Transitioning towards renewable energy (RE) is a central pillar of climate change mitigation. To meet globally agreed-upon climate targets, investments in RE need to more than double from $624 billion in 2024 to $1.4 trillion per year over the 2025-2030 period. While beneficial for the environment in terms of carbon reduction, these investments can involve environmental costs if solar, hydro, or wind plants destroy habitats or fragment ecosystems. This research constructs a novel global measure of the land-use-induced biodiversity impacts of RE plants to evaluate these trade-offs.

Data

The study utilizes a sample of 47,616 RE plant units across 190 countries, consisting of 13,556 hydro, 17,715 solar, and 16,345 wind units. These plants have a total capacity of approximately 1.61 terawatts (TW), which accounts for roughly 95% of the total utility-scale RE capacity in the S&P Market Intelligence (S&P MI) database by 2025. Satellite imagery from the Sentinel-2 mission and NASA’s Landsat programme was employed to identify and quantify land-use changes caused by construction. The research focused on plants that became operational between 2000 and 2025.

Methodology

To measure land-use-induced biodiversity impact, the researchers rely on the Species Threat Abatement and Restoration (STAR) score. Specifically, they employ the Threat Abatement STAR (START) subscore, which quantifies at a granular 5 by 5 kilometre level how much extinction risk could be reduced if current threats were removed. Land-use footprints were identified using remote sensing data or capacity-based approximations and spatially matched with START data. The analysis excludes offshore wind farms due to a lack of comprehensive underwater biodiversity data and sets the impact of rooftop solar to zero as it does not displace natural habitat.

Evidence on biodiversity impacts of RE plants

RE plants generate an aggregate biodiversity impact, calculated as the total START score, of about 5 billion. Solar and hydro account for nearly 90% of this overall impact, whereas wind contributes relatively little (7.2%). Solar plants have a large impact due to their extensive land requirements; the most common land use for solar is grass land (29.1%). In contrast, hydro plants (52.9% of total impact) are frequently sited in biodiversity-sensitive locations, with forested areas being their most common land type (42.2%).

Impacts have risen sharply over the past two decades, primarily due to rising land use per project rather than siting in more sensitive locations. Looking ahead, if planned plants are included, RE’s biodiversity impact increases by 50% to 7.5 billion. Long-term projections by the International Energy Agency suggest the START score could reach 53 billion by 2050, equivalent to the biodiversity value of the entire land area in Canada or Uganda.

Evidence on impact concentration and counterfactual analysis

Biodiversity impacts are highly concentrated and skewed; fewer than 1% of all plants account for nearly 70% of the overall biodiversity impact but provide less than 5% of the generation capacity. The ten plants with the highest START score account for 13.2% of the total impact. The Pachamama II Solar Project in Mexico has the largest overall impact with a START score of 120.8 million. Counterfactual analysis shows that relocating the worst 1% of solar and wind plants to optimal sites within a 10-kilometre radius could reduce their impact by 86.2%. This demonstrates that targeted siting policies could substantially reduce environmental costs without compromising capacity growth.

Evidence on impact heterogeneity across firms

The study examines differences in ownership and financing structures. Plants owned by listed firms have a two-thirds higher absolute biodiversity impact than those of unlisted firms, though this difference disappears after controlling for electricity generation, suggesting a scale effect. However, within the same ultimate parent company, plants of listed subsidiaries have significantly lower impacts than those held by unlisted subsidiaries. This suggests that parent firms may strategically allocate high-impact plants to private subsidiaries to shield them from public scrutiny.

Plants owned by financial firms have a 10% lower biodiversity impact compared to those of non-financial firms, often locating in less sensitive areas. Additionally, project-financed plants exhibit systematically higher biodiversity impacts (about 20% higher overall) compared to plants financed directly on corporate balance sheets. These findings imply that firms may be reluctant to hold high biodiversity risks on their own balance sheets due to investor and regulatory pressure.

Conclusion

While RE is essential for climate goals, its potential to harm biodiversity through habitat loss is significant and concentrated. The research highlights that mitigating the ecological footprint of RE does not require halting the energy transition. Instead, careful exploration of nearby, lower-impact alternatives and rigorous ex-ante biodiversity screening by investors and regulators can yield substantial environmental benefits. The concentration of damage among a few firms and projects suggests that targeted regulatory effectiveness is achievable.