Peer-reviewed publications

Wildlife-vehicle collisions are a major threat to biodiversity conservation and human safety, particularly in tropical regions with high wildlife diversity and expanding transportation networks. Despite growing efforts to reduce road impacts on wildlife, there is still limited evidence evaluating the effectiveness of large-scale mitigation strategies in tropical wetlands. Here, we evaluated the spatial and seasonal patterns of wildlife road mortality and assessed the effectiveness of wildlife underpasses and associated fencing along 280 km of the BR-262 highway in the Pantanal–Cerrado ecotone, Brazil. Roadkill monitoring conducted over six years recorded 9157 individuals from 238 vertebrate species, including 17 threatened species. Reptiles and mammals were the most frequently recorded taxa, with Caiman yacare, Dasypus novemcinctus, and Cerdocyon thous presenting the highest mortality rates. Using a Before-After-Control-Impact (BACI) design, we found that long fenced sections reduced road mortality of Caiman yacare by 70.4%, whereas no significant reductions were detected for Cerdocyon thous and Hydrochoerus hydrochaeris. Spatial analyses identified 589 overlapping roadkill hotspots, providing the basis for prioritizing mitigation measures in the most critical sections for medium- and large-bodied mammals, which are predominantly threatened species and are also associated with the most severe collisions involving road users. Based on these results, a large-scale mitigation plan was developed incorporating 170 km of fencing, 44 existing bridges, 22 existing underpasses, 10 new wildlife underpasses, seven artificial canopy bridges, six speed traps, eight jump-outs, and 36 rumble strips. Our findings provide one of the most comprehensive evaluations of wildlife mitigation planning in a tropical wetland ecosystem and offer a scalable framework for wildlife-friendly transportation infrastructure in biodiversity-rich regions worldwide.

In recent years, there has been growing interest in applying vision–language models (VLMs) to quantitative remote sensing. This study evaluates whether three commercial VLMs (GPT-4o, GPT-5.5, and Claude Sonnet 4.6) can detect and classify the severity of harmful algal blooms (HABs) from Sentinel-2 satellite imagery of western Lake Erie and compares them against classical machine learning classifiers (Random Forest (RF), Support Vector Machine (SVM), and eXtreme Gradient Boosting (XGBoost)) trained on both a three-band red, green, blue (RGB) composite representation of the imagery and a 10-band multi-spectral reflectance representation. Forty bloom events identified from the National Oceanic and Atmospheric Administration (NOAA) Harmful Algal Bloom Operational Forecast System (HAB-OFS) severity assessments were assembled into the evaluation dataset, spanning seven bloom seasons (2019–2025). For binary bloom detection, the VLMs did not match the classical RGB classifiers; their F1 scores (0.69–0.75) fell below the best RGB classifier (Random Forest, 0.76) and below a trivial always-present baseline (F1 = 0.77), and they carried false positive rates of 73–93% on bloom-absent images, against 27–40% for the RGB classifiers. The VLMs reached high recall by labeling most scenes as bloom-positive, which makes them operationally unreliable in this configuration. For severity classification, the VLMs assigned 60–70% of their predictions to the “moderate” category regardless of actual conditions and identified at most one of the two severe blooms, whereas the classical classifiers tracked the ground-truth distribution and delivered two to nearly three times the exact-match accuracy (0.44–0.59 vs. 0.20–0.225). The strongest method across all metrics was the multi-spectral SVM (F1 = 0.833, false positive rate 27%, accuracy 0.795). Switching the same SVM from RGB to multi-spectral features raised accuracy from 0.675 to 0.795, a 12-percentage-point gain that measures the spectral information carried by red-edge and shortwave infrared bands that are accessible through multi-spectral sensors but unavailable to standard VLM vision encoders. Feature-importance analysis showed that the multi-spectral classifiers ranked chlorophyll-specific indices, the Normalized Difference Chlorophyll Index (NDCI) and the Floating Algae Index (FAI), among their top predictors, the same signatures used in established operational algorithms, while the RGB classifiers relied on red-channel variability and green-dominant pixel fractions because RGB inputs cannot compute those indices. Two compounded limitations therefore constrain off-the-shelf VLMs for aquatic remote sensing: the limited spectral information available through standard RGB channels and a mismatch between the land-dominated training distributions of these models and aquatic optical conditions. Domain-specific classifiers operating on multi-spectral data remain the more suitable tools for continued development of HAB monitoring and water-quality retrieval.

Livestock production is a major driver of deforestation, carbon emissions, and biodiversity loss in South America, yet the livestock sector is often treated as a monolithic entity, obscuring how diverse actors, from smallholders to agribusinesses, drive distinct social-ecological impacts. To address this gap, we spatially relate livestock systems and deforestation frontiers across the 4.2 million km2South American Dry Diagonal (Chaco, Chiquitano, Caatinga, and Cerrado), one of the world’s most active deforestation regions. We find that distinct livestock systems leave distinct deforestation signatures. Capitalized cattle operations dominate rampant frontiers, where pasture expansion drives active, cohesive clearing. Small-scale cattle and Goat rearing systems concentrate in fragmented frontiers, where deforestation advances slowly and patchily. Industrial-intensive cattlesystems (feedlots and confinement areas often promoted as land sparing) also spatially coincide with rampant frontiers, likely acting as markers of capital-intensive agricultural landscapes where livestock and commodity crops are tightly coupled, rather than direct drivers of clearing. Overlaying these frontiers with Jaguar Conservation Units and corridors reveals contrasting risks for jaguars (Panthera onca), with very different implications for conservation. In rampant frontiers, jaguar populations face habitat loss and corridor severance, requiring protected areas, land-use planning, and zero-deforestation commitments. Conversely, in fragmented frontiers, the conservation challenge is fundamentally social: a high human-wildlife interface that requires community engagement, conflict prevention, and compensation mechanisms. By empirically linking livestock systems to deforestation frontier types, our approach bridges between generic livestock narratives and case-specific accounts of human–jaguar conflict. This offers a way to differentiate intervention contexts at the regional scale, clarifying which actors and frontier dynamics shape a given landscape and so which conservation responses might fit it.

Animal movement data have transformed our understanding of ecological systems and shaped conservation practice, but have limited influence on tracking progress towards international biodiversity goals. Existing biodiversity indicators adopted in frameworks such as the Kunming–Montréal Global Biodiversity Framework — the primary multilateral conservation agreement that aims to halt and reverse biodiversity loss by 2030 — are typically not responsive enough to detect biodiversity change in time to guide action, nor sufficiently biologically informative to explain or predict changes. In this Perspective, we provide seven reasons why movement data can help to tackle these limitations by adding biological realism, mechanistic understanding, and early-warning capacity to current and future indicators. Movement data already inform conservation efforts, from local management to global treaties for migratory species, and are increasingly helpful to uncover sources of environmental change, while enhancing monitoring capability and policy relevance. We recommend that the scientific community ground existing connectivity metrics in empirical movement data, develop new indicators that flag rapid change, and invest in modelling and attribution studies that use movement data to identify drivers of biodiversity loss and recovery.

Developed through the Zofnass Program at the Harvard Graduate School of Design, Climate and Biodiversity in Infrastructure proposes a high-level conceptual framework to holistically capture how integrated climate and biodiversity action is defined at the infrastructure project level. At its core, it proposes a set of criteria to jointly address two of the most significant, concurrent, and interconnected crises of our time: climate change and biodiversity loss. These criteria aim to mainstream biodiversity into climate, and vice versa, as a way to achieve multiple goals.

National governments and multilateral institutions face difficult challenges reconciling biodiversity, climate, and economic development goals. Here we integrate spatial biophysical and economic data with optimization methods to develop sustainable landscape efficiency frontiers that show maximally feasible combinations of biodiversity conservation, land-based climate mitigation, and net economic value from agricultural crops, livestock, and forestry production. We apply this approach in 146 countries and find large potential gains in biodiversity, climate, and economic development from improved land use and land management. Summing national level results show potential to increase climate mitigation by over 200 billion metric tons CO2 equivalents (>20% increase) or net economic value by over 350 billion USD (>80% increase) without loss in other objectives.

Marine protected areas (MPAs) are widely adopted to conserve marine biodiversity, yet their distributional impacts on resource-dependent communities remain underexplored. This study investigates socioeconomic inequality linked to MPAs in eastern Indonesia using a quasi-experimental design and multi-round household surveys of over 10,000 households across 180 coastal settlements in 10 MPAs (2010–2017). We assess poverty through objective (asset-based index) and subjective (perceived economic trends) measures. Findings reveal overall poverty reduction and persistent pre-existing inequalities by gender, age, occupation, and tenure rights. MPAs did not cause short-term economic losses or widen objective inequality but constrained perceived economic improvement, particularly among female-headed households. Community engagement moderated these effects: gender disparities were greater where female participation in groups was low. These findings suggest that promoting inclusive participation can help ensure more equitable conservation outcomes.

Community-based coastal governance is essential for biodiversity conservation and sustainable livelihoods, yet many communities continue to face significant challenges in establishing and sustaining effective governance systems. We sought to improve understanding of the current status of community-based coastal governance across five countries in the Southwest Indian Ocean (SWIO) region of Africa. Using the Elinor tool and data system, we examined how specific governance attributes relate to perceived social and ecological outcomes and sought insights for community-managed areas (CMAs) in the SWIO region. Progress toward strong community-based governance of CMAs has been uneven across the region and across governance attributes. Although some CMAs have established clearly defined management boundaries and property rights, many other CMAs are still working to improve some attributes, such as congruent regulations, transparency, accountability, and operational capacity. Perceived ecological and social outcomes correlated with different governance attributes. Perceived ecological outcomes were positively correlated with measures of property rights and local governance, whereas perceived social outcomes were correlated with local governance and operational capacity. The strength of CMA governance tended to improve in the first 5–10 years after establishment, then governance attributes plateaued or declined thereafter. These findings coincide with typical funding cycles, thus reinforcing the value of continued investment in community-based coastal governance. Specifically, our findings point to the importance of providing long-term capacity development and securing durable conservation finance to ensure the success of CMAs. More broadly, these findings contribute to a growing body of evidence of the importance of paying attention to when and how to leverage different types of support and capacity building.

Despite broad recognition of the building sector’s role in climate mitigation, current efforts remain insufficient, requiring a faster and more comprehensive deployment of climate mitigation strategies (CMS). To that aim, it is critical to understand under which contexts CMS are technically possible, and economically and socially viable. While much research exists on the building sector, many of the studies focus on specific world regions, interventions, or isolated feasibility dimensions, leaving limited understanding of whether these strategies are feasible to implement at the necessary pace and scale across diverse geographic and socio-economic contexts. This article systematically reviews and describes the peer-reviewed literature to characterize the distribution of knowledge on where, how, and under what conditions CMS can be deployed and scaled in the building sector to reduce greenhouse gas emissions in terms of geographical scope, building typologies, and dimensions of feasibility (technical, economical, institutional, socio-cultural). We find knowledge clusters on new low energy building technologies and appliances (focused on Europe and Asia) also in combination with renewable energy sources. Almost ninety percent of the knowledge base focuses on technical mitigation potential, alongside economic potential. However, less research is focused on socio-cultural and institutional dimensions that support CMS implementation. These findings highlight a concerning gap in knowledge on the factors that enable or inhibit CMS implementation. To fill this gap and inform necessary actions in the building sector, we recommend focusing future research on all feasibility dimensions from non-technical solutions and sufficiency interventions, socio-cultural and institutional feasibility across all strategies, in developing countries which are vastly understudied. In addition, research needs to cross silos – enabling evaluations of multiple CMS implemented in packages, rather than examining each individually.

Socio-environmental conflicts associated with large-scale economic projects in Indigenous territories are well documented, but conflicts in coastal areas remain largely unexplored, despite the crucial differences between inland and coastal ecosystems. While the empirical literature about socio-environmental conflicts in Indigenous territories in coastal areas is growing, the role of financial institutions and corporations in these conflicts is often overlooked. This study addresses this gap by examining the financial institutions and corporations associated with more than 400 reported socio-environmental conflicts affecting Indigenous Peoples in coastal areas worldwide. Our results show that most financial institutions supporting projects linked to conflicts are based in West Europe and North America, with North American financial institutions linked to multiple projects. The energy sector accounts for the highest number of socio-environmental conflicts, particularly those involving corporations operating across continents. Many European and North American corporations operate mostly in Africa, but also globally. Corporations with overseas operations are associated with a higher intensity of conflicts. These findings highlight the need to scrutinize financial investments and corporate practices driving socio-environmental conflicts in coastal Indigenous territories, where human rights are often at risk. The lack of due diligence and accountability in applying environmental and social safeguards weakens protections for both ecosystems and Indigenous communities. Strengthening the enforcement of national and international laws on environmental sustainability, human rights and Indigenous rights is therefore essential for coastal development projects.