Key points:
- Deforestation and climate change are reducing rainfall in the Amazon, leading to reduced river flows on key transport routes for commodities such as soy.
- We estimated the economic risks to Europe-bound soy exports through two key inland river ports, Manaus and Santarém, this century under two emissions scenarios associated with high land-use change and deforestation.
- By mid-century, between USD 300 million and almost USD 1 billion in soy trade annually would transit these ports during periods of critically low river levels: conditions under which vessel navigation could be constrained or delayed.
- On the upper end, this represents more than a third of the USD 2.4 billion in Europe-bound soy that transits these ports annually that could be exposed to disruption from declining river levels.
- Three of the world’s largest agricultural commodity traders, including Cargill, Amaggi and Bunge, could be exposed to up to USD 440 million in soy trade risk annually from port disruption by the 2060s.
- Recent cuts to climate and development aid in major economies risk reducing support for forest protection despite the visible economic consequences to global trade.
- Amazon deforestation, which the EU Deforestation Regulation aims to curb, disrupts the river trade that soy supply chains depend on, offering a quantifiable financial case for European importers to support the regulation.
Deforestation is worsening climate impacts in the Amazon, with knock-on effects for global supply chains
Rivers are an important means of transport for agricultural, timber and mineral products in the Brazilian Amazon. However, when water levels are low, rivers can become unnavigable for barges, requiring agricultural products to be rerouted through alternative ports. This extra journey time causes delays and increased expenses.
Recent climatic events have highlighted this risk. Throughout 2023-2024, climate change caused exceptional drought in the Amazon River Basin, further amplified by El Niño. One impact of this was drained waterways in the region’s largest city, Manaus, disrupting grain exports. Brazil’s waterway transport company, Hidrovias do Brasil, reported a net loss of more than USD 100 million in 2024,1 Or BRL 622 million. The figure is based on a 2024 exchange rate of USD 0.1863 to BRL 1. as drought disrupted key navigation routes. Declines of 33% in cargo volumes and 37% in revenue in its southern corridor operations contributed to the loss. Also in 2024, the Madeira River – the largest tributary of the Amazon – fell to a record low, forcing port closures, reducing cargo transport by 60%, and more than doubling travel times along a key Amazon shipping route. During the 2023 drought, river transport capacity in Amazonia was reduced by 40-50%, and transport tariffs increased by an estimated 20–50%.Â
While climate change was primarily to blame for the drought, deforestation also plays a part. Strong scientific and institutional evidence increasingly supports the conclusion that forest loss makes droughts more intense by reducing moisture cycling.
A 2026 Nature paper found that deforestation in the Amazon has already significantly altered rainfall patterns, with the southern basin experiencing an 8-11% decline in annual precipitation over the past four decades. Up to three-quarters of this reduction is directly linked to deforestation. Climate models tend to underestimate the extent to which deforestation disrupts moisture recycling, meaning that risks to the forest may materialise sooner than expected, according to the paper. The consequences were visible in both 2015-2016 and 2023-2024, when El Niño events combined with North Atlantic warming amplified two of the most severe Amazon droughts on record. In 2023-2024, temperatures were 2-4°C above average and wildfire-affected areas increased by 9%, ​​underscoring how deforestation-weakened hydrological systems amplify the impacts of climate extremes.
Deforestation goals at risk as major economies cut development and climate financing
Global forest loss means the world remains far off track the COP26 target of halting deforestation by 2030, according to the Forest Declaration Assessment. Deforestation rates in 2024 exceeded the target by 63%, largely driven by agricultural expansion and commodity demand such as soy and beef.
The supply chains for these goods are deeply embedded in Amazon transport corridors, linking deforestation directly to production of goods that depend on these corridors to reach their markets. At the same time, deforestation is weakening the Amazon’s hydrological system by reducing rainfall and increasing drought risk. This causes physical disruptions to river transport, from halted port operations to severe navigation constraints.
Despite growing recognition of forests’ role in climate stability and supply chain resilience, financing for forest protection remains far below what is needed. Closing this gap will require investment by 2030 of more than three-fold 2023 levels.
Yet international financing trends are moving in the opposite direction: recent cuts to climate and development aid in major economies such as the UK and France risk reducing support for forest protection and climate resilience in tropical regions, even as the economic and financial consequences of forest loss – through disrupted logistics, trade and supply chains – are becoming more visible.
Box 1: The compounding risks of river flow decline impact trade and could lead to more deforestation
When river flows fall below certain levels, sediment begins to accumulate in river channels, forming sandbars, islands and shallow sections. Without the regular high-flow pulses that would push this sediment downstream, channels can become progressively wider, shallower and more complex, with vegetation eventually establishing on river beds. All of this makes transport more difficult.
Deforestation compounds this risk: as forest cover declines, floodplains lose their capacity to trap incoming sediment. Recent modelling of Amazon floodplain sedimentation found that deforestation causes more sediment to remain in the main river channel, making it shallower and triggering channel migration or location shifts.
These changes have implications for port infrastructure. Restoring navigability requires ongoing maintenance dredging – and the Amazon River network already suffers from inadequate dredging and other navigational challenges.
Rerouting soy from the Amazon River to road transport is not a straightforward alternative to dealing with reduced water flows. Amazonian logistics rely on multimodal combinations of truck, barge and rail to transport goods from production areas to coastal ports. Road alone cannot substitute for the river leg because a purely overland route is either non-existent, in poor condition or vastly longer.
There is a further risk that logistics reorganisation driven by river disruption could accelerate deforestation. When river routes become unreliable, the economic incentive could shift toward opening new agricultural land closer to alternative transport corridors. This same dynamic has historically driven forest clearance along Amazon road infrastructure.
Europe is dependent on soy imported via the Amazon river network
Brazil is a global trade hub for soybean – the world’s largest source of protein for livestock feed. The country accounts for 42% of global production, with around 12% of the crop moving to international markets via barge on the region’s river network. The EU is critically dependent on soy imports, particularly for poultry and pig feed. Imported soybeans account for 84% of demand, with around 50% sourced from Brazil.
Two inland river ports in the Amazon Basin, including Manaus (a central logistics hub) on the Rio Negro and Santarém on the Tapajós River, are key for global soy trade. In 2022, 38.5% of all European-bound Brazilian soy transited these two ports, representing USD 2.4 billion in trade value.2Trade figures represent value not volume of soy as calculated from Trase Earth data. European importers are therefore heavily reliant on infrastructure that sits on a river system projected to experience significantly more frequent drought conditions this century.
While river levels are critical for port operations, their economic implications for Amazon soy exports have not yet been quantified. This gap leaves policymakers and industry stakeholders without a clear understanding of the potential magnitude of climate and deforestation-induced exposure.
Quantifying the financial risk of Europe’s exposure to Amazon-exported soy
In this Zero Carbon Analytics report we estimated the economic risk to Europe-bound3European destinations include: EU member states (Spain, Netherlands, Belgium, Portugal, Italy, Greece), the UK, and other European trading partners (Russian Federation, Turkey, Norway, Gibraltar). soy exports through Manaus and Santarém4The coordinates of the ports were verified using satellite imagery (Manaus: 3.1368°S, 60.0258°W; Santarém: 2.4154°S, 54.7381°W). Data on daily river discharge was extracted at each port location using bilinear interpolation across the four surrounding 0.5° × 0.5° grid cells (approximately 55 × 55 km at these latitudes), consistent with the spatial resolution of the ISIMIP3b discharge outputs. Other major soy export ports operating under tidal or oceanic conditions were not included, as navigability is not likely to be constrained by river discharge changes. using hydrological models and detailed trade data from Trase Earth.5Export volumes and free‑on‑board (FOB) values for soy from Brazilian ports in 2022 were obtained from Trase Earth. The dataset provides, for each port, the tonnes of soy shipped, the total FOB value in US dollars, and the breakdown by destination country. We used these data to estimate the economic value of soy trade that passes through Manaus and Santarém. To assess the risk posed to these ports by climate and land-use changes, we looked at two Shared Socioeconomic Pathways – SSP3‑7.0 and SSP5‑8.5 – both of which assume continued high rates of land-use change and deforestation. This makes them particularly appropriate for assessing river-flow risk in the Amazon Basin where forest cover loss compounds climate-driven drying.6Two global hydrological models were included: CWatM and WaterGAP2‑2e (DOI:10.48364/ISIMIP.230418). Each model was forced with five different global climate models (GFDL‑ESM4, IPSL‑CM6A‑LR, MPI‑ESM1‑2‑HR, MRI‑ESM2‑0, UKESM1‑0‑LL) to capture a range of possible future climate conditions. Simulations were available for the historical period 1981–2014 and for two future scenarios: SSP3‑7.0 (moderate to high emissions) and SSP5‑8.5 (very high emissions) from 2015 to 2100. Together, the combination of two hydrological models and five climate forcings yields a 10-member ensemble, run separately under two emissions scenarios, that captures structural and climate uncertainty. Historical baseline runs used the same models driven by bias‑corrected CMIP6 historical climate simulations.).
Trade at risk is estimated as the annual value of soy transiting these ports during projected low-flow conditions, equivalent to multiplying the 2022 baseline trade value – the latest available in the Trase dataset – by the fraction of the year spent below the discharge threshold needed to maintain waterway navigation, known as Q10.7This approach is a simplified application of the expected value of trade disrupted (EVTD) framework of Verschuur et al. (2025), in which EVTD = P × (T/365) × S × D, where P is hazard probability, T is trade value, S is disruption severity, and D is disruption duration in days. Our estimate sets P = 1 and S = 1, treating all projected low-flow days as full disruption events. Actual trade losses would depend on the degree to which vessels can still navigate at reduced load, whether shipments are delayed rather than cancelled, and the availability of alternative routing. Value at risk is calculated separately for each port before aggregation: Manaus low-flow days are multiplied by Manaus soy trade values and Santarém low-flow days by Santarém soy trade values. The values from the two ports are then summed to produce the combined exposure figure. This avoids applying a cross-port average discharge signal to port-specific trade values. 8We used the Q10 threshold, which is the value below which 10% of daily discharge falls, as a low-flow threshold. This has been used as a threshold in tropical river settings and is recommended for investigating low flows, having been applied across multiple studies; see Mukama et al. (2025) and Van Vliet et al. (2013). Thresholds were first computed per ensemble member (hydro model × climate forcing), then averaged to obtain a single threshold per port (Manaus: 48,967 m³/s; Santarém: 67,073 m³/s). Because each ensemble member has a slightly different baseline discharge distribution, the averaged threshold may not perfectly reflect any single model’s low-flow behaviour.
Our analysis shows that the number of days per year where river discharge falls below historically extreme low-flow conditions is projected to roughly double or triple by mid-century relative to the 1981–2014 baseline, depending on the level of emissions in each model.
Holding 2022 trade volumes constant, soy trade worth between USD 300 million and almost USD 1 billion a year would transit these ports during periods of critically low river levels by mid-century: conditions under which vessel navigation could be constrained or delayed (Figure 1). On the upper end, this represents exposure to trade disruptions for more than a third of the USD 2.4 billion in Europe-bound soy that transits these ports annually.9The wide range reflects uncertainty between the two hydrological models, though both agree on the direction of change: low-flow conditions at these ports will become more frequent and more prolonged across the century regardless of which model or scenario is used. Guimberteau et al. (2017) attribute large uncertainties in Amazon runoff projections to differences between land surface models, Global Circulation Models, and deforestation scenarios.
These figures capture international export exposure only: the same river infrastructure also supports domestic grain flows and essential supplies for riverine communities, meaning the total economic cost of low-flow disruption extends well beyond the trade values presented here. Our analysis also only captures the value of trade exposed to periodic low-flow events but does not account for the cumulative infrastructure costs driven by sustained reductions in river discharge.
Spain and the Netherlands, two of the EU’s largest soybean import and distribution hubs, could see disruptions to soy trade valued at up to 146 million and USD 72 million, respectively, because of low water levels at these ports by mid-century.10With a range of USD 30 million-146 million and USD 12-72 million, respectively. For the UK, up to USD 50 million in soy trade value may be disrupted annually over the same period.
Figure 1
Company-level risk exposure
Some of the world’s largest agricultural commodity traders are active across the supply chain. Among shipments with known exporters, Cargill, Amaggi and Bunge together account for 79.8% of soy trade (or 38% of all soy trade, including shipments with unknown exporters) through these ports to all destinations, not just Europe. Disruptions to river flow could mean these companies have a combined exposure of up to USD 440 million annually by the 2060s. Cargill — the largest exporter of soy through these ports — faces the greatest potential exposure, with up to USD 346 million in soy trade at risk annually by the 2060s (Figure 2). Amaggi and Bunge face exposures of up to USD 60.5 million and USD 30 million, respectively, over the same period, depending on the model. These companies face material financial exposure to the river infrastructure that connects their Amazon supply chain to markets.11Company-level exposure figures are derived from 2022 Trase Earth trade data, which records trade by exporter at each Brazilian port studied. Figures reflect FOB values held constant at 2022 levels and do not account for changes in trade volumes or market share since that date. These figures capture the value of soy transiting these ports during periods of low flow and do not account for rerouting, delays or additional shipment costs.
Figure 2
Protecting forests is essential to protecting trade routes
Efforts to address deforestation-linked commodity trade are underway. Supply-side measures such as the EU Deforestation Regulation (EUDR) will require importers to demonstrate that soy is not linked to post-2020 deforestation based on geolocated supply chains when it comes into force for large operators by 30 December 2026. Meanwhile, voluntary corporate commitments and the Soy Moratorium in Brazil are beginning to shape trade flows away from deforestation.
While these measures address the risks of further deforestation, they do not address the risks from deforestation – such as reduced rainfall and more frequent droughts – that are already affecting the region. The EUDR secures deforestation-free sourcing at the point of production, but it does not address the resilience of the river infrastructure within Brazil that connects compliant producers to export terminals.
Inland river ports exposed to low-flow disruption handle around 38.5% of all European-bound Brazilian soy. The Trase data shows that a substantial share – at least 41.6%12As the ‘zero deforestation commitment’ status of 52.3% of all soy going through these ports in 2022 was categorised as ‘unknown’, the 41.6% confirmed deforestation-free soy represents a minimum bound. – of soy transiting these ports is already covered by zero-deforestation commitments under the Soy Moratorium or corporate supply chain policies. These are the same goods most likely to meet EUDR requirements, meaning some of the most compliant supply chains are also among the most exposed to port disruption.
Physical disruption risk and regulatory compliance risk therefore fall on the same supply chains. As deforestation weakens the rainfall patterns that sustain river transport, disruption risk increases along these corridors.
For Europe’s livestock sector, reduced flows at Manaus and Santarém threaten not only trade volumes but also the reliable supply of compliant feed ingredients European pig and poultry production depends on. This has direct implications for feed costs and could translate into real costs for European producers and consumers. European meat consumption already exceeds sustainable health and environmental benchmarks, and this risk further exposes the fragility of supply chains.
What interventions would work?
Although the EUDR is designed to eliminate deforestation from EU supply chains, its effectiveness varies across different ecosystems. Because the EUDR targets deforestation rather than broader ecosystem conversion, large areas of native vegetation, such as in the Cerrado savannah – which has faced extensive deforestation for soy in recent years – may fall outside its full scope. Current policies don’t address additional risk if deforestation related to production and trade routes moves to other areas such as the Cerrado, which are not covered by the EUDR. These other regions may be prioritised by producers once port infrastructure becomes unreliable.
Additional policies will be needed to support deforestation-free supply chains. Recent analysis shows that when EUDR-style measures are combined with broader policy levers on consumption, production and trade incentives – including shifting diets away from high-impact commodities – modelled EU deforestation footprint reductions reach up to 85.3%, compared with 68.2% from high-ambition EUDR implementation alone.
Aligning deforestation-free sourcing with targeted investment in forest protection, such as the Tropical Forest Forever Facility (TFFF), would reinforce EUDR compliance while safeguarding the physical systems that enable trade.
Amazon forest protection is a necessary condition for protecting the river infrastructure that compliant supply chains depend on, and this analysis puts a number on that dependency.
