Modeling deforestation and examining trade-offs among carbon stocks and benefits to livelihoods and biodiversity for REDD+ planning in Kenya

Thesis

Reducing deforestation and degradation (REDD+) can lower anthropogenic carbon emissions by 9-11% and is therefore integral to mitigating climate change. REDD+ is a proposed scheme in which developed countries pay developing countries for reduced deforestation. Kenya is one such country participating in REDD+. As such, Kenya must establish a reference emission level (REL) against which it can compare avoided emissions from reduced deforestation. Developing a spatial model of future deforestation will determine the REL for Kenya and provide complementary information on where forests are at risk and drivers of deforestation. REDD+ also offers benefits of poverty alleviation and biodiversity conservation, referred to as "co-benefits." Since the spatial distributions of carbon and co-benefits may not be congruent, there may be trade-offs among them whereby increasing one decreases others. To assist Kenya in its REDD+ preparations, I build the first spatially explicit deforestation model for the whole of Kenya parametrized with national data. I apply a novel Bayesian inference method which propagates uncertainty in the estimated parameters to the predictions. In addition, I conduct a novel analysis of trade-offs among carbon and co-benefits in three dimensions. I find that predicted forest loss is 1.68% ± 0.06% from 2013 to 2018 (cross-validated AUC = 0.8). There is high deforestation in the Rift Valley and coastal regions attributed to drivers related to accessibility, extraction, and agriculture. Analysis of three-dimensional trade-offs reveals that REDD+ sites could simultaneously include 86% of the maximum possible number of people in poverty and 88% of the maximum possible average proportion of ranges of rare species with the smallest 25% of global ranges with a 10% reduction in carbon. The results presented here may help Kenya spatially target REDD+ interventions and address particular deforestation drivers as well as plan for securing high levels of carbon stocks and co-benefits.

Authors: Formica, A

• DOI: 10.5287/ora-5jxbovak4
• Type of award: MPhil
• Awarding institution: University of Oxford
• Language: English
• Keywords: co-benefits; deforestation; REDD+; spatially-explicit model; biodiversity; uncertainty; poverty; trade-offs; Kenya
• Subjects: Geography