Skip to main content
. 2019 Nov 14;179(5):1084–1097.e21. doi: 10.1016/j.cell.2019.10.008

Figure 4.

Figure 4

Projected Changes in Shannon Diversities by the End of the 21st Century

(A) Projected changes by the end of the 21st century relative to the beginning of the century (percent) for MPGs accounting for GAM models with high explained deviance (>60%). Projections were based on SST and chl a data simulated by the CMIP5 models and the GAMs (n = 13,000 for each combination of MPG and time frame; STAR Methods; Tables S1F–S1H; see Figure S12 for SD by grid cell). Copepods, photosynthetic protists, parasitic protists, and endophotosymbiont diversity (Shannon index) was modeled based on 18S rRNA gene metabarcoding data, size fraction 0.8–2000 μm, and diversity of heterotrophic and photosynthetic bacteria on 16S rRNA gene metagenomics data (size fraction 0.22–3 μm), all from the surface layer. Predicted Shannon values of 0 or less obtained at high latitudes, particularly for copepods and endophotosymbionts, were excluded.

(B) Latitudinal averages of values in (A) and their uncertainties. For visualization purposes, average anomalies for endophotosymbionts and copepods were drawn up to latitudes where values remain below 100%, and all plots show the averaged SD reduced by half. The x axis is not fixed. The last three panels refer to latitudinal averages of particulate organic carbon (POC) export at 100 m (Henson et al., 2012), the number of grid cells with a high marine fisheries catch (>200 kg km−2 year−1) (Watson, 2017), and marine protected area (MPA) latitude kernel density plots (Bruno et al., 2018; STAR Methods; Table S1I).