Solar geoengineering could redistribute malaria risk in developing countries

The study addresses whether and how solar geoengineering, specifically stratospheric aerosol injection (SAI), would affect human health through changes in malaria transmission risk. While climate change is known to impact health and vector-borne diseases, the health consequences of SRM remain largely unexplored. Malaria is a high-priority case due to its immense burden and a temperature-response that peaks around 25 °C, making it sensitive to cooling or warming. The authors hypothesize that SRM designed to offset greenhouse-gas-induced warming could have heterogeneous effects: potentially reducing malaria encroachment at high elevations (e.g., in East Africa) but increasing suitability in lowland tropics and parts of Asia by maintaining or returning temperatures to ranges optimal for transmission. The purpose is to quantify these potential redistributions of risk under realistic SRM scenarios compared to equivalent warming without SRM, thereby informing policy on potential health trade-offs and climate justice implications.

Prior work on SRM’s health impacts is sparse, with limited studies on air quality, UV-B exposure, and urban heat stress under aerosol injection. Climate change is consistently projected to alter vector-borne disease risks, with robust links between temperature and transmission for mosquito-borne pathogens via unimodal thermal response curves. Previous projections indicate substantial increases in populations at risk for dengue and Zika by mid-century and shifting yellow fever burden, but malaria has the largest global burden. For malaria, evidence suggests warming will expand Plasmodium falciparum into southern and high-elevation East Africa and possibly reduce transmission in central Africa and the Sahel, though uncertainties remain. Plasmodium vivax, responsible for a large share of cases outside Africa, has been underrepresented in climate studies, with evolving understanding of its thermal ecology linked to vector species (e.g., Anopheles stephensi). This background motivates examining SRM’s potential to alter malaria risk nonlinearly, given malaria’s relatively cool transmission optimum and the likelihood that SRM cools tropical regions disproportionately.

Study design: Compare malaria transmission suitability and populations at risk under two greenhouse gas emissions scenarios with and without SRM through 2070. SRM scenarios: (1) GeoMIP G3 offsets RCP4.5 warming via equatorial SO2 injection ramping from 2020 to stabilize global mean temperature at 2020 levels; (2) GLENS offsets RCP8.5 warming via four injection sites (30°N, 30°S, 15°N, 15°S) starting in 2020 to stabilize global mean temperature at 2020 levels, aiming to avoid overcooling the tropics. Climate models and data: G3 and RCP4.5 from HadGEM2-ES; GLENS and RCP8.5 from CESM1(WACCM). Three ensemble members per scenario were used to capture internal variability. Outputs (near-surface daily air temperature; monthly precipitation) were regridded to 1° resolution via bilinear interpolation. By 2070, HadGEM2-ES projects ~2.8 °C warming above 1850–1900 and CESM1(WACCM) ~3.3 °C, with SRM stabilizing at ~2020 levels (~1.1 °C above preindustrial). Malaria transmission modeling: Use temperature-dependent R0(T) framework to derive scaled transmission suitability (0–1) from laboratory-parameterized, life-history explicit models integrating: biting rate (a), vector competence (c), parasite development (PDR), mosquito development (D), egg-to-adult survival (s), adult mortality (u), and egg laying (E). Suitability is determined by daily mean temperature relative to empirically-derived bounds and optimum. Regional transmission systems and thermal bounds: - Sub-Saharan Africa (P. falciparum via An. gambiae complex): 17–34 °C bounds (canonical estimates). - South Asia (P. vivax via An. stephensi): 15.7–32.5 °C. - Americas (P. vivax; multiple vectors): 19.4–31.6 °C (assumed comparable to An. gambiae thermal ecology due to limited species-specific data). Environmental mask: Exclude deserts (≤250 mm annual precipitation) per projected precipitation to account for vector habitat constraints. Mapping suitability and seasonality: Count number of days per grid cell within transmission bounds annually; classify risk as unstable/epidemic (31–179 days/year) or stable/endemic (≥180 days/year). Comparisons: (1) 2070 SRM scenarios versus corresponding no-SRM climates (G3 vs RCP4.5; GLENS vs RCP8.5); (2) 2070 SRM versus present day (2020). Populations at risk (PAR): Regions based on modified Global Burden of Disease groupings; modeled falciparum in West, East, Central, Southern Africa; vivax in East, South, Southeast Asia; and Central, Tropical, Andean Latin America. Pair RCP4.5/G3 with SSP2 and RCP8.5/GLENS with SSP5 using SEDAC 1/8° urban/rural population projections per decade (2010–2100). Rasterized regional boundaries and summed populations in pixels meeting thermal and precipitation criteria within risk strata. Sensitivity and uncertainty: Used three ensemble members per scenario; acknowledge model-selection uncertainty and high variability in southern Asia outcomes. Descriptive healthcare overlay: Mapped Malaria Atlas Project prevalence changes (2000–2019) against 2020 R0(T) to qualitatively indicate where interventions may decouple suitability from burden (no direct incorporation into projections).

• SRM alters malaria suitability heterogeneously across the tropics. - Under SRM, high-elevation regions (e.g., Ethiopian highlands, East African Rift, Andes) show reduced suitability compared to no-SRM futures, potentially protecting populations from encroachment. - SRM could increase suitability in many lowland tropical regions by maintaining temperatures closer to malaria’s optimal range: notably parts of the Amazon basin (Brazil, Peru, Ecuador, Venezuela), Indonesia, west and central Africa, and southeast/Atlantic coasts of Africa. - In RCP8.5 without SRM, several of these regions warm sufficiently that average suitability declines toward near-zero by 2070; SRM (GLENS) reverses this, cooling up to ~1 °C relative to present, returning conditions toward near-optimal for transmission. - Compared to extreme warming (RCP8.5), SRM by 2070 would nullify a projected reduction of nearly one billion people at risk of malaria, effectively maintaining higher PAR than under no-SRM warming where heat would otherwise reduce suitability. - Relative reductions with SRM are projected in the Indian subcontinent and the Sahel; however, outcomes in southern Asia show high inter-model variability, limiting confidence in precise regional PAR differences. - Overall, SRM produces a patchwork of regional winners and losers; at best net neutral, and at worst increases malaria risk relative to no-SRM climate change in some regions.

The findings indicate that SRM is not guaranteed to improve health outcomes uniformly. Because malaria transmission responds nonlinearly to temperature with a relatively cool optimum, stabilizing or cooling tropical temperatures can elevate suitability in currently warm lowland regions while benefiting cooler highlands. Thus, SRM may create trade-offs: local benefits in East African highlands but increased risk in parts of west/central Africa, southern Asia, and the Americas. These heterogeneous effects challenge the notion that SRM would reduce climate injustice; instead, it could shift or maintain disease burdens in lower-income tropical regions often excluded from geoengineering decision-making. Given many health outcomes have nonlinear climate responses, optimizing SRM for health is difficult, and benefits will be context-specific. The study underscores the need for case-by-case health impact assessments, disaggregated by region, and for meaningful engagement of developing countries in SRM research and governance. It also highlights broader uncertainties from health system changes (e.g., post-COVID-19 recovery) that will modulate future malaria burden independent of climatic suitability.

This work provides an initial quantitative assessment that solar geoengineering designed to offset warming could redistribute malaria risk, with mixed regional outcomes and potential global net effects ranging from neutral to adverse relative to no-SRM futures. By 2070, SRM could negate a projected near–one-billion-person reduction in PAR that extreme warming would otherwise produce, while offering localized benefits in high-elevation regions. These results caution against assuming SRM will alleviate health inequities and emphasize integrating health considerations into SRM design and policy with leadership from developing countries. Future research should: incorporate fuller transmission dynamics and additional climate drivers (e.g., precipitation, hydrology), evaluate multiple SRM strategies and offset magnitudes, include multi-model ensembles for climate and disease, account for vector and parasite range shifts and interventions, and consider broader planetary health trajectories and health system capacities.

• Modeling simplifications: The R0(T) framework captures temperature-driven suitability but not incidence or burden; it omits factors such as vector and human population densities, intervention coverage, health system capacity, immunity, diagnostics, and behavior. - Climate variables: Precipitation is only used to mask deserts (≤250 mm/year); explicit hydrology and rainfall dynamics are not integrated into transmission suitability. - Scenario and model dependence: Results rely on two SRM scenarios (G3, GLENS) and two climate models (HadGEM2-ES, CESM1(WACCM)); inter-model structural uncertainty is acknowledged but not fully quantified. - Regional thermal bounds: Thermal limits for P. vivax in the Americas are assumed based on analogy due to limited species-specific data; vector–parasite thermal ecology may vary. - Vector and pathogen dynamics: Potential future shifts in vector ranges, adaptation, or coevolution, including expansion of An. stephensi and P. vivax in East Africa, are not modeled. - Population projections and region assignments introduce uncertainties; southern Asia results show high variability across ensemble runs. - Health interventions: No dynamic modeling of control measures or future elimination efforts; descriptive overlay of prevalence trends is not integrated into projections. - Termination effects and alternative SRM designs (e.g., abrupt deployment or termination) are not assessed.