Trophic amplification: A model intercomparison of climate driven changes in marine food webs

Abstract

Marine animal biomass is expected to decrease in the 21st century due to climate driven changes in ocean environmental conditions. Previous studies suggest that the magnitude of the decline in primary production on apex predators could be amplified through the trophodynamics of marine food webs, leading to larger decreases in the biomass of predators relative to the decrease in primary production, a mechanism called trophic amplification. We compared relative changes in producer and consumer biomass or production in the global ocean to assess the extent of trophic amplification. We used simulations from nine marine ecosystem models (MEMs) from the Fisheries and Marine Ecosystem Models Intercomparison Project forced by two Earth System Models under the high greenhouse gas emissions Shared Socioeconomic Pathways (SSP5-8.5) and a scenario of no fishing. Globally, total consumer biomass is projected to decrease by 16.7 ± 9.5% more than net primary production (NPP) by 2090-2099 relative to 1995-2014, with substantial variations among MEMs and regions. Total consumer biomass is projected to decrease almost everywhere in the ocean (80% of the world's oceans) in the model ensemble. In 40% of the world's oceans, consumer biomass was projected to decrease more than NPP. Additionally, in another 36% of the world's oceans consumer biomass is expected to decrease even as projected NPP increases. By analysing the biomass response within food webs in available MEMs, we found that model parameters and structures contributed to more complex responses than a consistent amplification of climate impacts of higher trophic levels. Our study provides additional insights into the ecological mechanisms that will impact marine ecosystems, thereby informing model and scenario development.

Fig 1

Main outputs of the Earth system models considered in the current study: mean change in Sea Surface Temperature (SST) (a) and percent change in Net Primary Production (NPP) (b) in the 2090s relative to the reference period 1995–2014, under SSP5-8.5 for GFDL.

Fig 2. Conceptual scheme showing the six total consumer biomass response types to changes in NPP.

(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig 3. Ensemble projections of NPP and total consumer biomass changes, relative to 1995–2014, under SSP5-8.5 and for all the considered MEMs.

(a) based on GFDL/MEMs combinations; (b) based on IPSL/MEMs combinations. All values are relative to the standardised reference period of 1995–2014. Vertical grey line indicates the last year of the historical period.

Fig 4. Spatial distribution of total consumer biomass response types, for the seven considered MEMs forced with GFDL-SSP5-8.5.

Each response type is estimated at a 1° × 1° resolution, from the change in consumer biomass and NPP expected at the end of the century, relative to the reference period (Figs 1 and S1A).

Fig 5. Magnitude of the different types of change for the seven considered MEMs forced with GFDL-SSP5-8.5.

Each response type is estimated at a 1° × 1° resolution, from the change in consumer biomass and NPP expected at the end of the century, relative to the reference period (Figs 1 and S1A).

Fig 6. Model agreement for biomass response types projected in the 2090s, under GFDL-SSP5-8.5 configuration.

Response types correspond to at least six out of seven models in agreement for the coloured cells, while grey cells indicate where fewer than six models project the same type of response and white cells where no data were available. The percentage numbers in the pie chart correspond to the relative surface areas.

Fig 7

Evolution of R ratio (total consumer biomass change divided by NPP change) as a function of SST increase relative to the reference period: where NPP is expected to decrease (a) or increase (b) under GFDL-SSP5-8.5. Grey lines separate biomass response types. (1), (2) and (3) refer to positive inversion, negative attenuation, and negative amplification, respectively, while (4), (5) and (6) refer to negative inversion, positive attenuation, and positive amplification, respectively. Colour shaded areas correspond to standard deviation.

Fig 8. Disaggregation of consumer biomass response across the food web.

The red gradient indicates changes in consumer biomass by weight classes, the purple gradient indicates changes in consumer biomass by length classes, and change in NPP is reported in black for the framework forced by GFLD (left column) and IPSL (right column) outputs under SSP5-8.5. Gradients light colours to dark colours correspond to small size classes to large size classes, respectively.