HomeOur World in Data

Living Planet Index: what does an average decline of 68% really mean?

An average isn’t helpful – we need to know where and what animal populations are in greatest danger. Some are doing much worse than the headline sugge

An average isn’t helpful – we need to know where and what animal populations are in greatest danger. Some are doing much worse than the headline suggests, but they get lost in the averages. This means we can’t focus our efforts on protecting the species that need it the most.  

We should also be careful about what the average actually tells us. It can, quite easily, be sensitive to outliers: populations that have seen a dramatic decline or increase. 

Let’s look at a simple example of how this might work: how would a population experiencing an extreme decline be reflected in an averaged result across several populations? Say we have five lizard populations. One of these populations has experienced a terrible 99% decline in numbers. The other four have not changed at all – they have just as many individuals as they did decades ago. When we calculate the average change across these five populations we get a 20% decline.3 

This would be the headline we see in the news. But actually this result is pretty useless. It does not tell us that most lizard populations are doing fine – we probably think they are shrinking. And worse, we are blind to the great peril of the one lizard population that is quickly vanishing. It has declined by 99%, but we only see the final result of a bad, but more moderate, decline of 20%. These averages don’t allow us to prioritize our conservation efforts where they’re most needed.

There are other hypotheticals that make this clear: let’s say we had an ecosystem where one population saw a decline of 99%, and 393 other populations each increased by 1%. Our final result would report an average decline of 50%.4 This is the difference between what we’d call a ‘catastrophic decline’ (a decline across most or all species) versus a ‘cluster decline’ (where it is a very specific set of species that are struggling). Our approach to tackling either of these scenarios would be very different.

This is not just borne out in hypothetical examples. It’s reflected in the LPI results. In a study published in Nature, Brian Leung and colleagues looked at how the LPI was affected by extreme declines (or increases) in a small subset of the studied populations.5 For this study they looked at the results of the 2018 LPI Report – we’ll look at a similar analysis for the 2020 update soon. It reported a 60% average decline in wildlife populations since 1970.6 

By looking at the population data underlying the LPI they found that this 60% average decline was driven by extreme losses in a small subset of populations. If you excluded the 2.4% most-strongly declining populations – which was 356 out of 14,700 – the result reversed from a 60% average decline to a slightly positive growth. In other words, 2% to 3% of populations were doing extremely badly, but it appears that most species were doing okay.

In the chart we see how the 2018 LPI result would have been affected by excluding the most extreme-negative populations. In red we see the final headline result of the report – a 60% average decline across the 14,700 populations. But as we exclude the most extreme negative populations, first 120 then 238 populations, we see that this average decline reduces significantly. Then, when we exclude the 356 most-severe populations, not only does the average decline reduce, it actually turns into a net positive. The abundance across these populations was, on average, increasing.