ABSTRACT
The present study aims to forecast the effect of climate change and anthropogenic pressures on Tunisia's Ichkeul Lake. This lake is among the most productive ecosystems in the Mediterranean and is particularly important for the non-migratory European eel, which serves as an excellent bioindicator of the ecosystem's health.
Random Forest (RF) and Cubist (CB) approaches were developed for this purpose using a dataset from the long-term (2010–2020) environmental and biological monitoring of Lake Ichkeul.
The results revealed pronounced seasonal variability of the environmental parameters. While the models’ outcomes highlighted their best predictive performance and revealed that the most important predictors of eel landings appear to be, salinity, turbidity, and water level. These results are consistent with the relationships found with Pearson correlation.
Our results proved that these variables should always be included in management measures and can be widely extended to other Tunisian coastal ecosystems.
This study has also proven that the sustainable approach, which combines the two models, can be used in decision making by civil authorities and other interested stakeholders.
