• Acronym: MEMIP
• Co-Chairs:
  Shin-ichi Ito (Japan)
  Harold (Hal) P. Batchelder (USA)
  

Modeling is a central approach for comparative analyses of ecosystems, i.e. concerning the structure, functioning and impact responses of marine ecosystems. It is important for process and modelling studies to identify if interrelationships amongst physical and biological variables are the same in different locations of whether certain relationships vary geographically or if the conclusions are dependent on the particular applied modeling tool.

Past PICES modeling activity has concentrated on the development of the NEMURO family of models. The strategy of NEMURO was to develop and apply the same model to multiple locations in order to remove the “model” confounding effect and isolate localized or species effects. This process is moving forward and applications of the NEMURO family of models are progressing in several ecosystems in the North Atlantic as well as in the North Pacific (NEMURO special volume in Ecological Modeling Vol. 202, ICES Annual Science Meeting, 2007).

Alternatively, when a single “correct” model cannot be identified a priori, a suite of models can be applied to the same system to determine not only which models are appropriate, but also the range of outcomes that may be expected. This is similar to the IPCC procedure for evaluating alternative climate models, a process that has been widely accepted. This project will implement the same model evaluation process except that we plan to use marine ecosystem models instead of climate prediction models. Thus, the idea behind this project is to apply multiple ecosystem models to the same location/species and to use an ensemble model forecast to identify and compare predicted and observed responses of marine ecosystem types to global changes. Other recent model comparison exercises have been undertaken using NPZ models (Friedrichs 2001, Friedrichs et al. 2007, Friedrichs and Hofmann 2001, Friedrichs et al. 2006, Hood et al. 2006) and Ecopath models (Taylor and Wolff 2007). Also Éva E. Plagányi (2007) recently conducted an in-depth qualitative analysis of the characteristics, data requirements and outputs of a large number of models appropriate for addressing management of fisheries in an ecosystem context.

The ability to evaluate the range of ecosystem response from different modeling approaches will produce valuable outcomes. Through this process we hope to be able to identify and characterize components of the major marine ecosystems which are likely to be affected at an early stage by global changes, to understand the responses to global change of each component of the ecosystem, focusing primarily on zooplankton which provide the prey base for upper trophic level fish species, and to use ecosystem models to identify and compare predicted and observed responses of marine ecosystem types to global changes. We will also be able to identify which of the candidate models are the most successful at hind-casting in each of the ecosystems chosen for study.

A key outcome of these comparisons will be to identify “early-warning” indicators of large-scale ecosystem changes, and to learn the extent to which these indicators are similar among a variety of ecosystems when multiple systems are analyzed. Early identification of potential indicators will provide opportunities for monitoring and assessment through planned field and modelling activities.

We will use several species of copepods and Pacific krill (Euphausia pacifica) as the modeled indicator species. Modeling the lower trophic level with minimally ecologically complex models makes the modeling task easier with respect to parameterizing and configuring multiple models. These candidate species are widely distributed in the North Pacific, are well studied, and have what we believe to be ecological equivalents in the North Atlantic, thus facilitating collaboration with North Atlantic colleagues. For many, there also exist substantial, high quality time series. The final decision of the indicator species on which to focus will be decided by the working group, once data sets are assembled and evaluated.