The present study focused on using benthic macrofauna to assess the risks of moving away from Good Ecological Status towards the functioning of an anthropized system (Nokoué-Benin), based on de ning reference values for macroinvertebrate metrics, stress thresholds and the responses of selected metrics to stressors. The approach used is a combination of a joint species distribution model and Bayesian networks. We used JSDM to select relevant metrics and generate posterior probabilities. We then converted these posterior probabilities into posterior response probabilities for each of the stress levels and fed them into a Bayesian network. To de ne the reference values of the metrics and the stress thresholds derived from the probability density diagrams for low-pressure levels, we used the responses predicted by the Bayesian network. An application of this approach was then carried out on a lagoon sampled during high and low water periods for three consecutive years (2004)(2005)(2006), with 33 macroinvertebrate taxa present in all seasons and sampling points, and measurements of 14 environmental parameters used as application data. This study demonstrated a reliable inference, the selection of seven relevant metrics and the de nition of reference conditions. The relevance of the metrics and ecological assessment criteria, despite the small sample size, suggests the potential for wider applicability of the approach to aquatic ecosystem monitoring and assessment programs in developing countries, which are generally characterized by a lack of monitoring data.