Identifying predictors of deforestation across Brazilian landscapes

Prof Felix Eigenbrod, Assoc Prof Alessandro Sorichetta, Prof Jean-Paul Metzger (University of São Paulo)

Deforestation in Brazil has global implications for both biodiversity loss and climate change mitigation. Brazil is also globally important for food production as the world’s largest producer of soy, beef, chicken, coffee and sugarcane1. Deforestation is largely driven by expansion of agriculture. However, the causes of agricultural expansion and deforestation vary by location – sometimes it is to meet the demand of a growing population, while at other times it is distant, rather than local demand that drives agricultural expansion. Furthermore, policies designed to protect forests in one location can lead to more deforestation in other areas, while growth of cities can displace crops, leading to deforestation elsewhere. Understanding how drivers of deforestation interact spatially and across scales is critical to sustainably managing agricultural landscapes, but remains poorly understand in Brazil and elsewhere.

In this PhD, you will identify causes of land use change across space and scales in Brazil using statistical modelling, high resolution land cover data and existing ecological, social and demographic data.The results of these analyses will then be combined with projections on threats to biodiversity and population growth to provide critical evidence for policy makers on sustainable management of the varied landscapes of Brazil.


The originality of the project is that you will be using  the excellent, freely available Mapbiomas 35 year land cover time series for Brazil (1985-2019; 30 m x 30m resolution), along with socio-economic and population datasets to test specific hypotheses about the effects of land use change within and across landscapes on  deforestation in Brazil using a before-after-control-impact (BACI) analytical approach. Initial work will focus on pairing similar landscapes (municipalities) that underwent rapid expansion of a key commodity (sugarcane) between 2006 and 2010 (Impact) with ones that did not (Control). You will then quantify changes in deforestation both before and after this time period both within the paired landscapes, as well as other adjacent landscapes. This will enable you to test how land use decisions at a landscape (municipality) scale indirectly affect changes in other landscapes (Chapter 1).You will also look at other potential causes of deforestation (i.e. urbanization, expansion of cattle ranching) using a similar approach (Chapter 2). This will enable you to identify critical spatial linkages between landscapes, and use this to identify more sustainable pathways for managing Brazil’s globally important forests (Chapter 3). Your project will benefit from our foundational work applied both globally2 and in Brazil3.

University of Southampton

The INSPIRE DTP programme provides comprehensive personal and professional development training alongside extensive opportunities for students to expand their multi-disciplinary outlook through interactions with a wide network of academic, research and industrial/policy partners. The student will be registered at the University of Southampton and hosted at the School of Geography and Environmental Science. Specific training will include advanced statistical and spatial modelling skills in R and GIS, programming and data management skills in R, expertise in cloud computing (Google Earth Engine). The student will also be trained in systematic analyses of the broad interdisciplinary literature in land use theory and landscape ecology relevant for the project, as well as training in academic publishing. In addition, the student will gain international experience through a visit to Metzger’s lab at the University of São Paulo, with a particular emphasis on forest conservation policy in Brazil.

Eligibility & Funding Details: 

Please see for details.

Background Reading: 

1. Lapola et al. 2014 Pervasive Transitions of the Brazilian Land-Use System. 2014. Nature Climate Change

2. Eigenbrod et al. Identifying Agricultural Frontiers for Modeling Global Cropland Expansion. One Earth, 2020

3. Ruggerio, … and Metzger. Election cycles affect deforestation within Brazil's Atlantic Forest. Conservation Letters 2021