Résumé: Size-based approaches are paramount tools for the study of marine food webs. Here, we investigated the relationship between zooplankton body size, stable isotope composition and trophic level (TL) along a large-scale onshore-offshore gradient in the western tropical Atlantic. Samples were obtained on the Brazilian continental shelf, slope and in oceanic waters (off Fernando de Noronha archipelago and Rocas Atoll) in September and October 2015. Zooplankton was sieved into five size fractions. Zooplankton was dominated by copepods, except for the largest (> 2000 mu m) size fraction, that showed a high biovolume of chaetognaths, decapods, and fish larvae. Maximum zooplankton abundance and biovolume was found at the continental slope. POM showed consistently lower delta C-13 than zooplankton, indicating a selective use of C-13-rich primary food sources by zooplankton. Particulate organic matter (POM) was more C-13-enriched in shelf areas (average: -22.8, -23.6 and -24.3% at the shelf, slope and oceanic islands, respectively), probably due to the higher abundance of diatoms nearshore. POM had delta N-15 values between 2.5 and 6.9% (average: 4.0, 4.9 and 4.2% at the shelf, slope and oceanic islands, respectively). Zooplankton delta N-15 and TL increased with body size. The delta N-15 of the 200-500 mu m size fraction was used as baseline for TL estimation. Oceanic areas (average baseline delta N-15 = 5.8% +/- 0.52, n = 14) showed a higher baseline delta N-15 than the shelf (average = 3.9% +/- 0.69, n = 9) and the slope areas (average = 3.1% +/- 0.93, n = 9). In spite of differing baselines, the delta N-15 data produced a consistent pattern of log-linear increase in TL with increasing size, in all areas. The choice of input trophic enrichment factor (TEF) values only slightly changed the log10 (body size) vs TL slopes, but this choice had a considerable effect on the estimates of predator/prey size ratio (PPSR) and predator/prey mass ratio (PPMR). Using a TEF above 2.3 leads to unrealistic PPSR and PPMR estimates. Overall average slope was 0.59 +/- 0.08 TL mu m(-1) with TEF = 2.3 and 0.42 +/- 0.07 TL mu m(-1) with TEF = 3.2.

Résumé: Ensemble niche modelling has become a common framework to predict changes in assemblages composition under climate change scenarios. The amount of uncertainty generated by the different components of this framework has rarely been assessed. In the marine realm forecasts have usually focused on taxa representing the top of the marine food-web, thus overlooking their basal component: the plankton. Calibrating environmental niche models at the global scale, we modelled the habitat suitability of 106 copepod species and estimated the dissimilarity between present and future zooplanktonic assemblages in the surface Mediterranean Sea. We identified the patterns (species replacement versus nestedness) driving the predicted dissimilarity, and quantified the relative contributions of different uncertainty sources: environmental niche models, greenhouse gas emission scenarios, circulation model configurations and species prevalence. Our results confirm that the choice of the niche modelling method is the greatest source of uncertainty in habitat suitability projections. Presence-only and presence-absence methods provided different visions of the niches, which subsequently lead to different future scenarios of biodiversity changes. Nestedness with decline in species richness is the pattern driving dissimilarity between present and future copepod assemblages. Our projections contrast with those reported for higher trophic levels, suggesting that different components of the pelagic food-web may respond discordantly to future climatic changes.

Résumé: Climate change effects on marine ecosystems include impacts on primary production, ocean temperature, species distributions, and abundance at local to global scales. These changes will significantly alter marine ecosystem structure and function with associated socio-economic impacts on ecosystem services, marine fisheries, and fishery-dependent societies. Yet how these changes may play out among ocean basins over the 21st century remains unclear, with most projections coming from single ecosystem models that do not adequately capture the range of model uncertainty. We address this by using six marine ecosystem models within the Fisheries and Marine Ecosystem Model Intercomparison Project (Fish-MIP) to analyze responses of marine animal biomass in all major ocean basins to contrasting climate change scenarios. Under a high emissions scenario (RCP8.5), total marine animal biomass declined by an ensemble mean of 15%–30% (±12%–17%) in the North and South Atlantic and Pacific, and the Indian Ocean by 2100, whereas polar ocean basins experienced a 20%–80% (±35%–200%) increase. Uncertainty and model disagreement were greatest in the Arctic and smallest in the South Pacific Ocean. Projected changes were reduced under a low (RCP2.6) emissions scenario. Under RCP2.6 and RCP8.5, biomass projections were highly correlated with changes in net primary production and negatively correlated with projected sea surface temperature increases across all ocean basins except the polar oceans. Ecosystem structure was projected to shift as animal biomass concentrated in different size-classes across ocean basins and emissions scenarios. We highlight that climate change mitigation measures could moderate the impacts on marine animal biomass by reducing biomass declines in the Pacific, Atlantic, and Indian Ocean basins. The range of individual model projections emphasizes the importance of using an ensemble approach in assessing uncertainty of future change.