Modelling global ecosystem structure and function on land and sea

Background/Question/Methods

Ecologists, conservationists, and policy-makers use models to explain and predict changes in the biosphere as a result of anthropogenic impacts. These models often take a statistical, correlative approach, fitting to observations without explicitly incorporating mechanisms. Process-based models that implement the fundamental equations describing a system and allow for emergent properties and behaviour offer a greater chance of furthering our structural understanding, along with greater confidence in predictive ability. Although such models are common in the physical sciences (e.g. general circulation models for understanding climate change), they are less frequently applied to ecosystems. In the terrestrial realm, process-based models tend to only include autotrophs (Dynamic Global Vegetation Models). In the marine realm, process-based models do exist, but tend to be regional, taxon-specific, and focussed on questions of fisheries or biogeochemistry, rather than ecosystems and biodiversity. Here we present a global, spatially-explicit process-based model of ecosystems that uses the same underlying and unified concepts to model both land and sea. Our aim is both to provide a 'proof-of-concept' for global process-based models of ecosystems, and eventually to provide a policy-relevant tool for examining the consequences of socioeconomic scenarios and trajectories over the coming century. The code for this model will be freely available to the ecological community for alteration and improvement.

Results/Conclusions

Modelled estimates of growth rates, generation lengths, mortality rates and other life history characteristics are broadly in line with empirical results from the literature. Emergent ecosystem properties such as biomass pyramids are also generally consistent with observational data. We conclude that the initial proof-of-concept of a dynamic marine-terrestrial global ecosystem model based on fundamental ecological processes and allowing for maximal emergence in outputs is feasible, though considerable effort remains to continue to address shortcomings and to further develop the model in terms of ecological realism.