Table 1.
Key terms.
Fig 1.
Global body size biomass spectrum.
A. Median carbon biomass (log scale) per log size as a function of body size with 95% confidence bounds (black dotted curves) cumulated across biological groups from 1000 bootstraps over within-group biomass and body size error distributions. Groups were organized from the least massive at the bottom to the most massive at the top for visibility on the log scale (ordered from top left to bottom right in color legend for group identity). Group biomasses are stacked so each group’s biomass is represented by its upper y-axis location minus its lower y-axis location (not by the upper y-axis location alone). See Tables 2–4 for within-group biomass uncertainties, and S3 Table; for icon sources. B. Median biomass in linear biomass scale. Confidence bounds are not shown here because they are so large as to obscure the median patterns on the linear scale.
Table 2.
Terrestrial body sizes and biomasses.
Table 3.
Marine body sizes.
Table 4.
Subterranean consumer body sizes.
Fig 2.
Body size biomass spectra within groups.
Thick black curve is the median log biomass, and black dotted curves are 95% confidence bounds from 1000 resamples from within-group size and biomass uncertainties. Groups are organized from lowest to highest biomass (A to AJ). For reference, the thin grey curve is the median cumulative log biomass of all groups.
Fig 3.
Body size biomass spectra by habitat realms.
See Fig 1 caption for description. A. Terrestrial. B. Marine. Subterranean prokaryotes are excluded. Thin grey curves are the median cumulative log biomass of the global biome.
Fig 4.
Rows represent habitat realms (A to E: All realms, F to J: Terrestrial, K to O: Marine). Columns represent regression model types: (A, F, K: Linear, B, G, L: Gaussian, C, H, M: Gaussian mixture 2, D, I, N: Gaussian mixture 3, E, J, O: Gaussian mixture 4). Grey curves represent 95% confidence intervals of the data, and blue curves represent 95% confidence intervals of the model from 1000 bootstraps. For linear models, regression slopes are mean power exponents ± standard deviations across bootstraps. R2 and AICc scores are means ± standard deviations across 1000 bootstraps.
Table 5.
Size-biomass power law exponents across realms and assumptions.
Assumptions correspond to sensitive analyses plotted in S1 Fig. Exponents and R2 result from 1000 bootstrapped linear regressions of log biomass on log size.