Modeling
This project applied machine learning and advanced regression techniques to model European anchovy (Engraulis encrasicolus) population dynamics, using environmental and fisheries-dependent data to forecast stock fluctuations and inform sustainable management strategies.
By integrating oceanographic variables with catch data, we developed predictive models that capture the complex drivers of anchovy recruitment and abundance.
Machine learning regression and classification models, time-series analysis, environmental variable selection, and cross-validation frameworks applied to fisheries-dependent and fisheries-independent datasets.
- Maximum Entropy (MaxEnt) models
- Environmental driver analysis
- Stock assessment recommendations
- Peer-reviewed publication
Our approach integrates species distribution modeling with population abundance data to develop robust predictive frameworks. We combine occurrence data with environmental layers to identify suitable habitat areas and link these with population dynamics.
By applying machine learning algorithms, we capture nonlinear relationships and interactions between environmental drivers and anchovy abundance, providing insights into the factors influencing stock fluctuations.
MaxEnt models over 28 years reveal habitat suitability patterns, highlight critical fish habitat and predict key environmental drivers influencing anchovy distribution.
Examples of environmental layers (from 2019) included in the models and demonstrated to influence anchovy distribution. Left panel shows anchovy occurrences, central panel presents the sardine distribution layer, and right panel illustrates sea bottom temperature dynamics.
3D visualization of the MaxEnt workflow represented.
Predicted habitat suitability (lag 1 year) and deep water formation was found to significantly influence anchovy distribution patterns, mediating anchovy population dynamics and abundance in the Adriatic Sea. This pattern was revealed through QGAM analysis.
Summary of environmental variable importance from MaxEnt outputs, and identified drivers of essential anchovy habitat in the Adriatic Sea from 1994 - 2021. These findings highlight the key drivers influencing predicted species habitat suitability.