The analysis of socio-ecological systems (SES) is increasingly approached with models because of their multiple sources of complexity: bi-directional interactions, feedback between social and ecological systems, processes occurring at several scales, links between different levels of aggregation... (Martin and Schlüter, 2015) .
Multi-agent models analyze the functioning of an SES through the study of interactions between social and ecological dynamics (Bousquet, 2001; Bousquet and Le Page C., 2004; Schlüter et al., 2014; Le Page, 2017). They are based on distributed modelling where components are represented as autonomous entities (agents) that interact with each other (Ferber, 1995; Weiss, 2013). Several dedicated modeling platforms such as NetLogo, Repast, Cormas or GAMA provide a development environment adapted to this paradigm.
Spatial and temporal dimensions are inherent in multi-agent models because they allow to build a virtual simulation environment, implicit or explicit (Le Page, 2017) and to synchronize the different processes modeled using a scheduler (Wooldridge, 2009). With regard to their scalar flexibility and the possibility of implementing anthropogenic and natural dynamics at different levels of abstraction, multi-agent systems have demonstrated their ability to address complex interaction systems (Filatova et al., 2013).
Despite the diversity of modelling environments and the wealth of specifications proposed to describe human-environment interactions, the spatio-temporal restitution of these interactions remains an important scientific barrier in socio-ecosystem approaches. Many authors have shown that the coherence of spatial and temporal scales is both a key point and one of the main pitfalls in restoring the variability of anthropogenic and environmental signals in models (Liu et al., 2007 ; Filatova, 2013 ; Truong, 2015). Ecosystem dynamics are not always synchronized with the evolution of human activities but are subject to complex cycles that remain particularly difficult to model. Environmental science research supports this view and shows that the main limitations are inherent in the difficulties of simplifying complex natural processes within a model (Levin, 1998; Pascual, 2005; Leles et al., 2016; Shimoda and Arhonditsis, 2016).
Aims
This workshop aims to exchange on modelling practices, the consideration of scales, and the integration of spatial and temporal data in marine science research. Through feedback and exchanges in working groups, the following themes will be addressed:
Model formulation: moving from observation to modeling
The tools that can be mobilized: which framework, which platform?
Changes in scale: emergence, multi-level approaches, integration of spatial and temporal scales in models
Link to observations: calibration, validation
Organization
The workshop is free of charges and will take place over 2 days, on 15-16 October 2019 (see the programme) at the Pôle Numérique Brest Iroise (PNBI) in Plouzané (see practical information).
The format of the workshop will be as follows:
Introductory plenary presentations by invited researchers on modelling strategies
Round table of participants in the form of two slides defining the topics of study and interest for modelling
Group work: on publications related to the introductory presentations and discussion topics
Demonstration of simulation tools and platforms
Target audiance
The workshop is open to all researchers interested in Breton modelling and/or modelers, with invited researchers (list in the process of being finalised).
For practical reasons, the number of participants is limited to 80.
