Sealion Projects

The 73%, one dot per population

Every monitored wildlife population as its own dot, so you can see the crowd the Living Planet Index average comes from.

dataviz · desktop-firstData as of 1 Nov 2026Published 1 Nov 2026wildlifeliving-planet-indexdataaverages

Interactive explainer

Sources and method

The question

The 2026 Living Planet Index reports a 73% average decline in monitored wildlife populations since 1970, and it is routinely retold as "73% of wild animals are gone". It is not a headcount. It is an average of rates of change across tens of thousands of monitored populations, some collapsing, many stable or growing. People wonder because the number is scary and the wording keeps slipping, so this piece shows the crowd the average comes from, one dot each.

The data

The Living Planet Database public release of 2016 (ZSL / WWF, via figshare, CC BY 4.0) was downloaded on 2026-11-01. Of 11,067 rows, 11,047 populations with at least two counts were kept. For each: gaps between counts were filled log-linearly, zeros received an offset of 1% of the series mean, and the mean of year-to-year log10 changes became its rate, stored to four decimals with species, class, region and first and last year. The gold line is the unweighted mean of those logs, converted to a percentage a year. The 2026 database itself requires a data-use agreement, so 2026 figures appear only as cited context.

The form

Every population is a dot on one axis of change, so the reader sees the crowd, not a bar. A bar of the index cannot show that the average (about −2% a year here) sits far from where the median population sits (about zero), nor that 48% fall while 49% rise. The crowd splits by region or group on demand, and the "set aside steepest" control removes the most extreme collapses, so the gold line visibly moves. Search places an animal the reader knows among them.

What it leaves out

This recomputation is unweighted and approximate: the official LPI weights by species richness and uses a different, larger release, so the 2016 data's average here should not be read as the 2026 figure. A few very short series with near-total drops dominate the mean, which is itself a finding but also a method artefact. Monitored populations are not a random sample of wildlife, and rates beyond ±10% a year are drawn in an edge strip.

Sources

A focused prototype from Sealion Projects. Anything it saves stays in this browser, on this device.