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

impacted by climate change with high biomass losses. These species play a crucial role in the stability of food webs and ecosystems in general [28, 29] and are the main target of most fisher- ies worldwide [30, 31]. Because high trophic level species play key ecological and socio-eco- nomic roles, it is important to better understand the underlying mechanisms to face the challenges that the 21st century will pose, including biomass loss and food security. In this study, we examine the spatial and temporal patterns of trophic amplification in the global ocean, initially based on the definition of Lotze et al. (2019). We consider nine MEMs that have been forced by two CMIP6 ESMs under the high emissions scenario (SSP5-8.5) to investigate the trophic amplification process across a variety of marine ecosystem models with different assumptions, structures, and processes. We analyse how climate-induced NPP changes (ESM outputs) may propagate to higher trophic levels and explore spatial variations across the global ocean using a classification of biomass responses (MEMs outputs) adapted from Chust et al. (2014). Furthermore, we investigate the potential warming effects on the pro- jected trophic amplification. Finally, we track trophic amplification through the food web, from primary consumer to high trophic levels, using all available MEMs outputs. 2. Materials and methods 2.1 Changes in biomass, primary productivity and temperature This study analysed global projections from ESMs and MEMs available from CMIP6 and Fish- MIP, respectively [3, 5].

Outputs from two ESMs from CMIP6 were considered: IPSL-CM6A-LR developed by the Pierre Simon Laplace Institute (IPSL, [32]) and GFDL-ESM4 developed by the Geophysical Fluid Dynamics Laboratory (GFDL; [33]). ESMs integrate the interactions between the atmo- sphere, ocean, land, ice and biosphere to simulate the state of regional and global climate and their changes under a wide variety of greenhouse gas emissions pathways [10]. Greenhouse gas emissions pathways that are adopted by the Intergovernmental Panel on Climate Change (IPCC) are the Shared Socioeconomic Pathways (SSPs, [34]). Projections by ESMs under stan- dardised simulation experiments and SSPs are available from Earth System Grid Federation (ESGF,https://esgf-node.llnl.gov/projects/esgf-llnl/). CMIP6 ESMs data used to force FishMIP MEMs was processed to a two-dimensional hori- zontal regular 1˚ x 1˚ grid (https://data.isimip.org). Here we use this data on an annual time resolution from 1950 to 2099, under SSP5-8.5 where the concentration of greenhouse gases will rise throughout the 21st century. This choice of scenario has been elected in order to easily disentangle mechanisms within each MEMs and better showcase as well as understand models’ reaction to their forcing variables. Outputs from nine MEMs were gathered from Fish-MIP (fishmip.org, [5], data extraction were made the 25 of November 2022): APECOSM, BiOeconomic mArine Trophic Size-spec- trum model (BOATS), Dynamic Bioclimate Envelope Model (DBEM), Dynamic Benthic Pelagic Model (DPBM), EcoOcean, EcoTroph, FishErIes Size and functional TYpe model (FEISTY),

Macroecological, and Zooplankton Model of Size Spectra (ZooMSS). MEMs were forced using a set of environmental outputs from the two ESMs based on a no-fishing and SSP5-8.5 scenario [3, 5, 27]. All MEMs used the IPSL outputs, while only seven MEMs used the GFDL outputs, due to specific and non-compatible data requirements of APECOSM and DBPM. The nine selected MEMs differ in how they represent ecosystem structure, composition, and trophodynamics. For instance, ecosystem components are represented by size classes (BOATS and Macroecological), functional groups (EcoOcean), trophic levels (EcoTroph), spe- cies (DBEM), or a hybrid between size-classes and traits (APECOSM, DBPM, FEISTY, and ZooMSS), whereas food web links can be represented by trophic networks, diet composition PLOS ONE Trophic amplification revealed PLOS ONE | https://doi.org/10.1371/journal.pone.0287570 August 23, 2023 3 / 23