The Living Planet Index: Does It Rest on a Statistical Bias?

The Living Planet Index, the reference figure of the COPs, is built on a structural statistical flaw

The figure has entered the collective unconscious of climate negotiators: minus 73% of wild vertebrate populations between 1970 and 2020. Minus 85% for freshwater species. These data, produced by the WWF in its Living Planet report, have punctuated every biodiversity COP for years, fed government arguments, justified financial commitments, and served as the basis for reports from the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES). A study published in 2024 in Nature raises an uncomfortable question: what if these figures, without being false in their intent, are constructed on a structural mathematical flaw?

This is not an attack on the idea that biodiversity is in crisis. It is an examination of how we measure this crisis, and what we risk when the measuring instrument is defective.

The essentials

The WWF’s Living Planet Index records a 73% decline in wild vertebrate populations between 1970 and 2020, and 85% for freshwater species.
A study published in Nature in 2024 documents a systematic mathematical bias: a few populations in extreme collapse weigh disproportionately on the global average, producing an overestimation of actual decline.
Other independent indicators confirm a serious decline in biodiversity, but far less brutal than what the LPI suggests.
The issue is not whether biodiversity is doing well, but building indicators robust enough to withstand political counterattacks and guide effective policies.

An indicator becomes the official truth of biodiversity

The Living Planet Index (LPI) has existed since 1998. The WWF produces it in partnership with the Zoological Society of London, aggregating thousands of time series on wild vertebrate populations across the world. In 2024, the database covered more than 34,000 populations representing approximately 5,500 species. The ambition is legitimate: to create a global thermometer of the state of life, comparable from year to year.

This thermometer has a precise method. For each monitored population, the relative variation is calculated between a starting year and the current year. These variations are then aggregated into a geometric mean, first by species, then by taxonomic group, then by region, before producing a global index. The result: a single figure, easily communicable, that claims to sum up the state of wildlife on the planet.

This communicability is precisely what has made it successful. Negotiators at COP15 biodiversity in Montreal in 2022, who adopted the Kunming-Montreal framework aimed at protecting 30% of land and oceans by 2030, had the figure in mind. IPBES reports cite it. Environmental ministries in dozens of countries use it in their official communications. A single figure that sums up everything—that’s the ideal tool to convince busy decision-makers.

The problem is that single figures that sum up everything are often the ones that lie most cleanly.

The mathematical flaw at the heart of the calculation

The study published in Nature in 2024 identifies a structural problem in the LPI’s aggregation method. To understand what is at stake, we need to look at what a geometric mean does when applied to wild population data.

Let’s take a simplified example. Imagine ten vertebrate populations monitored over fifty years. Nine of them have stable or slightly declining trajectories, say between -10% and -20% over the period. The tenth has suffered a catastrophic collapse: a dried-up lake, a razed forest, a locally extinct endemic species. Its decline reaches 98%. The geometric mean of these ten populations will produce a figure dominated by this extreme case. Result: a global index that says “vertebrate populations declined by 60%,” when nine out of ten populations lost between 10 and 20%.

This is not a calculation error. It is a known mathematical property of geometric means applied to asymmetric data. The LPI aggregates thousands of populations, some of which have trajectories of extreme collapse. These few cases mechanically weigh on the overall result in a disproportionate manner. The Nature study documents that this bias is systematic and significant in the WWF database.

This bias has a technical name: it is linked to the asymmetry of the distribution of rates of change. When a population can theoretically lose 100% of its individuals, but can gain only a few hundred percent, the distribution is structurally asymmetric. A geometric mean on such a distribution produces an index that does not reflect the median experience, but the experience of the most extreme cases.

The authors of the study do not conclude that biodiversity is doing well. They conclude that the LPI as it is currently calculated overestimates the magnitude of average decline. The difference is not trivial.

What other indicators really say

It would be tempting to conclude from this methodological critique that the alarm about biodiversity is exaggerated. This would be committing the opposite error: invalidating the diagnosis because a thermometer is incorrectly calibrated.

Other indicators, constructed with different methods, converge on a finding of serious decline. The Biodiversity Intactness Index, developed by researchers at Newcastle University and published in Nature, estimates that terrestrial ecosystems have on average lost around 15% of their integrity compared to their natural reference state. This is concerning. It is far from the 73% of the LPI.

The International Union for the Conservation of Nature (IUCN) now classifies 44,000 species as threatened with extinction out of 157,000 assessed in its Red List. This is approximately 28% of assessed species, with overrepresentation of amphibians (41% threatened), corals (33%), and sharks and rays (37%). These figures tell a real crisis, documented species by species, not an extrapolation of aggregated averages.

What the convergence of these indicators reveals is that the biodiversity crisis is certain and serious, but it is also far more complex and geographically heterogeneous than a single global figure suggests. Certain regions and certain taxonomic groups are under intense pressure. Others are holding steady or, in a few documented cases such as Africa’s large mammals after the protection policies of the 1990s, are partially recovering.

Nuance is not a gift to climate skeptics. It is a condition for effective policy: conservation that does not know where to concentrate its efforts because it is guided by a biased global figure is conservation that wastes already insufficient resources.

When a biased indicator structures international negotiations

The political stakes are here. The Kunming-Montreal framework adopted in December 2022, sometimes called “the Paris Agreement of biodiversity,” relies on mobilizing 200 billion dollars per year by 2030 for nature. It commits signatory countries to protecting 30% of land and marine areas by the same deadline. These commitments were negotiated in a context of urgency that the LPI helped construct.

No one suggests that these commitments are disproportionate. The protection of 30% of lands and oceans is defended by solid arguments that do not depend on the LPI. But when activists and delegates wave the figure of minus 73% as ultimate proof of urgency, and when that figure is subsequently contested on serious statistical grounds, the risk is real.

That risk is credibility. Opponents of any conservation policy have every interest in seizing on the methodological flaw to discredit not only the indicator but the entire case. “If their flagship figure is wrong, why believe them on anything else?” The argument is rhetorically powerful even if it is intellectually dishonest. The flaw in the LPI’s calculation method says nothing about the real state of biodiversity. But in a public debate where figures are weapons, a weapon that turns against you is dangerous.

We find a similar dynamic in other scientific fields with high political stakes. Economists who measure inequality know that the choice of indicator (Gini, interdecile ratios, share of the top 1%) produces different narratives about the same reality. Climatologists have long debated the best way to aggregate temperature anomalies to produce a global index. In these debates, methodological robustness is not an academic luxury. It is a political protection.

Scientists working to correct course

The critique published in Nature is not an isolated attack. It is part of a methodological discussion that has existed within the ecology community for several years. Researchers such as those behind the Biodiversity Intactness Index or the Living Planet Index critical circle have published work pointing to the same limitations: sensitivity to extreme values, underrepresentation of certain geographic regions (notably the tropics and Southeast Asia, where data are scarcer), overrepresentation of vertebrates at the expense of invertebrates and plants.

The WWF and the Zoological Society of London themselves have acknowledged some of these limitations in their reports. Recent editions of Living Planet include more developed methodological notes, and the authors recommend not interpreting the LPI as a measure of absolute abundance of wild species, but as a trend indicator. This nuance, which appears in technical annexes, generally disappears in public communication.

Research teams are working on alternative or complementary indicators. The PREDICTS project of the Natural History Museum in London compares species assemblages between impacted zones and reference zones rather than monitoring individual populations. The UNEP Biodiversity Scenarios group develops models that integrate anthropogenic pressure projections with local species inventories. These approaches are slower to produce results and harder to communicate than a single figure, but they are potentially more robust.

The issue is not to replace the LPI overnight. It is to build a dashboard, a set of complementary indicators that overlap and mutually correct each other. What we already do for economic indicators, where no serious decision-maker relies on a single figure, whether GDP or unemployment. What we have not yet managed to impose in public debate on biodiversity, where the LPI reigns alone. The same question arises moreover in other domains of scientific exploration, as shown by the effort to document major discoveries of the deep sea: measuring life is always dependent on the methods we give ourselves to accomplish it.

Statistical rigor and political credibility are the same thing

The debate over the LPI raises a broader question about how science produces figures for public policy. There is a real tension between communicability and precision. A single figure is mobilizing. A dashboard of twelve indicators with confidence intervals is more accurate but difficult to make exist in international negotiations.

This tension cannot be resolved by the goodwill of researchers alone. It requires a conversation between scientists and policymakers about what an indicator can and cannot promise. The limit of GDP as a measure of well-being took decades to enter political minds, and alternatives like the HDI or indicators of national gross happiness remain marginal despite their merits. The same shift must occur for biodiversity.

What is at stake is not merely the precision of a calculation. It is the ability of the conservation movement to maintain its credibility over the long term. The commitments of the Kunming-Montreal framework will be evaluated in 2030. If the LPI remains the reference indicator, and if voices continue to point out its statistical flaws, each assessment risks being transformed into a methodological debate rather than an evaluation of conservation policies.

Conversely, if the scientific community manages to build a robust and recognized dashboard by 2030, it will have a tool that resists counterattacks, that allows distinguishing regions and groups that are progressing from those that are declining, and that guides investments where they are most effective. The 200 billion annually promised in Montreal deserve better than an indicator that does not know how to measure where the money should go.

Biodiversity is in crisis. This point is not in debate. What is in question is our ability to measure this crisis with enough precision to act usefully. Building measurement instruments equal to the challenge: perhaps this is the priority that the critique of the LPI should bring to light.

Sources

Living Planet Index, Wikipedia — https://en.wikipedia.org/wiki/Living_Planet_Index
WWF, Living Planet Report 2024 — https://www.wwf.org.uk/sites/default/files/2024-10/living-planet-report-2024.pdf
IUCN, Red List of Threatened Species — https://www.iucnredlist.org
IPBES, Global Assessment of Biodiversity and Ecosystem Services (2019) — https://ipbes.net/global-assessment