Coca cultivation and deforestation in the Peruvian Amazon: A multiscale analysis using spatial regression models

Abstract

Coca cultivation is expanding rapidly into the Amazon biome, raising concerns about its role in deforestation. While this relationship is well-studied in Colombia, Peru remains understudied in spatial modeling analyses. This study addresses this gap by employing spatial regression models to assess the association between coca cultivation and deforestation across the Peruvian Amazon at the national and subnational level. First, we quantify trends: between 2016 and 2020, coca cultivation expanded by 40 %, from 43,900 to 61,800 ha, while annual deforestation exceeded 250,000 ha. Coca increasingly overlaps with Protected Areas, buffer zones and Native Communities. Second, using a spatial regression framework, we found coca cultivation to be a statistically significant driver of deforestation at the national level, with each hectare of coca associated with an increase of 0.02–0.06 ha of deforestation. Forest integrity showed the strongest protective effect, while elevation and slope also reduced deforestation risk. Contrary to expectations, Protected Areas did not exhibit significant deterrent effects and Native Communities were positively associated with deforestation. Third, we analyzed two subnational hotspots: VRAEM and Callería. In VRAEM, coca showed no significant link to deforestation, while in Callería, this association was positive but only significant in one model specification. Forest fires were a dominant driver in both areas, followed by a modest effect of the agricultural-livestock mosaic. Our results contribute to a nuanced understanding of how coca cultivation drives deforestation in a spatially varying manner, challenging the conventional wisdom that coca cultivation exerts a consistent impact on deforestation irrespective of local context.