Fig. 1: Population declines, increases and fluctuations over time occur across all latitudes and biomes within the freshwater, marine and terrestrial realms.
From: Rare and common vertebrates span a wide spectrum of population trends
Results include 9286 populations from 2084 species. The lack of biogeographic patterning in vertebrate population trends was also apparent on a UK scale (Supplementary Fig. 13 and Supplementary Table 3). The numbers in the legend for d–g and on the x axis in c show the sample sizes for realms and biomes, respectively. The μ values of population trend a, b, d, e, h and the σ2 values of population fluctuation c, f–g are from state-space models of changes in abundance over time for each population. d and f show the distribution of population trends across realms including raw values (points) and boxplots (including the mean, first and third quartiles and boxplot whiskers that cover 1.5 times the interquartile range). e, g and h show the effect sizes (centre of error bars) and the 95% credible intervals of population trends e, h across realms and biomes, and fluctuations across realms g. For variation in fluctuations across biomes, see Supplementary Fig. 8. The three estimates in e and h refer to different analytical approaches: population trends calculated using linear models (circles), state-space models (μ, triangles) and population trends (μ) weighted by τ2, the observation error estimate from the state-space models (squares). The five estimates in g refer to different analytical approaches, where the response variables in the models were: (1) the standard error around the slope estimates of the linear models of abundance versus year (circles), (2) half of the 95% confidence interval around the μ value of population change (triangles), (3) half of the 95% confidence interval around μ weighted by τ2, (full squares), (4) the process noise (σ2) from the state-space models and (5) the standard deviation of the raw data for each population time series (empty squares). The process noise is the total variance around the population trend minus the variance attributed to observation error. See Supplementary Table 2 for model outputs. Icon credits: tree by FayraLovers, wave by Setyo Ari Wibowo, mountain and stream by Nikita Kozin.
