Bayesian modelling tool to assess the risk of mass development and the consequences of macrophyte removal

What happens if we remove macrophytes?

The first task in developing a Bayesian Network (BN) model is to produce a causal flow chart (without feedback loops), e.g. as shown in Figure 1 below. Such flow chart can be co-constructed from the scientific literature, available data and knowledge from stakeholders. The nature and amount of knowledge to link the boxes (nodes) and provide probabilities of effects may be variable, from mathematical equations, data mining (descriptive statistics), expert or local knowledge. It is important to be transparent so that the user can appreciate the assumptions made and check out the source of information used. 

 

MadMacs derived probabilities from general knowledge in ecology (food-web and alternative stable state: macrophyte clear water to phytoplankton green pea soup) and a global review of the effect of plant removal on ecosystem structure and function (Thiemer et al. 2021). The BN remains largely hypothetical and should only be used for illustrative purposes. The BN provides a mathematical platform to determine the probabilities of phytoplankton blooms as a function of a set of interrelated causes. Scientists and water managers can manipulate the BN to quantify this risk under different scenarios, e.g. to simulate alternative desirable ecosystem services.

Scientists and water managers may also set the risk of phytoplankton blooms (endpoint) to a specific target and see how probabilities are affected backwards throughout the whole BN, identifying key nodes on which the set target depends.

BNs provide an educational tool informing the discussion between scientists, managers and users of aquatic ecosystems: everyone learns everyone teaches.

Do you want to play?

The BN works with the software Netica, a free version is available here. The free version allows you to use (but not modify) the BN. Once you have installed Netica, download the MadMacs netica file. You should be able to see the BN as here.

For more information see Thiemer et al. (2021) and supplementary information therein.