Scholarlink

<p>Dengue remains a major mosquito-borne public-health threat, and climate change is expected to reshape the geographic suitability of its principal vectors. We developed India-wide species distribution models for <i>Aedes aegypti</i> and <i>Aedes albopictus</i> using a presence–background MaxEnt framework driven by high-resolution (~ 1&#xa0;km) CHELSA bioclimatic predictors. Occurrence records were compiled from GBIF and subjected to transparent quality control and de-duplication on the modelling grid, yielding 301 training presences for <i>Aedes aegypti</i> (363 raw → 341 quality-filtered → 301 thinned) and 180 for <i>Aedes albopictus</i> (224 raw → 210 quality-filtered → 180 thinned). To reduce multicollinearity, predictors were screened using Pearson |r|&gt;0.7 and retained based on permutation-importance prioritization, resulting in 8 predictors for <i>Aedes aegypti</i> and 11 for <i>Aedes albopictus</i>. Model performance was evaluated using 5-fold spatial block cross-validation (1°×1° blocks) and a background sensitivity analysis (5,000/10,000/20,000 points).</p><p>Cross-validated discrimination was stable across background sizes (AUC ≈ 0.746–0.747 for <i>Aedes aegypti</i>; ≈0.721–0.741 for <i>Aedes albopictus</i>), and we additionally report PR-AUC to reflect class imbalance under presence–background modelling. The fitted models were projected under CMIP6 SSP126, SSP370, and SSP585 scenarios to map baseline suitability and potential future shifts. These outputs provide a reproducible, spatially robust evidence base to support forward planning of vector surveillance and targeted control, while recognizing that suitability mapping is an environmental proxy and not a mechanistic transmission model.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dengue vector niche modeling with future SSP climatic scenarios over India using the machine learning-based MaxEnt model

  • 期刊论文

  • 发表日期：

    2026年4月14日

查看全文
  • Subrahmanya Hari Prasad Peri

摘要

Dengue remains a major mosquito-borne public-health threat, and climate change is expected to reshape the geographic suitability of its principal vectors. We developed India-wide species distribution models for Aedes aegypti and Aedes albopictus using a presence–background MaxEnt framework driven by high-resolution (~ 1 km) CHELSA bioclimatic predictors. Occurrence records were compiled from GBIF and subjected to transparent quality control and de-duplication on the modelling grid, yielding 301 training presences for Aedes aegypti (363 raw → 341 quality-filtered → 301 thinned) and 180 for Aedes albopictus (224 raw → 210 quality-filtered → 180 thinned). To reduce multicollinearity, predictors were screened using Pearson |r|>0.7 and retained based on permutation-importance prioritization, resulting in 8 predictors for Aedes aegypti and 11 for Aedes albopictus. Model performance was evaluated using 5-fold spatial block cross-validation (1°×1° blocks) and a background sensitivity analysis (5,000/10,000/20,000 points).

Cross-validated discrimination was stable across background sizes (AUC ≈ 0.746–0.747 for Aedes aegypti; ≈0.721–0.741 for Aedes albopictus), and we additionally report PR-AUC to reflect class imbalance under presence–background modelling. The fitted models were projected under CMIP6 SSP126, SSP370, and SSP585 scenarios to map baseline suitability and potential future shifts. These outputs provide a reproducible, spatially robust evidence base to support forward planning of vector surveillance and targeted control, while recognizing that suitability mapping is an environmental proxy and not a mechanistic transmission model.