “biodiversity intactness” is down by 24% from some baseline; there must therefore be some thing whose intactness can, in principle, be preserved or degraded. The sentence therefore invites us to treat biodiversity as a conserved quantity: something that can be diminished, offset, stabilised, or restored.

This kind of language is entirely familiar. We are repeatedly told that biodiversity loss must be halted, reversed, or offset. These claims presuppose that “biodiversity” names a single, trackable object, whose magnitude can be measured, compared, and defended over time. But the living world does not present us with anything like this.

What we call biodiversity ranges over many distinct biological features and processes, operating at different organisational levels and governed by different dynamics, each requiring its own forms of description and explanation. There is no unified thing here that could be pointed to, measured, or lost in the way that organisms, populations, traits, abundances, ranges, or lineages can be.

This essay develops a cumulative argument. I begin by showing that “biodiversity” lacks identity conditions and cannot function as a single object. I then trace the methodological consequences of that indeterminacy, first by clarifying why pluralism is not the issue, and then by unfolding the resulting failures of measurement, scale, and composability. I go on to show how these failures propagate into policy through institutional uptake and operationalisation, before becoming explicit in targets, trajectories, and compensatory schemes. I end by reframing the question of what, in practice, is being governed and on what grounds.

1. Identity Failure

Species richness, evenness, total abundance, phylogenetic diversity, functional diversity, genetic variation, and compositional turnover all capture different aspects of biological systems. They are not interchangeable. They are not necessarily correlated with one another, and when they are, they are sometimes inversely related. There is no non-arbitrary way to present them as a single object without first making motivated decisions about weighting, scale, taxonomic scope, and level of organisation. The world does not dictate those choices.

“Biodiversity” does not consistently refer to the same kinds of things from one context to the next. It has no stable extension.

Nor is there any natural unit, scale, or level of organisation at which biodiversity exists. It has no determinate metric.

There are only indices, each emerging from specific methodological and normative decisions. The term slides indifferently between non-equivalent measures that track different things and vary independently, as if they were perspectives on a single object.

There is no coherent basis for treating losses of one biodiversity surrogate as being offset by gains of another; the loss of a species, for example, is not offset by increased genetic diversity in those remaining. Nor is there any coherent basis for treating losses in one ecological system as compensated by gains elsewhere.

This indeterminacy is not accidental. In introducing the term for the 1986 National Forum on BioDiversity, W. G. Rosen explained that he chose it to signal that discussion would not be focused on a single topic but range over many biological and ecological concerns and concepts.

We should not therefore expect, and we do not discover, a principled answer to the question “Do these two systems have the same biodiversity?” unless we specify which surrogate we mean (species richness, genetic diversity, extinction risk, etc.), at what scale, and for which taxa. Two systems can be identical with respect to one measure and radically different with respect to another. “Sameness” depends entirely on stipulation, because the term “biodiversity” provides no criterion for deciding which differences count.

The problem persists through time and across scales. If a system loses species but gains functional diversity, has its biodiversity decreased, increased, or stayed the same? The concept supplies no rule for tracking identity through change: for example, it provides no criterion for determining what would count as the persistence or transformation of “biodiversity” in a process of succession.

The same failure appears across spatial scales. A landscape may gain species while its component plots lose them, yet there is no identity-preserving relation linking the “biodiversity” of plots to that of landscapes. Ecological explanations of turnover do not repair this gap. The term slides freely from plots to landscapes to continents to the planet, even though diversity measures at different scales are neither additive nor comparable.

In short, the term has no clear identity conditions.

2. Pluralism Is Not the Problem

This is not simply because biodiversity is heterogeneous. Many useful scientific concepts tolerate internal diversity. Species, for example, is understood in different ways across biological contexts, and gene has carried multiple meanings across molecular, developmental, and evolutionary biology. In each case, however, particular usages supply local and purpose-bound criteria of sameness and difference. Boundaries are fuzzy and lineage continuity is imperfect, but disagreement concerns which criteria to privilege, not whether there is any principled basis for treating disparate measures as tracking the same thing.

Abstraction and pluralism are not themselves problematic.

“Biodiversity” differs in kind. Its internal diversity is not managed by fixing identity within contexts, but by leaving identity indeterminate altogether, allowing non-equivalent surrogates and scales to be treated as if they referred to a single conserved quantity. The problem is that “biodiversity” permits unconstrained substitution across measures, scales, and levels of organisation while continuing to function rhetorically as if it named a single quantity. Nothing in the concept limits which differences may be traded against which others, or on what grounds.

This argument does not speak against the protection of living systems, organisms, or ecological processes; it concerns the conceptual role played by “biodiversity” in attempts to justify and organise that protection.

3. Methodological and Normative Slippage

At times “biodiversity” functions descriptively, as a summary of biological variation. Elsewhere it stands in for processes such as extinction, turnover, or homogenisation. In policy contexts it often serves as a proxy for normative judgements about “intactness”, “health”, or “degradation”. These are categorically different kinds of claims. The umbrella term allows them to be treated as continuous, muddling transitions between description, process, and value.

As a result, claims about biodiversity change are invariably index-relative, scale-dependent, and value-laden. To say “we’ve lost 24% of biodiversity” does not report a fact about the world. But it does present the output of a convention-dependent index as if it were that kind of fact. The number itself and its apparent precision are artefacts of methodological and normative choices, not a property of the system being described.

Policy documents provide clear illustrations of this malleability.

SDG 15 opens with the injunction to “halt biodiversity loss”, as if biodiversity named well-defined global quantity. A few lines later, that loss is operationalised using the Red List Index, which is a summary of aggregate extinction risk across species. Elsewhere, biodiversity is invoked to designate places (“important sites for biodiversity”), but what makes a site “important” remains undefined because we do not know which aspect of biological variation should be prioritised. In Target 15.6 it becomes genetic resources and benefit-sharing; in Target 15.9 it acquires “values” to be integrated into national accounting; and in SDG 14 it appears as a contributor to development, food security, and livelihoods. Apparent continuity of purpose (“do something about this good”) is produced by repeated changes of referent.

This plasticity is enabled by the foundational policy definition itself. The Convention on Biological Diversity defines biodiversity as “the variability among living organisms”, including diversity within species, between species, and of ecosystems. “Variability” is not a single property but a family of statistical descriptors, and the definition specifies neither what is to vary, nor how it is to be measured, nor at what scale. It explicitly spans organisational levels that are not composable: genetic variation, species richness, and ecosystem diversity are not parts of a single quantity, share no property that would make comparison meaningful, and do not sum to anything coherent.

Embedded in institutional practice, this conceptual permissiveness shapes what is measured, funded, and defended.

4. Institutional Uptake and Inferential Failure

The definition supplies no identity conditions. We are no wiser after reading it about what would count as more or less biodiversity, how to determine that two systems have the same biodiversity, or how trade-offs across levels might be handled. It does not specify what “biodiversity” is a summary of, or what can and cannot be inferred from it. It therefore fails to fix a stable extension of the term.

The definition succeeds institutionally by leaving the referent open: its indeterminacy makes it portable across conservation biology, economics, policy, and ethics. But in no coherent sense does “biodiversity” summarise biological, economic, policy, or ethical facts.

The subsequent equivocations are unavoidable. “Biodiversity” means extinction risk in one context, protected-area coverage in another, genetic resources in a third, and something like ecological “health” or “intactness” in a fourth. The term functions as a portmanteau, jumbling together incommensurate analytical and normative ideas while seeming to name a single thing. Because “biodiversity” includes ecosystems, and by standard ecological definition ecosystems include their inanimate substrates, then the boundary of what counts as “biodiversity” thins out gradually, until almost anything of interest can be brought within it.

Terms such as biodiversity, ecosystem services, and natural capital typically originate in science as broad, heuristic constructs. Policy and funding agencies take them up rapidly because their conceptual flexibility allows decision-makers within those institutions to align heterogeneous aims. These terms, together with policy-coined terms such as nature-based solutions, all of them lacking conceptual discipline, are fed back with institutional authority into scientific discourse in funding calls, indicators, and reporting frameworks.

Careers, resources, and legitimacy come to depend on fluency in the sanctioned vocabulary. Conceptual looseness proves useful, and persists. A large literature grows around the terms, refining methods and applications while leaving the underlying concepts largely untouched. The referent may be elusive, but semantic discipline is a luxury the system does not reward. Questioning meaning becomes at best tedious. The label stabilises. We know what biodiversity means; let’s not waste time reopening settled questions. Or perhaps, We know the term is ill-defined; better not to reopen a debate everyone knows cannot be resolved.

That biodiversity is operationally effective as a coordinating label does not resolve this difficulty. Heuristic success does not repair failures of reference. A construct may facilitate measurement, coordination, or communication while remaining incapable of bearing the inferential and justificatory weight placed upon it.

In practice, biodiversity does not determine what must be measured, how different measures relate to one another, or what would count as success or failure. Nor does it genuinely structure reporting frameworks, targets, or funding priorities. Instead, it functions as a label under which disparate indicators and targets are assembled without being integrated. In that limited procedural sense, it may be described as operationally effective.

This indeterminacy may conceivably have enabled the CBD. Precisely because “biodiversity” did not fix its referent, it allowed actors with divergent scientific commitments, political interests, and normative priorities to converge on a shared instrument of negotiation. In that limited procedural sense, the term proved operationally effective. But the same feature that made agreement possible leaves the concept unable to support the inferential and justificatory roles it is later asked to play. Much the same institutional function might have been served by the “Convention on Biota”, using a domain label that enables agreement by deferring specification. But unlike “biodiversity”, “biota” does not so readily masquerade as a single measurable quantity. Its indeterminacy would therefore express itself less as spurious aggregation than as initially unresolved scope and selection.

5. Targets, Trajectories, and Spurious Quantities

Beyond any operational effectiveness, the difficulty with “biodiversity” arises when the term is treated as if it named an object whose magnitude could be measured, tracked, aggregated, and defended. Ambitions such as the commitment to “halt biodiversity loss by 2011”, or to “ensure that Europe’s biodiversity will be on the path to recovery by 2030”, presuppose a single measurable quantity whose trajectory could, in principle, be stabilised or placed on a path.

Yet “biodiversity” itself provides no basis for determining whether such ambitions have been met. Success or failure can only ever be diagnosed relative to particular surrogates and scales. Outputs of one or another index are then presented as facts about the world; losses measured under a given surrogate are taken to be losses of a single conserved quantity; and policy language of halting, reversing, offsetting, or compensating presupposes persistence through change.

I started this essay with a quote referring to a “biodiversity intactness” index. This index, or rather its name, provides a useful illustration of how the term “biodiversity” functions in practice. The index does not attempt to measure biodiversity in any general sense. It quantifies changes in the estimated average abundance of a narrow subset of indigenous taxa relative to a modelled baseline. That this quantity is intelligible is not in question (though I doubt that it measures anything ontological). But calling it an index of “biodiversity intactness” allows its output to be read as a statement about biodiversity as such, rather than about one particular surrogate, at one scale, under one set of assumptions. The name supplies rhetorical continuity where conceptual continuity is lacking.

The Convention on Biological Diversity does not, in its implementation machinery, treat “biodiversity” as a single measurable quantity. Recent decisions, including those agreed at COP16 (Rome, February 2025), link new funding commitments to expanded requirements for planning, monitoring, reporting, and review, implemented through sets of indicators. Progress is specified, compared, and defended by surrogate measures tracking different features at different scales. This allows claims of success, failure, and equivalence to be grounded in different, principled but non-commensurable measures.

The result of using “biodiversity” as if it were a thing one can measure is not empirical error so much as inferential misalignment: the term is asked to do justificatory work that, by its construction, it cannot do. From there, it is a short step to treating biodiversity as something whose protection requires justification.

6. Why “Biodiversity” Cannot Justify Conservation

The cost of this conceptual permissiveness is that it makes it impossible to answer, in any clear or stable way, the question “Why should we conserve biodiversity?”

That question assumes there is some clearly defined thing to conserve. But in some ways “biodiversity” resembles the Oort cloud: a diffuse halo of non-equivalent measures, organisational levels, and normative roles, weakly held together by a name. There is no single thing to conserve unless it is Gaia, perhaps, or life on Earth, or the extended human habitat. Different assumptions about the question then tacitly select different surrogates, scales, and values. All we consistently require of the answer is that it maintains the appearance that we are defending a single, unified good.

If biodiversity means species richness, then conservation might be justified by preventing extinctions. If it means genetic variation, the justification shifts, perhaps, to preserving evolutionary potential. If it means ecosystem diversity, the argument could perhaps turn on maintaining landscape heterogeneity or system-level processes. If it refers to functional diversity, the concern could relate to those slippery ideas of “resilience” or “service provision”. If it stands for intactness or health, the justification would be openly normative. These may all be defensible aims, but they are not interchangeable, and they may conflict.

Because the term conceals these switches, conservation arguments routinely slide between rationales. A loss is measured using one index, explained using a different process, and perhaps even justified using a third set of values. What looks like a single argument for conserving biodiversity is in fact a bundle of partially overlapping arguments for conserving particular features, processes, or outcomes. The umbrella term suppresses the need to say which of these is doing the justificatory work.

Trade-offs between levels and measures cannot be articulated honestly if the object of conservation is left indeterminate. Policies that increase species richness may reduce genetic diversity; actions that preserve ecosystem structure may accelerate turnover at the population level. Without a clear account of which aspect of biological variation matters, and why, such conflicts are treated as technical problems rather than substantive disagreements about values and priorities.

Humans have never aimed to conserve biological diversity in the round. Much “biodiversity” has always been treated as something to be reduced, excluded, or eliminated. Humans clear forests, eradicate predators, suppress pathogens, control weeds and pests. “Invasive species”, “antibiotic resistance”, and “crop pests” are forms of biodiversity that conflict with human purposes. When we ask why biodiversity should be conserved, we already have in mind a filtered subset of biological variation that we judge to be desirable.

By treating biodiversity as a single good, conservation language masks the fact that conservation is always about choosing which forms of biological variation are to be promoted, tolerated, suppressed, or destroyed. Those choices are irreducibly normative. They depend on judgements about risk, utility, justice, livelihoods, historical responsibility, and future options. They cannot be derived from biodiversity as such, because biodiversity includes the very things we routinely fight to remove.

The result is a form of moral and analytical laundering. Normative commitments about what ought to be preserved, for whom, and at what cost are smuggled into policy under the guise of a single, quasi-natural quantity called “biodiversity”. The rhetorical power of the term allows disparate motivations to be aligned without being examined, and disagreement to be postponed unresolved.

7. What, Then, Are We Governing?

The concern here is not whether living systems should be protected, but how appeals to “biodiversity” are used to justify and organise that protection. As a collectivising banner, the term has clear advantages: it allows disparate concerns to be gathered under a shared heading, facilitates coordination across institutions and disciplines, and lowers the threshold for political engagement.

But the same indeterminacy that makes this possible also creates its central danger. By masking heterogeneity and suppressing conflict, “biodiversity” invites distinct aims to be treated as if they were expressions of a single quantity, encouraging substitution, aggregation, and trade-off where none are conceptually warranted. What enables coordination also enables confusion about what is being protected, for whom, and on what grounds.

Indeed, because “biodiversity” admits such a wide range of interpretations, decisions can in practice be taken without close attention to the living world at all. The term can be satisfied procedurally through indicators, accounts, or designated priorities, without any particular engagement with organisms, populations, or ecological processes. We see this happening with “ecosystem services”. This term refers to the ways humans extract benefits from, and assign value to, their surroundings. The ecological substrate is almost incidental; the service might be delivered by technology.

This does not show that conservation lacks justification. It shows that “biodiversity” itself cannot be the justification. Abandoning it as a single justificatory object does not preclude coordination. It requires replacing a fictive unity with explicitly plural, context-bound conservation aims. Any defensible answer to “Why conserve biodiversity?” must therefore abandon the umbrella term and specify what is actually being conserved, for whom, and why those things matter.

The question, then, may be a different one: if not biodiversity, what exactly are we governing, and how do we justify the actions taken on its behalf without losing coherence or legitimacy?

1. Clements, H.S., Biggs, R., De Vos, A. et al. A place-based assessment of biodiversity intactness in sub-Saharan Africa. Nature 649, 113–121 (2026). https://doi.org/10.1038/s41586-025-09781-7