Supporting code and dataset for Deep learning predicts climate and anthropogenic impacts on African Cerambycidae beetles: contributions to SDGs 12 and 15

This repository contains code and data for the research paper "Deep learning predicts climate and anthropogenic impacts on African Cerambycidae beetles: contributions to SDGs 12 and 15". The study models the dynamics of two indicator species (Morimus funereus and Tithoes maculatus maculatus) in Cameroon using a deep neural network, assessing their contribution to Sustainable Development Goals (SDGs) 12 (Responsible Consumption) and 15 (Life on Land). ## Authors1. NGNANIYYI ABDOUL¹*2. TAGNE TAKOUTCHOUANG RICK²3. TALIPOUO ABDOU³4. MOHAMADOU DALAILOU HOUSSEINI⁴5. PENANJO STÉPHANIE⁵6. TEIKEU NGUEVEU ERIC DONALD⁶7. SEINO RICHARD AKWANJOH⁷ ### Affiliations¹ Applied Biology and Ecology Research Unit (URBEA), Department of Animal Biology, Faculty of Science, University of Dschang, P.O. Box 67, Dschang, Cameroon² Research Unit of the Scientific and Educational Center in Specialized Informatics, Department of Electronic Computer Engineering, Faculty of Computer Technology, Ryazan State Radioengineering University, Gagarina St. 59/1, Ryazan, Russia³ Laboratory of Zoology, Department of Animal Biology and Physiology, Faculty of Science, University of Yaoundé 1, P.O. Box 337, Yaoundé, Cameroon⁴ Higher Institute of Agriculture, Forestry, Water and Environment (ISABE), University of Ebolowa, P.O. Box 118, Ebolowa, Cameroon⁴ Laboratory of Botany and Ecology (LBE), Department of Plant Biology, Faculty of Science, University of Yaoundé 1, P.O. Box 812, Yaoundé, Cameroon⁶ Soil Analysis and Environmental Chemistry Research Unit (URASCE), Department of Earth Sciences, Faculty of Science, University of Dschang, P.O. Box 67, Dschang, Cameroon⁷ Laboratory for Genetic Toxicology and Orthopterology (LAGTO), Department of Zoology, Faculty of Science, University of Bamenda, P.O. Box 39, Bambili-Bamenda, Cameroon *Corresponding author: NGNANIYYI ABDOULEmail: ngnaniyyi@gmail.comORCID: orcid.org0000-0002-3897-3057 ## DatasetThe dataset includes climate and anthropogenic variables for predicting Cerambycidae beetle populations: ### Input Variables:- precipitation: Precipitation data- temperature: Temperature data- NDVI: Normalized Difference Vegetation Index- HWV: Harvested Wood Volume (m³/ha/year)- PUM: Pesticide Mass Used (kg/ha/year active substance)- THN: Total Habitat Number ### Target Variables:- Sp1: Morimus funereus population density- Sp2: Tithoes maculatus maculatus population density ## Code StructureThe main analysis is implemented in `cerambycidae_deep_learning.py` which includes: Data loading and exploratory analysis2. Data preprocessing and transformation3. Deep learning model architecture4. Model training with callbacks5. Model evaluation and performance metrics6. Feature importance analysis (permutation and SHAP)7. Comparison with traditional machine learning methods8. Results saving for publication ## Requirements- Python 3.7+- TensorFlow 2.4+- scikit-learn- pandas, numpy- matplotlib, seaborn- shap ## Usage1. Place your dataset file in the same directory2. Update the file path in the code if necessary3. Run the script: `python cerambycidae_deep_learning.py` ## ResultsThe code generates:- Publication-quality figures (PDF format)- Performance metrics (Excel format)- Trained model (Keras format)- Processed data (PKL format)- Training history (CSV format)