This study quantifies the relative contributions of material crystallinity and environmental factors to the marine degradation rate of poly(ε-caprolactone) (PCL). PCL sheets with controlled crystallinity were exposed at six Japanese coastal locations for 6–15 months. Machine learning regression (CatBoost, R² = 0.60) combined with SHAP analysis revealed that water temperature contributed most strongly to degradation rate variation, followed by depth and total nitrogen, while crystallinity showed moderate influence. These findings demonstrate that environmental conditions substantially drive degradation rate variability in realistic marine settings, providing insights for field-relevant material design and environmental compatibility assessment.
- Hironori Taguchi
- Takako Kikuchi
- Keiji Tanaka