A generic model for Antarctic marine food webs: a tool for data summary and scenario testing
Résumé
Understanding the functioning of Antarctic marine ecosystems and projecting their possible responses to environmental drivers require analysis of extensive data sets, mathematical models to integrate these data, and assessment of data uncertainty on the model projections. In this study, a generic food web model with 22 trophic groups was used to describe the structure and functioning of two Antarctic marine food webs that are representative of a region with heavy sea ice (southern region of the Western Antarctic Peninsula, sWAP) and one with little sea ice (South Georgia, SG). For each food web, ecosystem metrics were calculated using Monte Carlo methods that take into account uncertainty in the data used to define model parameters. Common scenario simulations were used to investigate possible changes in the structure of these food webs to changes in sea ice extent and biomass of dominant food web components. The uncertainty associated with data sets from the two regions was similar, average sWAP CV = 0.5 and average SG CV = 0.46. In both systems, microzooplankton consume 2 times more small phytoplankton production than large phytoplankton production, while mesozooplankton and macrozooplankton consume 3 times more large phytoplankton production. Overall, the microbial part of the food web consumes a large percentage of the primary production in both systems (68% in sWAP and 46% +- 16 in SG), suggesting that improved definition of these processes is needed. Simulations show that a reduction in Antarctic krill (Euphausia superba) biomass corresponds to an almost linear reduction in the biomass of top predator (penguins, crabeater seals, baleen whales) biomass in both systems (coefficient of correlation = 0.90). This result is in agreement with common understanding of the importance of Antarctic krill as diet component for top predators. Simulated shifts from large to small phytoplankton caused decreases in top predator production in the sWAP (coefficient of correlation = 0.82) but not at SG, where modelled zooplankton groups were also less sensitive to phytoplankton type. More diet data for all zooplankton groups are needed to investigate if this is due to actual differences in the food web structure between the two regions or to uncertainty in the data used to parameterize the models. This generic food web model structure provided a useful framework for consistent comparisons of the responses of two ecosystems to common environmental drivers. As such, it provides a basis for input to larger integrated ecosystem models that include the physics and biogeochemistry of the Southern Ocean. This type of common framework will facilitate the comparison among different regional food webs and the integration in larger ecosystem models for the Southern Ocean.