The paper found that simple agricultural interventions led to an increase of more than 60% in cassava yields and a spill-over effect of improved cassava variety uptake in non-intervention zones. Household surveys conducted at the end of the 8 year project implementation period revealed that households that received agricultural interventions had more favorable attitudes toward forest zoning and conservation. The surveys also showed that farmers in the intervention domain practiced less land-intensive field and fallow management strategies compared to those practiced in the non-intervention domain.
However, an 18 year time series analysis of Landsat satellite data revealed that agricultural expansion persisted in areas both with and without intervention assistance, and there is risk of a rebound effect. Approximately 70% of the tree cover loss that occurred outside of the agricultural areas was located within a 3 km buffer zone surrounding the outermost edges of the agricultural areas, which suggested that the majority of tree cover loss was caused by agricultural expansion. Within that 3 km buffer, average annual tree cover loss during the post-intervention period was higher in the intervention domain compared to the non-intervention domain (0.17% yr−1 compared to 0.11% yr−1 respectively, p < 0.001), suggesting risk of a rebound effect. The disconnection between household perceptions of zoning adherence and actual behavior indicates the importance of strengthening governance structures for community-based monitoring and enforcement.