Indigenous resistance to colonisation impeded deforestation in the Southern Atlantic Forest for 300+ years

Freg J. Stokes & Barney Harris

• Timóteo da Silva Verá Tupã Popyguá
• Ará Rete (Sandra Benites)

• COMPLEG team
• and many more…

• …looking forward to telling you about this research
• …just to say the archaeology and history of Latin America is not a specialism of mine
• …however the 2 + 1/2 years I’ve been working on this project it’s been a real privilige and an honour to work on research with a direct contemporary significance and a real world impact, particulary in terms of Indigenous rights and the protection of forests

About the team

Our co-researchers

Guaraní co-researchers: Timóteo da Silva Verá Tupã Popyguá (Photo: Edu Simões) & Ará Rete (Sandra Benites); Freg Stokes, Phil Riris (PI), Patrick Roberts (Co-PI)

The Atlantic Forest

A biodiversity hotspot

• ~20,000 species of plants
• ~300 species of mammals
• ~700 species of amphibians and reptiles

Indivisible from humans

• “Nature is part of us” Karai Djekupe (Thiago), Guarani youth leader
• Tekoa: the place where the Guaraní way of life is possible

• …also some of highest levels of endemism in the world, so many of these species can only be found in the Atlantic Forest.
• …conceptually, these stats relate very much to an empirical, Western perspective of the environment.
• …Guarani cosmoperception of the environment is different, with no comparable dualism between human and nature.
• … encapsulated by Guarani word Tekoa, which means ‘place where life is possible’ and can be used to refer to both the village and its surrounding forests.

A history of exploitation

• Only ~10% of Atlantic Forest remains
• Deforested by Spanish and Portuguese colonists (from 16th C. onwards)
• Coffee, cocoa, cattle, gold, silver, soy, wheat

Indigenous resistance

• Key Indigenous figures led resistance against the Spanish, such as the Guaraní leader Ñezú (1628).
• Major battles included Mbororé (1641)
• Spatial distribution of resistance important

• … a Jesuit-Guaraní victory against the Portuguese.
• … e.g. Influenced location of modern Brazil-Argentina border.

Question:

Can we model the relationship between Indigenous resistance and forest survival?

Geospatial modelling

Benefits of formal spatial modelling frameworks

• Identify complex spatial relationships
• Quantify the impact of resistance
• Enable counterfactual simulations
• Feed into policy in a way that qualititative history cannot

• …not otherwise obvious to the human eye, or might be laborious to identify.
• …in comparison to other environmental covariate or spatially distributed historical phenomena.
• …assuming modelling is successful, it’s possible to simulate counterfactual scenarios under different environmental and historical conditions
• …quantititive data can interface with evidence-based policy making and help drive changes

Data collection

Resistance events

• Archival visits
• Collated data on Indigenous resistance events
• Consultation with Indigenous communities

• …10+ archives in Italy, Spain, Portugal, Brazil, Argentina
• …16th-19th C., inc. location, date, key figures
• …for wider meaning and perspective about resistance

Data collection

Deforestation

• Declassified CIA satellite imagery
• Digitisation of surviving forest (~1965)
• Difference with RESOLVE Biomes data

• … 1st-gen CORONA sattelite, found in archive in Maryland, georeferenced in QGIS using common landmarks.
• … Freg painstakingly digitised surviving forest visible in the images, to produce the earliest available forest survival maps
• … if we take the geometric sym difference of this with the predicted Atlantic Forest biome we get…

Key datasets

• Binomial deforestation (response)
• Indigenous resistance events (predictor)
• Reasonable spatial correlation visible, however…

Least cost path analyses

• Colonisers needed routes through the Atlantic Forest
• Mobility impacted by topography, river flow strength and direction + rapids
• Spatial impact of resistance on colonial routes local and dispersed

• Freg’s historical research indicated that Indigenous resistance had a spatially dispersed impact on colonial routes through the landscape
• This is because resistance frequently occurred around access points to key rivers or in nodal points of the landscape in terms of movement

Building the model

A reproducible approach using R

Modelling with R

forest survival ~ indig. impact + envir + anthro

• {targets} aids reproduciblity + HPC
• {spatialRF} explanatory spatial random forest models
  • Spatial predictors (MEMs) capture spatial structure within data
  • Built-in diagnostic tools and spatial cross-validation procedures

{tidyverse} + {terra} = (spatial) data wrangling and geoprocessing

Sampling strategy

• Forest survival is binomial (1/0 = survived/deforested)
  • Highly spatially contiguous
  • Highly spatially autocorrelated
  • Many observations (~35,000)
  • Difficult to model!

• Bootstrap aggregation (bagging)
• 242 individual spatial RF models
  • ~100–140 observations each
  • 40km resolution lattice used to enforce min. distance
  • Oversampling of small fragments = class balance & less sampling redundancy

Results

• Mixed in terms of AUC_ROC (< 0.7)
• Some model have spatially correlated residuals (Moran’s I <= 0.05)
• Caution when interpreting results

Predictions

• Closer to Indigenous resistance => higher chance of forest survival

Simulation

• Counterfactual scenario with no Indigenous resistance
• 20–22% increase in deforestation (minimum)

Concluding thoughts

• Ultimately the Atlantic Forest was heavily deforested…
• however, a similar story continues to play out in the Amazon today…
• Freg will share our research with Guarani communities in Misiones, Argentina
• … and work with Brazilian Guaraní researcher Ivanilde Kerexu to assess how research might help support future forest conservations strategy

Thank you

• bharris@bournemouth.ac.uk
• @barneyharris@sciences.social
• Barney Harris

• Built by a human using revealjs & quarto
• Quarto theme by Emil Hvitfeldt