Table 2.

Pathways for cocoa re-agro-forestation of pasture land to reduce greenhouse gas emissions. For details on calculations and references see Methods section

Mechanism	Potential carbon additionality	Assumptions	Reversals
1. Avoided deforestation emissions: Due to reduced loss of native forest converted to cocoa as result of forest protection. Conventional cocoa expansion has caused native forest loss in many parts of the tropics	In the order of 75 Mg C ha−1 above and below ground carbon stocks. Calculated as: 135 Mg C ha−1 carbon emissions from forest conversion minus carbon stocks of cocoa agroforests established at site, estimated at 60 Mg C ha−1. Includes 20 % deduction for heterogeneity in deforestation patterns and incomplete conversion	Forest would not be converted or degraded in the absence of cocoa planting	Market leakage due to displacement of cocoa planting to other forest areas if not applied at national, regional or global scale. Unsatisfied cocoa demand is met through intensification of existing farms
2. Re-agro-forestation sequestration: establishment of cocoa agroforests on previously cleared pasture land	In the order of 60 Mg C ha−1 above and belowground carbon stocks, highly dependent on practices used	Cocoa agroforest is managed to reach and maintain time-averaged carbon level of 60 Mg C ha−1. No emissions from removal of secondary vegetation. Planted area would not otherwise revert to forest. No net carbon flux from soils	Net emissions could occur if cocoa agroforests displace restoration of native secondary forests which may store more carbon. For displacement of cattle see (5)
3. Avoided fire emissions: better fire management on agroforestry farms to avoid damage to cocoa may reduce escaped fires in nearby native forests	Unknown but could be significant. The extent of escaped fires in the Amazon has high annual variability and can result in 10-80 % emissions of forest carbon stocks (Alencar et al. 2006; van der Werf et al. 2009; Pütz et al. 2014)	Re-agro-forested areas are located in proximity to natural forest, ideally along forest boundaries	None
4. Absorption of avoided deforestation leakage: to the extent that cocoa has larger labor demand per hectare than ranching, cocoa agroforestry can absorb local labor leakage from other avoided deforestation strategies	Unknown, depending on the extent of leakage which can range from <10 % to >90 % (Murray et al. 2004)	Displaced workers from cattle, logging or other sectors linked to deforestation are willing to work in cocoa	If cocoa expansion attracts additional labor to the region, a future slump in the cocoa economy could release labor and trigger additional deforestation
5. Changes in per-hectare emissions of greenhouse gases from agricultural and pasture management (carbon footprint)	Highly dependent on local circumstances. Positive impact if cattle numbers are reduced or management of remaining pasture area intensified	Soil fertility after pasture use is not so strongly degraded as to require large amounts of mineral fertilizer	Fertilizer application to cocoa causes greenhouse gas emissions but is small if naturally fertile soils are planted