Ecological risk assessment
Ecological risk assessments using Bayesian Network-Relative Risk Model
Assessing ecological risks at scales relevant for managers is challenging as it requires to consider a large variety of stressors and impacts on a number of habitats and communities. The Bayesian Network-Relative Risk Model (BN-RRM) framework is particularly adapted in this case as it can take into account multiple stressors (anthropogenic and natural alike) and response types over large spatial scale.
The model weights the relative importance of each stressor on a set of defined ecological endpoints (e.g. species richness). For this, a conceptual model linking sources of stressors (e.g. agriculture and vessel traffic) to stressors (e.g. herbicides and alien marine pests) and then stressors to ecological endpoints (e.g. crustacea and diatom richness) is built (see Figure below). Using prior knowledge and field data, Bayesian Networks then calculates the probability of a specific response (e.g. high diatom richness) occurring according to the state of stressors influencing it.
Prior analysis, the study area is divided into discrete sub-regions to account for the spatial variability of the stressors and biological communities. The relative importance of each stressor and their risks to the system is examined independently for each area. BN-RRM can then be used to test different management scenarios.
Introduction of eDNA biomonitoring data
Recently, Graham et al. (2019) demonstrated the relevance of including eDNA data into such models to predict the relative richness of benthic taxonomic groups. Here, we propose to introduce eDNA biomonitoring data into BN-RRM for modelling effects of land-based contaminants (i.e., pesticides and nutrients) and marine alien species on coastal communities. In this context, eDNA biodiversity data will be used to determine ecological endpoints such as species richness of a given taxonomic group, but also for assessing the presence of non-native species (i.e. evaluating a potential stressor).