Abstract
In the face of rapid global change and an uncertain fate for biodiversity, it is vital to quantify trends in wild populations. These trends are typically estimated from abundance time series for suites of species across large geographic and temporal scales. Such data implicitly contain phylogenetic, spatial, and temporal structure which, if not properly accounted for, may obscure the true magnitude and direction of biodiversity change. Here, using a novel statistical framework to simultaneously account for all three of these structures, we show that the majority of current abundance trends estimates among 10 high-profile datasets, representing millions of abundance observations, are likely unreliable or incorrect. Our new approach suggests that previous models are too simplistic, incorrectly estimating global abundance trends and often dramatically underestimating uncertainty, an aspect that is critical when translating global assessments into policy outcomes. Further, our approach also results in substantial improvements in abundance forecasting accuracy. Whilst our results do not improve the outlook for biodiversity, our framework does allow us to make more robust estimates of global wildlife abundance trends, which is critical for developing policy to protect our biosphere.
Competing Interest Statement
The authors have declared no competing interest.
Footnotes
t.f.johnson{at}sheffield.ac.uk, a.beckerman{at}sheffield.ac.uk, d.childs{at}sheffield.ac.uk, christopher.griffiths{at}slu.se, pcapdevila.pc{at}gmail.com, c.clements{at}bristol.ac.uk, marc.besson{at}obs-banyuls.fr, richard.gregory{at}rspb.org.uk, dele12{at}uqo.ca, gavin.thomas{at}sheffield.ac.uk, karl.evans{at}sheffield.ac.uk, t.j.webb{at}sheffield.ac.uk, r.freckleton{at}sheffield.ac.uk
Funding This work is funded by UKRI-NERC Grant NE/T003502/1
Data and code availability All code will be made publicly available upon acceptance. All data is publicly available; we will provide instructions on data access upon acceptance.