This repository contains the data and code associated with the study “Thermal stress drives seagrass fragmentation in the Mediterranean Sea.”The study integrates physiological modelling, sea surface temperature (SST) datasets, and satellite-based habitat mapping to quantify the effects of cumulative thermal stress on the spatial structure and cover of Posidonia oceanica meadows across the Mediterranean basin. Specifically, the repository includes: Example notebooks to compute Stress Degree Days (SDD) and thermal stress from daily SST data. Example notebook and R script to calculate fragmentation indices, including the gap-weighted fragmentation index (FI) and the composite landscape fragmentation index derived from standardized spatial metrics. Derived datasets containing: Basin-wide SDD and thermal stress estimates for the historical period (2000–2020). Projected SDD and thermal stress values under RCP4.5 and RCP8.5 scenarios (2006–2100). Predicted P. oceanica habitat maps for the analyzed Mediterranean regions, obtained using the CAMELE deep learning model. The CAMELE model (Convolutional neural network for Automated Marine Ecosystem Labelling) can be freely downloaded from this Zenodo repository and used for large-scale benthic habitat mapping across the Mediterranean Sea. The raw satellite imagery used to generate habitat maps cannot be shared due to license restrictions under Planet’s Education and Research Program.