Performance of Different Modelling Techniques in Testing the Impact of Environmental Variables on Eel Landing in Ichkeul Lake, a RAMSAR Wetland and UNESCO Biosphere Reserve

Sabrine Sahbani,Béchir Bejaoui,Ennio Ottaviani,Sihem Benabdallah,Eva Riccomagno,Enrico Prampolini,Donata Melaku Canu,Hechmi Missaoui,Cosimo Solidoro

DOI: https://doi.org/10.1016/j.rsma.2024.103587

IF: 2.166

2024-05-29

Regional Studies in Marine Science

Abstract:Advanced modeling techniques, including Random Forest (RF) and Cubist model (CB), were used to assess the relationship between environmental factors and European eels ( Anguilla anguilla ) abundance and to provide insights into the lake's ecological status while considering climate change and anthropogenic influences. A comprehensive dataset, attained through extensive environmental and biological monitoring for the period 2010-2020, was employed. The performance of the models is carried out using key metrics including the root mean square error (RMSE), coefficient of determination (R2), and mean estimation error (MAE). In addition, a sensitivity analysis was conducted to ascertain the relative significance of the thirteen input variables in shaping the predictions of the models. The precision of the CB and RF models in predicting eel landings surpassed that of Multiple Regression. In the training dataset, the CB model achieved R 2 =0.55, RMSE=7.68 tons, and MSE=6.20 tons, and the RF model achieved R 2 =0.56, RMSE=7.20 tons, and MSE=5.56 tons. High accuracy was maintained on the testing dataset, with CB achieving R 2 =0.73, RMSE=5.13 tons, and MSE=5.89 tons, and RF achieving R 2 =0.73, RMSE=5.81 tons, and MSE=4.67 tons. The scatter plot between predicted and measured eel landings indicated that the RF model tends to overestimate lower values and underestimate higher values of eel landings, while the CB model gave better performance in this context. Further, the carried sensitivity analysis using the CB model unequivocally identified three pivotal factors – water level, salinity, and turbidity level – as the most influential determinants governing the landing of eels in this ecosystem. Thus, the CB model is considered to be more promising for interpreting the relationship between environmental parameters and eel landings, which could be used by managers for an effective lake management strategy.

ecology,marine & freshwater biology