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

Moreover, seven out of nine of the MEMs we considered are forced by temperature and pri- mary and/or secondary producers [5, 35], without considering variables such as deoxygenation and acidification, which may act in combination to warming and magnify trophic amplification. 4.3 Various mechanisms towards consistent model responses? As shown in S1 Table, all models considered in this study differ and the variability observed among MEM biomass response types reflects the different ways in which forcing variables (temperature and primary production/biomass or plankton concentration) can be integrated into MEMs with different fundamental assumptions [19]. Despite differences in the total consumer biomass responses to changes in NPP, six MEMs exhibit agreeing temporal and spatial patterns from the most complex one, APECOSM, to models of intermediate complexity, FEISTY, ZooMSS and BOATS, to the simplest ones, Macroecological and EcoTroph. Although there are many differences between these MEMs, our study suggested that temperature-driven processes may be pivotal drivers to explain tro- phic amplification patterns. While temperature has a direct and negative impact on biomass in EcoTroph and Macroecological, interactions of multiple temperature-dependent processes drive the responses of APECOSM, FEISTY, ZooMSS and BOATS. Within these four models, different rates such as growth, mortality and basal metabolism can be affected by temperature. Taken together, these processes generally cause a warming-induced decrease in biomass. For

example, in FEISTY, basal metabolism increases faster than ingestion with temperature increases, so warming reduces the scope for growth, leading to a decrease in biomass [51]. In BOATS, warming influences the representative size of primary producers and sets the scope for growth of fish communities [48], warming causes an increase of the growth, reproduction and mortality rates leading at food web scale to a decrease in biomass [19, 37]. The differences in the processes which are affected by temperature may explain the differences in the biomass responses in terms of magnitude, type of response and spatial pattern. For instance, in APE- COSM, the temperature-induced changes in growth and metabolic rates, which determine the changes in biomass, varies with food limitation (more important in limited food regions than in not limited food regions). These temperature-associated mechanisms may cause the observed negative amplification where NPP is projected to decrease while they lead to positive attenuation or positive inversion (depending on warming magnitude and associated impact) where NPP is projected to increase. In contrast to the six aforementioned models, DBEM, DBPM and EcoOcean differ in terms of amplification (no amplification at global scale in DBEM and EcoOcean) and spatial pattern in biomass response to change in NPP (Figs 4 and S4). Their response to temperature changes may explain, at least partly, the absence of trophic amplification and their differences with other models. For example, in EcoOcean, environmental conditions and notably temperature

drive habitat suitability which can cause animals to relocate, and some animals can move across several 1-degree grid cells in a month. Thus, observed changes in biomass in a grid cell are only partially attributable to trophic cascades, which makes it harder to pinpoint trophic amplification [40]. For DBEM, a mechanistic species distribution model, temperature affects production by altering the individual growth of each modelled species that subsequently affects its population dynamics. Temperature and other environmental variables drive the spatial dis- tribution of species. However, in this model, organisms do not interact except with primary producers [19, 52]. This could be a potential reason for its weaker trophic amplification, sup- ported by the fact that DBEM has the lowest or near lowest ratio values (Figs 6A and S7A). Specifically, NPP is a key parameter for the carrying capacity. As a result, warming-induced PLOS ONE Trophic amplification revealed PLOS ONE | https://doi.org/10.1371/journal.pone.0287570 August 23, 2023 15 / 23