View in EDS

Modeling desert locust population dynamics: A climate driven approach using machine learning.

Saved in:
Bibliographic Details
Title: Modeling desert locust population dynamics: A climate driven approach using machine learning.
Authors: Mamo, Dejen K.1,2 (AUTHOR) dejenketema@dbu.edu.et, Kinyanjui, Mathew N.3 (AUTHOR), Siewe, Nourridine4 (AUTHOR)
Source: Ecological Modelling. Jan2026, Vol. 511, pN.PAG-N.PAG. 1p.
Subjects: Desert locust, Population dynamics, Pest control, Boosting algorithms, Machine learning, Atmospheric models, Long short-term memory
Geographic Terms: Ethiopia
Abstract: The desert locust (Schistocerca gregaria) remains one of the most destructive migratory pests, with swarms capable of devastating vegetation, crops, and pasturelands, thereby threatening food security across arid and semi-arid regions. Motivated by the 2019–2022 upsurge in the Afar region of Ethiopia, we develop and analyze a hybrid climate-driven mathematical model that integrates mechanistic population dynamics with machine-learning-based climate predictions. The model comprises a nonlinear system of differential equations, incorporating temperature- and rainfall-dependent developmental rates, vegetation-mediated survival, and density-driven phase transitions between solitarious and gregarious forms. Climate inputs are generated using a Long Short-Term Memory (LSTM) network for daily temperature and an Extreme Gradient Boosting (XGBoost) algorithm for monthly rainfall, both trained and validated against satellite and ground observations. Analytical results for the autonomous case reveal three ecologically meaningful equilibria: (i) an extinction state, always unstable; (ii) a locust-free equilibrium, locally stable when the basic offspring number N 0 < 1 ; and (iii) a coexistence equilibrium, stable when N 0 > 1. Simulations driven by fitted climate data reproduce outbreak dynamics consistent with FAO Locust Watch reports, showing that temperatures of 25–35 °Celsius and rainfall of 40–100 mm/month enhance vegetation growth, suppress mortality, and accelerate population buildup, thereby triggering swarm formation through gregarization. Sensitivity analysis further highlights the central role of vegetation in modulating mortality, with low vegetation thresholds prolonging survival and amplifying outbreak magnitude. Overall, this study demonstrates the critical interplay between climate, vegetation, and behavioral phase transitions in shaping locust dynamics. By combining machine learning-enhanced climate forecasts with biologically grounded modeling, the framework offers a predictive, data-driven tool to support early warning, risk mapping, and climate-informed intervention strategies for sustainable desert locust management. • Hybrid ODE–ML model integrates climate forecasts with locust life-cycle dynamics. • LSTM and XGBoost improve temperature and rainfall prediction accuracy. • Vegetation–mortality feedback and phase transitions drive outbreak dynamics. • Global sensitivity analysis identifies vegetation and egg survival as key drivers. • Framework enhances early warning and intervention for desert locust management. [ABSTRACT FROM AUTHOR]
Copyright of Ecological Modelling is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:The desert locust (Schistocerca gregaria) remains one of the most destructive migratory pests, with swarms capable of devastating vegetation, crops, and pasturelands, thereby threatening food security across arid and semi-arid regions. Motivated by the 2019–2022 upsurge in the Afar region of Ethiopia, we develop and analyze a hybrid climate-driven mathematical model that integrates mechanistic population dynamics with machine-learning-based climate predictions. The model comprises a nonlinear system of differential equations, incorporating temperature- and rainfall-dependent developmental rates, vegetation-mediated survival, and density-driven phase transitions between solitarious and gregarious forms. Climate inputs are generated using a Long Short-Term Memory (LSTM) network for daily temperature and an Extreme Gradient Boosting (XGBoost) algorithm for monthly rainfall, both trained and validated against satellite and ground observations. Analytical results for the autonomous case reveal three ecologically meaningful equilibria: (i) an extinction state, always unstable; (ii) a locust-free equilibrium, locally stable when the basic offspring number N 0 < 1 ; and (iii) a coexistence equilibrium, stable when N 0 > 1. Simulations driven by fitted climate data reproduce outbreak dynamics consistent with FAO Locust Watch reports, showing that temperatures of 25–35 °Celsius and rainfall of 40–100 mm/month enhance vegetation growth, suppress mortality, and accelerate population buildup, thereby triggering swarm formation through gregarization. Sensitivity analysis further highlights the central role of vegetation in modulating mortality, with low vegetation thresholds prolonging survival and amplifying outbreak magnitude. Overall, this study demonstrates the critical interplay between climate, vegetation, and behavioral phase transitions in shaping locust dynamics. By combining machine learning-enhanced climate forecasts with biologically grounded modeling, the framework offers a predictive, data-driven tool to support early warning, risk mapping, and climate-informed intervention strategies for sustainable desert locust management. • Hybrid ODE–ML model integrates climate forecasts with locust life-cycle dynamics. • LSTM and XGBoost improve temperature and rainfall prediction accuracy. • Vegetation–mortality feedback and phase transitions drive outbreak dynamics. • Global sensitivity analysis identifies vegetation and egg survival as key drivers. • Framework enhances early warning and intervention for desert locust management. [ABSTRACT FROM AUTHOR]
ISSN:03043800
DOI:10.1016/j.ecolmodel.2025.111363