To achieve the global ambition to keep global warming below 1.5 °C, considerable actions are needed to reduce greenhouse gas emissions across all sectors. The agriculture sector accounts for about 20% of total emissions. The growing international demand for agricultural commodities is driving up emissions, not just from land-use change, but also from transportation to foreign markets. This insight explores the emissions footprint of five forest-risk commodities, with a focus on how emissions from the maritime transport of commodities compare to those associated with deforestation.
This analysis draws on best available data on maritime transport from the newly available Global Shipping Watch platform and on deforestation emissions from articles by Pendrill et al. of 2019 in Environmental Research Letters and in Global environmental change, as well as Trase. We focus on a subset of commodity-production country contexts – cocoa and coffee from Brazil, Indonesia and Peru; beef and soy from Brazil; and palm oil from Indonesia. Datasets for both emissions from shipping and deforestation are complete in these selected contexts. Exploring differences in emissions can help identify priority commodities, production areas and emissions sources to more efficiently take action to address climate change.
Cocoa deforestation emissions trump shipping emissions
In 2019, nearly 453,000 tonnes of cocoa (calculated in cocoa bean equivalent) were transported by sea from Brazil, Indonesia and Peru to international markets according to shipping records from Global Shipping Watch. The transport was associated with 40,714 tonnes of CO2 emissions (tCO2). Shipping emissions naturally increase with traded volumes. However, a comparison of emissions intensities (shipping emissions per tonne of cocoa (tCO2/t)) shows significant differences across origin markets. Emissions estimates range from 0.04 tCO2/t for cocoa from Brazil, to 0.08 tCO2/t from Peru, and 0.10 tCO2/t from Indonesia.
It is worth noting that, despite these differences, in the three countries, shipping emissions were just a fraction (less than 1%) of the emissions from deforestation associated with cocoa production. Deforestation emissions ranged from 921,000 tCO2 (14tCO2/t) in Brazil to 3.4 million tCO2 (49tCO2/t) in Peru.
Figure. Shipping and deforestation emissions by country and commodity. Bubble size indicates the volume of the commodity traded by sea. Shipping emissions are for trade in 2019, and data come from Global Shipping Watch. Deforestation emission values are averaged across 2016–2018 and come from Pendrill et al. 2019.
Coffee shipping emissions considerably lower than deforestation emissions
The maritime transport of coffee (calculated in coffee beans equivalent) produced in Brazil, Indonesia and Peru was associated with nearly 200,000 tonnes of CO2 emissions, close to five times more than for cocoa. Shipping emissions per tonne of traded coffee were also higher, with coffee from Indonesia producing 0.32 tCO2 per tonne of coffee, Peru followed at 0.13 tCO2/t and Brazil at 0.06 tCO2/t.
Similar to the pattern observed for cocoa, maritime transport emissions make up less than 1% of those from deforestation for coffee from Indonesia (131 tCO2/t) and Peru (20 tCO2/t). Despite exporting the most coffee within the group, coffee production in Brazil was associated with considerably lower deforestation emissions (0.24 tCO2/t). Deforestation emissions, averaged over 2016–2018, were estimated at 668,000 tCO2 per year in Brazil, still about four times those from shipping. Another study, generalised for the entire coffee supply chain, estimated land-use change emissions between 0–35 tCO2/t and averaging at 9 tCO2/t (WWF 2022).
Indonesian palm oil deforestation emissions eclipse those from maritime transport
In 2019, 38 million tonnes of palm oil (calculated in crude palm oil equivalent) produced in Indonesia were transported by sea. Maritime transport produced 1.1 million tCO2, equating to 0.03 tCO2/t. Like other commodities, transportation emissions were less than 1% of deforestation emissions, which were an estimated 373 million tCO2 per year when averaged across 2016–2018, or 8.7 tCO2/t.
Brazilian beef and soy deforestation emissions dwarf those from shipping
We also examined data for beef and soy produced in Brazil, one of the world’s top exporters of the two commodities. The maritime transport of 68 million tonnes of soy (calculated in bean equivalent) was associated with more than 2.6 million tonnes of CO2. Factoring in volume, the shipping emissions intensity for soy was 0.04 tCO2/t, similar to that of cocoa.
In 2019, 2.4 million tonnes of cattle and beef products (in carcass equivalent) were shipped, generating 114,123 tCO2 of maritime transport emissions. Emissions per tonne of cargo varied by product, ranging from 0.07 tCO2/t for frozen boneless beef to 0.35 tCO2/t for live cattle.
Maritime transport emissions are dwarfed by the scale of deforestation emissions associated with the production of soy (94 million tCO2) and beef (509 million tCO2). Findings from a greenhouse gas emissions study from WWF similarly estimated land-use changes for the soy supply chain at an average of 1.6 tCO2/t and from the beef supply chain to range from 0–90 tCO2/t.
Comparing emissions data from deforestation-risk commodities
While the Pendrill et al. dataset provides comprehensive deforestation emissions estimates for many commodities, its global nature means that there are trade-offs with its accuracy. In particular, the dataset uses mostly national-level data, and estimates are based on relative land-use expansion. This means that some deforestation could be attributed to certain commodities incorrectly. There are few alternatives, such as Trase data, that use spatially explicit commodity data at the pixel level to connect forest conversion with specific commodity land uses and are likely more accurate. However, their coverage is limited to a few country-commodity contexts.
To test the sensitivity of our results, we compare deforestation emissions from the Pendrill dataset with those from Trase for the available contexts. For Brazilian beef, the Pendrill dataset tended to overestimate the deforestation area and emissions associated with the production of beef. Using Trase data, deforestation emissions fall to 76 tCO2 per tonne of beef, down from 210 tCO2/t when calculated using Pendrill data. In contrast, Trase data reports higher estimates for deforestation emissions for Brazilian soy. For palm oil produced in Indonesia, Trase estimates 1.6 million ha of palm oil-driven deforestation in 2018 compared to 610,300 ha from Pendrill et al. 2020.
Table. Estimated deforestation emissions and emissions intensities from beef and soy from Brazil. Trase data for soy is for 2017 and for beef is for 2018. *The tCO2/t estimate uses volume estimates from Trase, which differ from those provided by Global Shipping Watch.
Considerably lower shipping emissions underline importance of sustainable land-use practices
The results of this insight show that emissions from the maritime transport of internationally traded commodities tend to remain considerably lower than emissions from deforestation associated with their production (<3% of deforestation emissions), highlighting the importance of sustainable land-use practices and deforestation reduction activities. However, given the urgency of the climate crisis, emissions in all areas must be reduced. This insight provides new evidence that can inform discussions on how to reduce the emissions footprint of internationally traded agricultural commodities, including from their transport. Further research could explore if trends hold across other commodities and origins, and how factors like destination market and commodity characteristics explain differences in shipping emissions for the same commodity from different origins (as seen for cocoa and coffee). Lastly, future research should examine how transport emissions vary by product type, as results from the beef sector suggest this could be an important factor.