Week 2: Build mizer models

Now that we have a good understanding of the basic principles behind sized based models, during this week we will start developing our own multi-species models. Our goal at the end of this week is to have a basic set of steady state parameters for our own model. In this week we will only focus on model properties at steady state or equilibrium conditions and will not yet explore temporal dynamics.

You are welcome to use your own model system during this week, assuming you have the data. Alternatively, we will focus on a Celtic Sea. The Celtic Seas is characterized by a diversity of habitats, such as an extensive slope, canyons, ridges, and seamounts. The commercial fisheries in the Celtic Sea target a large number of stocks. We will only focus on 17 species and confine our modelling for ICES (International Council for Exploration of the Seas) areas 7b,c,e-k. For those participants who will focus on the Celtic Sea model, our goal at the end of the week is for each participant to develop their own model parameterisation. We will then collect these alternative parameter sets and use them as a model ensemble next week.

The material is split into 4 tutorials:

  1. Finding species parameters
    Here we will explore the main species parameters that should be provided by the user to start building a multi-species model. We will also look at assumptions and defaults that mizer uses to fill non-essential parameter values. Your task will be to collect essential parameters from FishBase or other sources.

  2. Create your first model
    Now that you have species parameters you can build the first model. In this part we will focus on achieving the correct species abundances and growth rates.

  3. Refine your model: diet data
    In this tutorial we will introduce the tuneParams() shiny gadget which makes it much easier to make changes to model parameters and observe their effects on the system. We will use it to explore how the diet emerges in the model from the interplay between predation preference and prey abundance and adjust the model parameters to reproduce observed diets.

  4. Refine your model: landings data
    Finally we will tune the model parameters to make the model predictions for the catches agree with landings data, in particular the size distribution of the landings. We will again be using the tuneParams() gadget.

To get your worksheet repository for this week, follow this link: