Geographic distribution of isolated indigenous societies in Amazonia and the efficacy of indigenous territories

Abstract

The headwaters of the Amazon Basin harbor most of the world's last indigenous peoples who have limited contact with encroaching colonists. Knowledge of the geographic distribution of these isolated groups is essential to assist with the development of immediate protections for vulnerable indigenous settlements. We used remote sensing to document the locations of 28 isolated villages within the four Brazilian states of Acre, Amazonas, Roraima, and Rondônia. The sites were confirmed during previous over-flights and by image evidence of thatched-roof houses; they are estimated to host over 1,700 individuals. Locational data were used to train maximum entropy models that identified landscape and anthropogenic features associated with the occurrence of isolated indigenous villages, including elevation, proximity to streams of five different orders, proximity to roads and settlements, proximity to recent deforestation, and vegetation cover type. Isolated villages were identified at mid elevations, within 20 km of the tops of watersheds and at greater distances from existing roads and trails. We further used model results, combined with boundaries of the existing indigenous territory system that is designed to protect indigenous lands, to assess the efficacy of the existing protected area network for isolated peoples. Results indicate that existing indigenous territories encompass all of the villages we identified, and 50% of the areas with high predicted probabilities of isolated village occurrence. Our results are intended to help inform policies that can mitigate against future external threats to isolated peoples.

Fig 1. Yanomamö village.

Images from an isolated Yanomamö settlement in 2012 and 2014, as identified with remote sensing. Satellite images courtesy of DigitalGlobe, Inc.

Fig 2. Maps of Amazon Basin study area in Brazil.

Study area includes states and locations of isolated villages (triangles, panel A), and locations of isolated villages and existing indigenous reserves (crosshatch, panel B). Panel C illustrates MaxEnt model results, with red areas indicating higher probabilities of occurrence for isolated tribes. Panel D includes both MaxEnt model results and overlays of existing indigenous reserves. Panel E illustrates areas with MaxEnt occurrence probabilities that are higher than areas in which isolated tribes have actually been observed (minimum training presence), and panel F includes the overlay of existing indigenous reserves.

Fig 3. Marginal response curves.

Marginal response curves for the top three variables included in the MaxEnt model of occurrence of isolated tribes in the Amazon Basin of Brazil. Plots reflect the dependence of predicted site suitability on each variable, and on dependencies induced by correlations between variables.