Consistent cooling benefits of silvopasture in the tropics

The study investigates whether expanding silvopasture—trees deliberately integrated into pasturelands—can provide meaningful local cooling in tropical regions and thereby help communities adapt to heat. Agroforestry is recognized for carbon sequestration and co-benefits such as improved soil fertility, water availability, biodiversity support, and food security, yet its large-scale cooling services have not been quantified. Heat exposure already harms labor productivity, human health, and livestock in many tropical areas, with impacts projected to worsen. The authors focus on silvopasture because livestock heat stress is increasing, temperature regulation is a known benefit of silvopasture, and pasturelands are common targets for increasing tree cover. The analysis targets the tropical Americas and Africa (excluding tropical Asia due to limited silvopasture) and aims to answer four questions: (1) What are the cooling benefits of silvopasture across tropical pasturelands? (2) Do cooling benefits depend on spatial extent (patch size) or intensity (woody carbon density)? (3) Where are potential cooling benefits highest? (4) How much woody carbon is required to achieve these benefits? The study uses woody carbon density and satellite-derived surface temperature data to quantify cooling relative to intact forests at the same latitude.

Prior work shows agroforestry stores significant carbon globally and can offer co-benefits including improved soil fertility, water access, food security, and biodiversity support. Agroforestry has been identified as a potential contributor to climate mitigation and adaptation goals, including NDCs and SDGs. Trees cool locally via shade and evapotranspiration; preventing tropical deforestation is known to reduce warming and protect worker productivity, but less is known about cooling from adding trees to agricultural lands at large scales. Reviews note a lack of broad-scale quantification of agroforestry’s cooling benefits. Heat exposure negatively affects labor productivity, health outcomes, and livestock performance, underscoring the importance of potential cooling services in tropical rural settings where access to air conditioning and water is limited. The study builds on these insights by providing a continental-scale assessment of silvopasture’s cooling potential.

Data sources: Aboveground woody biomass density on pasturelands was taken from Chapman et al., partitioned specifically for agroforestry on pasture and agricultural lands. Biomass was converted to woody carbon density by multiplying by 0.47. These data were aggregated from 30 m to 1 km pixels by averaging biomass over pastureland and weighting by the pastureland fraction within each 1 km square. Only pixels with >50% pastureland were included. Biome masks from Dinerstein et al. were used to exclude montane grass and shrublands, deserts and xeric shrublands (where silvopasture is not viable or has biodiversity risks), and mangroves. Temperature data were annual mean daytime land surface temperatures (LST; ~1:30 PM local time) from the MODIS Aqua satellite for 2018 at 1 km resolution.

Forest Equivalent Temperature (FET): To quantify cooling, FET was defined as the difference between a pixel’s annual mean daytime temperature and the average annual temperature of intact tropical/subtropical dry and moist broadleaf forests at the same latitude (controls selected for homogeneity and low variability). Negative FET indicates cooler than intact forests, positive indicates warmer. Using a single year (2018) introduces interannual variability noise, but the ENSO state was neutral, limiting bias.

Analyses: Pixels were binned by woody carbon density to examine FET distributions. Linear regressions quantified the relationship between FET and woody carbon density separately for the Americas and Africa. A flood-fill algorithm identified contiguous silvopasture patches (sharing at least one boundary) to compute within-patch mean FET and woody carbon density. Patches were categorized by area: small (<10 km²), medium (10–33 km²), large (33–100 km²), and extra-large (>100 km²). Correlations between patch size and within-patch FET, and between within-patch woody carbon density and FET, were assessed, with best-fit slopes reported by patch class (Table 1). To evaluate climate change context, projected local warming to 2050 was derived from 24 CMIP6 models under SSP5-8.5, comparing 2045–2055 to 2015–2025.

Action maps: Using ΔT = mΔC (m = slope of FET vs carbon density; −1.11 °C per 10 tC/ha for the Americas, −0.83 °C per 10 tC/ha for Africa), the additional woody carbon needed to counteract projected local warming was estimated. A more realistic scenario was mapped by increasing only those pasture pixels with current woody carbon density below the biome median (for silvopasture >5 tC/ha) up to the biome median. The resulting FET change was divided by projected warming to yield the fraction of warming offset. Results were summarized by region and country. Uncertainties considered include single-year temperature data, zonal inhomogeneities, and satellite product uncertainties.

• Cooling increases with woody carbon density. Silvopasture pixels are cooler on average than low- or no-woody-carbon pasturelands; FET declines as woody carbon increases, with distributions shifting toward intact forest temperatures.
• Linear relationships: Best-fit slopes show FET decreases by 1.11 °C per 10 tC/ha in the Americas and 0.83 °C per 10 tC/ha in Africa (p < 0.001 for both), despite substantial FET spread, indicating robust biophysical cooling effects across regions.
• Patch size dependence is weak. In the Americas, patch area and within-patch FET are weakly negatively correlated (r = −0.07, p < 0.001), implying slightly cooler larger patches; in Africa, correlation is weakly positive (r = 0.05, p < 0.001). An order-of-magnitude increase in patch area changes FET by only −0.79 °C (Americas) or +0.29 °C (Africa). Thus, cooling does not require large contiguous patches.
• Within-patch density matters most. Across patch-size classes, mean within-patch woody carbon density is negatively correlated with FET (p < 0.001). Slopes range from about −1.11 to −2.84 °C per 10 tC/ha depending on region and patch class; strongest effects are often in smaller patches (e.g., Africa small patches: slope −2.37 °C per 10 tC/ha; Americas small/medium: −2.69/−2.84 °C per 10 tC/ha; large and extra-large in the Americas not significant).
• Current stocks and requirements: Existing silvopasture woody carbon is ~0.20 GtC in the Americas and ~2.55 GtC in Africa. To offset projected 2050 warming fully under high emissions, average additions of ~21 tC/ha (Americas) and ~18 tC/ha (Africa) would be needed—an upper bound that may exceed ecological or productivity constraints in some systems.
• Realistic expansion potential: Increasing low-density pasture pixels to their biome median silvopasture woody carbon yields mean cooling of 0.59 °C (IQR 0.48–0.67) in the Americas and 1.06 °C (IQR 0.95–1.29) in Africa, and additional carbon storage of ~0.49 GtC (Americas) and ~4.79 GtC (Africa). The Sahel shows particularly high potential, with >50% of projected 2050 warming offset in many areas. Countries with the greatest total storage potential include Sudan, Chad, and Somalia.
• Context: CMIP6 models project local warming to 2050 of ~0.51 °C/decade (Americas) and ~0.56 °C/decade (Africa) under SSP5-8.5, highlighting the relevance of silvopasture cooling for adaptation.

The results demonstrate that increasing woody carbon density in silvopasture provides substantial and scalable local cooling across tropical pasturelands, directly addressing the research questions. Cooling benefits correlate strongly with woody carbon density rather than patch size, implying that even smallholders can achieve meaningful temperature reductions by intensifying tree cover rather than expanding contiguous areas. This has important implications for human and livestock heat stress, potentially improving labor productivity, reducing heat-related illnesses and injuries, and mitigating livestock performance losses in increasingly hot climates. The conservative FET metric does not capture additional shade-related reductions in ground-level temperatures, so realized on-the-ground cooling may be larger. These findings support integrated strategies that advance climate adaptation and mitigation (via carbon storage) alongside biodiversity and development goals, aligning with policy mechanisms such as NDCs and SDGs. Action maps identify hotspots where silvopasture expansion could offset a substantial fraction of projected warming, guiding targeted interventions and investments. Overall, the study suggests that silvopasture intensification is a practical, distributed, and socially beneficial pathway for climate resilience in rural tropical regions.

This study provides the first continental-scale quantification of silvopasture’s cooling benefits across the tropical Americas and Africa, establishing robust linear relationships between woody carbon density and local cooling and showing that benefits are largely independent of patch size. Realistic increases in woody carbon to biome median levels could offset a significant fraction of projected mid-century warming while sequestering substantial carbon, with notable potential in the Sahel. The results can inform policies and programs that aim to simultaneously address climate adaptation, mitigation, biodiversity conservation, and rural well-being. Future research should: (1) evaluate cooling effects in croplands, (2) assess the influence of tree spatial arrangement and species on cooling, (3) incorporate belowground/soil carbon dynamics, (4) validate satellite-based metrics with field measurements of near-surface air temperature and thermal comfort, and (5) use multi-year, higher-resolution datasets to reduce uncertainty and capture interannual variability.

Key limitations include: (1) Temporal constraints: Only one year (2018) of MODIS daytime LST was used; lack of silvopasture initiation dates precluded before–after analyses. (2) Scope: Analysis was limited to pasturelands; results may differ for croplands. (3) Metric: FET is based on canopy/surface temperature relative to intact forests at the same latitude and serves as a proxy for near-surface air temperature; shading effects likely mean on-the-ground cooling is underestimated. (4) Spatial and ecological filters: Certain biomes (montane grass/shrublands, deserts/xeric shrublands, mangroves) were excluded; ecological constraints may limit feasible woody carbon increases in some regions. (5) Spatial heterogeneity and uncertainty: Zonal inhomogeneities and interannual climate variability introduce noise; satellite-derived temperature and biomass products carry inherent uncertainties, though not expected to bias results. (6) Practice specificity: The framework uses carbon stocks and does not differentiate silvopasture designs or species that may yield varying cooling or productivity trade-offs.