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

groups generally better able to relocate away from warming areas, dampens trophic amplification. 4.5 Analysing trophic amplification based on the change in NPP comparison with total consumer production In our study, we focused on the changes in NPP relative to consumer biomass. Yet, the gross production of the living biomass is also a key parameter of ecosystem functioning and one of the main drivers of their resilience [28, 29]. In particular, fisheries sustainability relies on the harvested part of the gross production rather than the biomass. A few studies like Du Pontavice et al (2021) and Petrik et al (2020) have investigated this difference between production and bio- mass and showed a larger decrease in biomass than in production. BOATS, EcoTroph, and Macroecological, three of the MEMs considered in this study, introduce an explicit relationship between these two parameters. In EcoTroph, production and biomass are linked by the trophic flow kinetics, i.e. the speed of biomass transfers through the food web [56, 57], which accelerates with temperature therefore leading to a larger decrease in biomass than in production. In Macroecological, biomass at any given size corresponds to the ratio of total production at that size, by the mass-specific production rate [44], whereas in BOATS, biomass spectrum is esti- mated directly from fish production spectrum and mortality rate [37]. Under CMIP6, the addi- tional biomass decrease between total consumer production and consumer biomass is about 5%

and 15% under GFDL and IPSL, respectively, for both EcoTroph, FEISTY, and BOATS and about 15% and 5% under GFDL and IPSL, respectively for FEISTY (S8 Fig). Thus, an approach based on the productivity of ecosystems and not their biomass results in less trophic amplification, which suggests a smaller impact on fisheries potential catches. 4.6 Low trophic level, key MEMs driver, estimation uncertainty and consequences In our study, we find that depending on low trophic level forcing in each MEM (low trophic level biomass vs. low trophic level production), the magnitude of biomass changes relative to primary productivity is very different (S9 Fig), which is consistent with the results highlighted in Heneghan et al (2021). The magnitude of the biomass response of MEMs is, on average, larger for MEMs driven at low trophic level by production than those driven by biomass forc- ing (S1 Table and S9 Fig). The nature of this difference, thus, could firstly lie in the inconsis- tency that arises from 1-way forcing with biomass or production [19]. This could dampen the consumer decline when forcing with biomass. For example, the comparison of 1-way and 2-way forcing with APECOSM supports this assumption, although the small difference in terms of global biomass decline might not explain all [50]. This study highlighted the relative changes in consumer biomass compared to changes in NPP to investigate trophic amplification. Hence, we compared a production (NPP) to a bio- mass (consumer biomass). In a supplementary analysis, we explored the changes of low tro-

phic level drivers of each MEM (other than NPP; i.e., phytoplankton and zooplankton biomass) which have a very different spatial structure and temporal trend (S9 Fig) compared to NPP [5]. At the global scale, low trophic level (i.e., phytoplankton and zooplankton) biomass is projected to decrease under IPSL while NPP is projected to increase.. One of the mecha- nisms driving this inversion (increase in NPP, decrease in phytoplankton biomass) is when temperature-driven phytoplankton metabolic costs exceed what can be met with the increase in NPP. Under GFDL, the decrease in phytoplankton and zooplankton biomass is projected to be slightly stronger than that of NPP. PLOS ONE Trophic amplification revealed PLOS ONE | https://doi.org/10.1371/journal.pone.0287570 August 23, 2023 17 / 23