Fig. 4: Illustration of the degree of trophic amplification of phytoplankton declines.

To explore how NBSS slopes translate into trophic amplification (a) we compared three NBSS-Chl a relationships namely: our best estimate (red line in Fig. 2a), a best-fit positive linear relationship (blue line) and the fixed NBSS value of −1 first proposed¹⁵ (green line). These relationships were used to estimate in b the respective supportable biomasses of fish as a percentage of phytoplankton (see Methods). The colour coded histograms provide an indicative example of the percentage decline in fish biomass (in units of g C m⁻²) that would result from Chl a values reducing from 1 to 0.5 mg Chl a m⁻³, as indicated by the vertical lines on the logarithmic Chl a axis. c CMIP6 outputs of phytoplankton biomass from an average of two Earth System Models (global scale, 1^o grid cells;) were used as input terms to drive our empirical model (red in a, b) to estimate the carrying capacity of fish. Changes in these (1990–1999 to 2090–2099) are related to those of phytoplankton carbon. The red circle denotes the global 7.5% decline in phytoplankton biomass which relates to a 19% decline in supportable fish biomass. Note the clustering of points below the red 1:1 line, showing negative trophic amplification. d Contemporary (1990–1999) reference model ensemble distribution mapped⁶⁶ for phytoplankton carbon. e Percentage change in phytoplankton carbon between the 2090–2099 period and the 1990–1999 reference. f 1990–1999 illustrative estimate of supportable biomass of fish, calculated as above (see also Methods), g change in estimated supportable biomass of fish (1990–1999 to 2090–2099). Large areas of bright red close to low- and mid-latitude continents show substantial declines in currently important fishing areas. See Supplementary Fig. 4 for comparable plots of Chl a and results from a wider ensemble of Earth System Models.