Satellite derived urbanization and vegetation indices predict divergent seasonal distributions of Aedes aegypti and Aedes albopictus Larvae in an urban landscape of central India.

سال انتشار: 1405
نوع سند: مقاله ژورنالی
زبان: انگلیسی
مشاهده: 23

فایل این مقاله در 18 صفحه با فرمت PDF قابل دریافت می باشد

استخراج به نرم افزارهای پژوهشی:

لینک ثابت به این مقاله:

شناسه ملی سند علمی:

JR_JADT-20-3_008

تاریخ نمایه سازی: 22 آبان 1405

چکیده مقاله:

Background: Urbanization and vegetation strongly shape the ecology of the dengue vectors Aedes aegypti and Ae. al­bopictus, yet their fine-scale seasonal distributions in rapidly growing Indian cities remain poorly characterised.Methods: We integrated six satellite and terrain-derived environmental predictors with longitudinal larval data from ۲۹ sites in Bhopal, central India. Seasonal larval distributions and environmental gradients were represented using inverse distance weighting (IDW) and analysed with exploratory Random Forest (RF) regression and SHAP. Environmental associations were further evaluated using partial least squares regression (PLSR) and LMG hierarchical partitioning, while leave-one-site-out (LOSO) validation assessed the complete IDW-RF spatial reconstruction workflow and Mo­ran’s I assessed residual spatial autocorrelation.Results: Exploratory RF/SHAP identified species- and season-specific patterns. Aedes aegypti showed stronger associa­tions with urban-intensity variables, particularly population density and built volume, whereas NDVI variables had the highest relative importance for Ae. albopictus across both seasons; dry-season NDVI was also important for Ae. aegypti. LOSO-RF showed limited out-of-sample performance (R²= ۰.۰۸۹ to −۰.۴۹۶). PLSR leave-one-out cross-validation showed modest monsoon performance (Q²= ۰.۲۳–۰.۲۹) and weak dry-season performance. LMG supported broad spe­cies-level differences in relative predictor contribution, while residual spatial autocorrelation was not significant in three of four PLSR models. Seasonal mapping showed broader monsoon hotspots for Ae. aegypti and more localized distribu­tions for Ae. albopictus.Conclusion: Integrating remote sensing, GIS, exploratory RF/SHAP and complementary multivariate methods provides a practical framework for spatially explicit characterisation of species-specific environmental associations that may in­form vector surveillance.Background: Urbanization and vegetation strongly shape the ecology of the dengue vectors Aedes aegypti and Ae. al­bopictus, yet their fine-scale seasonal distributions in rapidly growing Indian cities remain poorly characterised.Methods: We integrated six satellite and terrain-derived environmental predictors with longitudinal larval data from ۲۹ sites in Bhopal, central India. Seasonal larval distributions and environmental gradients were represented using inverse distance weighting (IDW) and analysed with exploratory Random Forest (RF) regression and SHAP. Environmental associations were further evaluated using partial least squares regression (PLSR) and LMG hierarchical partitioning, while leave-one-site-out (LOSO) validation assessed the complete IDW-RF spatial reconstruction workflow and Mo­ran’s I assessed residual spatial autocorrelation.Results: Exploratory RF/SHAP identified species- and season-specific patterns. Aedes aegypti showed stronger associa­tions with urban-intensity variables, particularly population density and built volume, whereas NDVI variables had the highest relative importance for Ae. albopictus across both seasons; dry-season NDVI was also important for Ae. aegypti. LOSO-RF showed limited out-of-sample performance (R²= ۰.۰۸۹ to −۰.۴۹۶). PLSR leave-one-out cross-validation showed modest monsoon performance (Q²= ۰.۲۳–۰.۲۹) and weak dry-season performance. LMG supported broad spe­cies-level differences in relative predictor contribution, while residual spatial autocorrelation was not significant in three of four PLSR models. Seasonal mapping showed broader monsoon hotspots for Ae. aegypti and more localized distribu­tions for Ae. albopictus.Conclusion: Integrating remote sensing, GIS, exploratory RF/SHAP and complementary multivariate methods provides a practical framework for spatially explicit characterisation of species-specific environmental associations that may in­form vector surveillance.

نویسندگان

Deepanker Das

ICMR-National Institute for Research in Environmental Health, Bhopal, Madhya Pradesh, India, School of Biosciences, Engineering and Technology, VIT Bhopal University, Madhya Pradesh, India

Siddhartha Maiti

School of Biosciences, Engineering and Technology, VIT Bhopal University, Madhya Pradesh, India

Devojit Kumar Sarma

ICMR-National Institute for Research in Environmental Health, Bhopal, Madhya Pradesh, India