From Suitability to Invasion Risk: Integrating Correlative and Mechanistic Models for Predicting the Potential Distribution of the Red Palm Weevil

Abstract

The red Palm Weevil (RPW) Rhynchophorus ferrugineus is the most devastating pest of palm trees and is responsible for significant economic and biodiversity losses worldwide. The larvae bore deep into the trunk or crown for feeding, resulting in extensive tunneling and mortality of the trees. In this study, we developed a hybrid modeling framework for RPW that integrates ecological niche modeling (ENMs) with a temperature-driven risk index (RI). The ENM was developed from the RPW occurrence data, while the RI was driven from temperature-dependent life history traits of RPW. The two approaches were combined through a weighted additive method to enhance predictive performance and biological realism. The correlative SDM achieved the highest predictive performance (AUC = 0.984 ± 0.002; TSS = 0.891 ± 0.009), whereas the mechanistic RI performed substantially lower (AUC = 0.755 ± 0.009 and TSS of 0.457 ± 0.017). Hybrid performance increased with greater SDM contribution, with the 90% SDM–10% RI hybrid (Hybrid90) achieving performance comparable to the SDM (AUC≈0.983 ± 0.002; TSS≈0.893 ± 0.009). Bootstrap analyses showed overlapping 95% confidence intervals between the SDM and Hybrid90 across all evaluation metrics, while the RI remained consistently inferior. Model rankings were also robust across suitability thresholds. Spatial projections revealed high suitability across Southeast Asia, North Africa, the Middle East, the Mediterranean Basin, and coastal Brazil, consistent with known invasion patterns. Additionally, large areas of coastal Africa, Europe, and the Americas were identified as climatically suitable but remain partially uninvaded, indicating substantial invasion risk under the current scenario. Future climate scenarios suggest persistence of current hotspots alongside increased stability and expansion potential, particularly in East and West Africa. Our results demonstrate that integrating correlative and mechanistic information enhances the ecological interpretability of species distribution models without compromising predictive performance. Although the correlative SDM remained the most accurate predictor, the optimal hybrid model achieved comparable performance while incorporating biologically meaningful constraints on species persistence. This transferable framework supports biologically informed invasion-risk assessment, climate-change adaptation, and proactive surveillance and management of invasive pests, including the red palm weevil in regions that remain uninvaded.