In 50 years, the average vertebrate population has shrunk by 73%. But within that collapse, something unexpected is happening — some species are coming back, some ecosystems are healing, and a growing body of evidence suggests we know how to make it happen deliberately. This is that story.


The world is losing wildlife — but not everywhere

The Living Planet Index has tracked vertebrate populations since 1970. The headline number is disappointing - there is a 73% decline on average across monitored species. But hidden within that global average are species that are coming back — animals pulled from the brink by targeted conservation and rewilding efforts. These recoveries don’t make the news. The data below does.

While these recoveries are real, the differences in population scale make some species hard to see — population units vary across species in the Living Planet Database (counts, densities, and indices are all used). Switching to a logarithmic scale reveals the directional trend for all six species - every one of them has grown by more than 150% since monitoring began.

The decline is global — but the gaps between regions tell a deeper story

Not all regions are losing wildlife at the same rate. Some have stabilised. A few show tentative signs of recovery. The Living Planet Index, indexed to 100 in 1970, reveals stark regional divergence — and points to where conservation efforts may be gaining ground.

Europe’s partial recovery is something noteworthy. It coincides with decades of rewilding initiatives which includes reintroducing apex predators, restoring floodplains and allowing farmland to return to nature. The question is no longer whether rewilding works. The question is how fast and across which dimensions of an ecosystem does recovery actually happen.

Rewilding works — but not all dimensions recover equally

Rewilding is not a single action — it is a spectrum of interventions across trophic complexity, natural disturbance, connectivity, and human land use. Data from measured rewilding sites across Europe shows that while some ecological dimensions recover quickly, others take much longer. The heatmap below scores each site across multiple rewilding dimensions at baseline and after intervention.


The ecosystem responds

Species returning is one signal. But does the wider ecosystem actually heal? Data from a controlled rewilding experiment tracks invertebrate communities — the base of most food webs — across three treatment types over 35 months. Rewilded plots show measurable gains in both abundance and species richness compared to unmanaged controls.

Australia is leading the way

While rewilding is gaining momentum globally, Australia has developed one of the world’s most systematic approaches — a network of fenced, predator-free sanctuaries managed by the Australian Wildlife Conservancy. These sites have become lifelines for mammals not seen in the wild for over a century. Explore the map below to see where recovery is happening and which species have been brought back.

References

Living Planet Index Database. (2024). LPD 2024 public data [Data set]. Zoological Society of London. https://www.livingplanetindex.org/data_portal

Torres, A., Fernández, N., Selva, N., Revilla, E., & Delibes, M. (2018). Data from: Measuring rewilding progress [Data set]. Figshare. https://figshare.com/articles/dataset/Data_from_Measuring_rewilding_progress/7133195/1

Baker, N. J., et al. (2023). Invertebrate and detritivore data from rewilding experiment [Data set]. Dryad. https://datadryad.org/dataset/doi:10.5061/dryad.ncjsxkt1r

Ritchie, H., & Roser, M. (2024). Living Planet Index by region [Data set]. Our World in Data. https://ourworldindata.org/grapher/living-planet-index-by-region

Australian Wildlife Conservancy. (2024). Safe havens: Fenced sanctuaries. https://www.australianwildlife.org/sanctuaries/

Lecture notes - MATH2270

Acknowledgements

Visualizations were created in R (v4.3) using the following packages: tidyverse, plotly, leaflet, readxl, and scales. Claude was used for assistance with the structuring and planning of the assignment, and for manually transcripting the data for chart 5 (Sanctuaries in Australia) as there was not a straightforward dataset available for the specific task.