How to read this
The chart shows a modelled Wildlife Population Index starting at 100 in 1970.
Each curve represents the predicted trajectory of vertebrate populations
under the active threat combination — how fast they are expected to decline
(or recover) on average per year from 1970 to 2040.
The dashed line is the no-threat baseline: without human pressures,
vertebrate populations in the dataset trend slightly upward. This is the world
we have interrupted. The shaded band shows the 80% credible interval —
the range of likely outcomes given model uncertainty.
The gold dot near 2025 marks the real-world measured value from the
Living Planet Index (~31), showing how well the multi-threat model tracks reality.
The paradox: common threats ≠ worst threats
Habitat loss and exploitation affect the most populations (27% and 31%
respectively) and dominate global conservation policy. Yet in the Bayesian
model, disease, invasive species, and climate change cause faster per-population
decline when present — despite affecting far fewer populations today.
This counterintuitive finding challenges where conservation funding flows.
The threats attracting the most attention are not the fastest killers.
The rarest threats can be the most lethal.
The no-silver-bullet finding
The model's most important policy implication: removing any single threat
from a multi-threat scenario produces only marginal improvement. Simulations show
that only the simultaneous removal of all threats can meaningfully reverse
predicted declines. Conservation strategies targeting one threat at a time —
the dominant global approach — are insufficient to halt the trajectory.
Interaction types
Additive (~85% of combinations globally): the combined impact
equals the sum of individual effects. Even without amplification, this produces
catastrophic totals when individual slopes are steep.
Antagonistic (~8–22%): threats partially offset each other —
the combined impact is less than the sum of parts. The mechanism is often unclear,
but populations still decline significantly under antagonistic combinations.
Synergistic (rare globally; more common in amphibians and some
terrestrial systems): threats amplify each other beyond their individual effects.
The amphibian disease + invasive species combination is the clearest documented example.
Global context
The 2019 IPBES Global Assessment (Díaz et al.) found approximately
1 million animal and plant species threatened with extinction —
more than at any previous point in human history. Over 75% of Earth's land
surface has been "significantly altered" by human activity. About 66% of the
ocean area experiences increasing cumulative human impacts.
The Living Planet Index (WWF/ZSL) tracks 34,836 populations of 5,495 vertebrate
species. Between 1970 and 2020, monitored vertebrate populations declined by an
average of 69%. Freshwater populations declined the most steeply
(83%), followed by terrestrial (69%) and marine (56%) populations.
Data & method
Population slopes (Δ log abundance per year) are posterior medians from a
multilevel Bayesian linear mixed model fit to 3,129 vertebrate population
time series (25,054 datapoints) from 1,281 species worldwide, spanning
1950–2019 (Living Planet Database). Confidence bands show 80% credible intervals.
The model includes spatial and temporal autocorrelation and phylogenetic
random effects. Threats were classified using six IUCN categories. Only
populations with ≥10 years of monitoring and ≥5 data points were included.
Interaction terms (pairwise and triple) capture non-additive effects.
Sources
Capdevila P., O'Brien D.A., Marconi V., Johnson T.F., Freeman R., McRae L.,
Clements C.F. (2026). Halting predicted vertebrate declines requires tackling
multiple drivers of biodiversity loss.
Science Advances, 12, eadx7973.
doi:10.1126/sciadv.adx7973
Díaz S. et al. (2019). Pervasive human-driven decline of life on Earth
points to the need for transformative change.
Science, 366, eaax3100.
doi:10.1126/science.aax3100
Living Planet Report 2022. World Wildlife Fund & Zoological Society of London.
Population time series data: Living Planet Database (LPD 2024 public release).