Extended Data Fig. 4: Larger population fluctuations cause less precise estimates of the Living Planet Index (LPI) in nonlinear population trajectories.
From: Random population fluctuations bias the Living Planet Index

a and b, Simulated sets of populations each with 500 species that increase from 100 to 160 individuals along concave-up, linear and concave-down trajectories with low (a) and high (b) population fluctuations. c and d, The accompanying trends in LPI for increasing populations with low (c) and high (d) fluctuations; the confidence intervals around the LPI are negligible because the starting (solid horizontal grey lines) and ending (dashed horizontal grey lines) populations are identical in all sets. e and f, The LPI in the final year of the simulation, 2020. Here, coloured vertical lines correspond to the LPI from the simulated data (c and d) and the dashed black line is the true value based on the actual final populations. The distribution is the density of LPI values from a null model that approximated linear declines by randomly reshuffling the order of population changes (100 times), while keeping the starting and end values constant.