To assess industrial development threat and risk on Indigenous lands, we used spatial datasets at 1-km resolution for the (1) geographical extent of Indigenous lands; (2) human modification of terrestrial lands used as proxy for ecological integrity; (3) land suitability maps for future development expansion by commodity-based and extractive economic sectors that provide development pressure indices; and (4) a novel national-level index, referred to as the Authority-Capacity-Support Index, which combines six national-level datasets to capture the social, economic and political contexts that can either promote or hinder continued stewardship of these lands by IP. All geospatial analyses were conducted in the Mollweide projection, an equal-area map projection, using ArcGIS 10.8.1 software (www.esri.com) with the Spatial Analyst extension. We applied bilinear resampling method for continuous raster data and nearest-neighbor method for discrete data (vector data were first projected and then converted to raster datasets). For national-level analyses, we used country boundaries sourced from the Global Administrative Areas (GADM) spatial database version 2.8 (https://gadm.org/).
Indigenous lands. We used the boundaries of Indigenous lands mapped by Garnett, Burgess 1, who identified Indigenous lands across 87 countries or politically distinct areas. This dataset represents the most comprehensive assessment of terrestrial lands where IP have customary ownership, management, or governance arrangements in place, regardless of legal recognition. It is based on 127 publicly available sources, including cadastral records, participatory maps, and census data. We adopt their definition of IP as those who identify as having “descended from populations which inhabited a country before the time of conquest or colonization [and] who retain at least some or all of their own social, economic, cultural and political institutions” (see Key Terms in Supplemental Glossary) As discussed in Garnett, Burgess 1, we note the practical and ethical challenges of defining IP and the implications for mapping their lands (see their SI for further details). As a result, the Indigenous lands map should not be used to identify specific territories or legal claims, nor should areas without delineation be interpreted as lacking IP’ presence, claim or interest. We also acknowledge that voids in these maps do not necessarily imply an absence of IP, but rather areas for which their presence cannot be determined from publicly available geospatial resources.
Development Threat on Indigenous Lands. We calculated a development threat score for each Indigenous land cell as a function of its ecological condition or degree of naturalness from the Human Modification (HM) dataset, and its cumulative development pressure from the Development Potential Index (DPI) dataset as follows:
Human Modification of Indigenous Lands. The ecological integrity of Indigenous lands can be measured by the extent of their modification by specific human activities known to negatively impact ecosystems 67. We adopt this approach using the published global Human Modification (HM) map 11. The HM is a geospatial map of the extent and estimated intensity of impacts from 13 anthropogenic stressors associated with human settlement (population density, built-up areas), agriculture (cropland, livestock), transportation (major roads, minor roads, two-tracks, and railroads), mining, energy production (oil wells and wind turbines), and electrical infrastructure (powerlines and night-time lights). We note that although the HM captures many human impacts, it does not include timber production or selective logging, pastureland, recreational use, and hunting. HM is a 0–1 metric that reflects the proportion that each 1-km2 land area is modified by human activities based on the median year of 2016. Following 11, 67 and based on the distribution of HM values globally and in protected areas, we categorized the modification of each cell as very low (0.00 ≤ HM ≤ 0.01), low (0.01 ≤ HM ≤ 0.10), moderate (0.10 < HM ≤ 0.40), high (0.40 < HM ≤ 0.70), and very high (0.70 < HM ≤ 1.00) (Fig. 1a). Low modified lands represent natural or semi-natural areas that are no more than 10% modified and have less than two overlapping human stressors; moderately modified lands are >10 to 40% modified and have less than three overlapping human stressors; and highly modified lands are human-dominated areas with over 40% modification with five or more overlapping stressors. Based on these categories, we created corresponding “naturalness scores” ranging from 1 to 5, assigning 5 to the most intact, natural lands (i.e., very low HM) and 1 to the most modified lands (i.e., very high HM).
Development Pressure on Indigenous Lands. We used published development potential indices (DPIs) 12 for renewable energy, oil and gas, mining, and agricultural sectors and created an additional urban DPI based on global urban growth projections from 2020 to 2050 68. DPIs are global, spatially-explicit land suitability maps at 1-km resolution that depict development pressure from the potential expansion of renewable energy (concentrated and photovoltaic solar power, wind power, and hydropower), oil and gas (conventional and unconventional), mining (coal, metallic and non-metallic mining), and agriculture (crop and biofuels expansion) sectors. Each DPI has standardized 0–1 values that indicate low to high suitability for future industrial development expansion based on a) sector-specific land constraints on development (e.g., suitable land cover, slope); b) land suitability for sector expansion based on resource availability (sector-specific yields); and c) siting feasibility of new development (e.g., ability to transport resources or materials, access to demand centers, and proximity of existing development).
For each DPI, we binned the range of values into six pressure categories based on standardized z-score ranges following 12 to characterize the development pressure on Indigenous lands as very low (≤10th percentile), low (>10th – 25th percentile), medium-low (>25th – 50th percentile), medium-high (>50th – 75th percentile), high (>75th – 90th percentile), and very high (>90th percentile). We calculated z-scores by mean-standardizing values per country under the assumption that national-level domestic demand will drive national-level resource extraction. Because urban DPI was derived from urban expansion probabilities based on population growth projections that were more restrictive than the DPI values (e.g., excluded suitable areas like flat land, near roads and existing urban areas once demand was met), we binned the upper 50th percentile and lower 50th percentile of non-zero urban DPI values into high or very high pressure categories, respectively. Based on pressure categories, we then assigned pressure scores ranging from 1 (very low category) to 6 (very high category) to each DPI, and then created a cumulative development pressure map across all DPIs by retaining the max cell score across all individual DPIs (Fig. 1b). All lands without a pressure category were assigned a score of 0.
To evaluate sector-specific drivers on Indigenous lands, we identified regions of high development pressure based on selecting cells with high or very high categories within any of the DPI maps (orange and red areas depicted in Fig. 1b). We determined the sector driver based on which of the five sector categories (i.e., renewable energy, oil and gas, mining, agriculture, or urban) had the highest pressure score for that cell (Fig. 1c). To illustrate, if a given cell had only one sector (e.g., agriculture) classified as very high (pressure score of 6), that sector was the identified driver. If a given cell had multiple sectors with similar scores and no clear max score, we assigned it as a having multiple sector drivers. We highlight a sector as the majority driver in a country if it made up >50% of the high development pressure estimate for Indigenous lands.
Development Threat on Indigenous Lands. We estimated development threat on Indigenous lands as a function of their human modification categories (using the 1 to 5 naturalness scores) multiplied by their cumulative development pressure categories (using the 1 to 6 pressure scores). This produced a 1-km resolution threat map for all Indigenous lands with scores ranging from 0 to 30, where 0 identifies Indigenous lands lacking any development pressure and 30 identifies Indigenous lands in the best ecological condition (i.e., least modified = 5) and the highest potential for industrial expansion (i.e., very high development pressure = 6). For visualization purposes, we binned the range of threat scores into six categories (very high, high, moderate, low, very low, none) using Jenks Natural Breaks (Fig. 1d).
Authority-Capacity-Support Index. National-level indices have been developed to assess risk and guide prioritization for a variety of purposes, including natural hazards or climate-related events 13, 14, 62, food system sustainability 69, food insecurity (https://fews.net/), commodity supply chains 70, and nature conservation 71. To our knowledge, our Authority-Capacity-Support index is the first global composite index that captures the socio-economic and political contexts with relevance to continued stewardship of Indigenous lands by IP confronting industrial development pressures. We selected a suite of national-level indicators that can either encourage or hinder the development of localized enabling conditions. We focused on national-level indicators that characterize contexts that: (1) strengthen and secure IP’ rights and decision-making authority (Authority sub-index); (2) promote the capacity for their engagement and representation in decision-making processes (Capacity sub-index); and (3) provide capital and support for Indigenous-led conservation and sustainable development (Support sub-index) (Fig. 2).
We used published data from publicly available datasets that were global in extent and based on the most-up-to-date information (data sources from 2013 (n =1), 2015 (n =1), 2018 (n = 4)). We prioritized indicators that were applicable to the concept in question and had theoretical and empirical support or broad utilization in other global prioritizations. Prior to selecting the final set of indicators, we tested for scale reliability among the full suite of indicators considered for each of the ACS sub-indices using Cronbach’s alpha (α). This test provided a measure of internal consistency within the suite of indicators selected and a statistic that allows for evaluation of their suitability as complementary measures of the same sub-component. Based on calculations of Cronbach’s alpha (α), internal reliability was considered “acceptable” for indicators of the Authority sub-index (α = 0.742) and “good” for indicators of the Capacity and Support sub-indices (α = 0.848 and 0.889, respectively) (Supplemental Figure 2). We considered three, four, and six national-level indicator datasets for the Authority, Capacity, and Support sub-indices, respectively, and ultimately retained two per index after consideration of their correlation statistics (See Supplemental Tables 2-4, Supplemental Figure 1).
Authority. For the Authority sub-index, we retained LandMark’s index of IP’ legal security 72 and World Bank’s Political Stability Index 73. The LandMark index provides a measure of how well a nation’s laws support IP’ rights over their land and resources 74 (see Supplemental material for further details on the components of various indices). Tenure security is noted to increase the motivation for collective action to sustainably manage territory and resources 75, 76. Thus, secure rights over territories and resources can enable better outcomes for IP confronting development pressure by empowering them with the legitimacy and authority to act as decision-makers, increasing their accountability for the decisions made, and improving their capacity for collective action in the sustainable management of local resources 44, 45. We view political stability as complimentary to LandMark’s indicator of IP’ legal security. The Political Stability Index is based on perceptions of political instability or politically motivated violence and considers instances of civil unrest, ethnic and international tension, armed conflict, violent demonstration, and internal/external conflict 77. We expect political stability to have a positive and complementary effect on the strength and security of IP’s rights and decision-making authority since more stable national governments tend to uphold their constituents’ rights and help to enforce and mitigate potential conflicts. In contrast, unstable governments can result in periodic abuses 78, loss of recognition and enforcement of IP’ rights, and even (re)appropriation and dispossession of IP’ homelands 78.
Capacity. For the Capacity sub-index, we included World Bank’s Control of Corruption Index 73 and the World Resource Institute’s Environmental Democracy Index 79. The Control of Corruption Index reflects perceptions of the extent to which a government can address problems with public power exercised for private gain, as well as the potential for capture by elites and private interests. We expect control of corruption to be positively correlated with the capacity for IP’ representation in the decision-making process, as higher values suggest a context favoring fairness and equity, where protections for the most marginalized and vulnerable exist to discourage power imbalances 77. There is also evidence to suggest that control of corruption is important to environmental outcomes, given that countries with stronger control of corruption exhibit significantly lower environmental destruction over time 80. The Environmental Democracy Index characterizes the extent to which a country’s citizens enjoy access to information, participation, and justice in environmental matters. It consists of different legal indicators concerned with the development and implementation of relevant legislation (e.g., laws and regulations governing freedom-of-information, requirements for consultation and environmental impact assessment, regulations on extractive industries, etc.), as well as indicators that assess their efficacy in practice 79. Higher values for environmental democracy suggest a context that promotes the capacity for both engagement and representation in decision-making processes.
Support. For the Support sub-index, we included the United Nation’s Human Development Index (HDI) 81 and country-level investments in the UN’s Sustainable Development Goals (SDGs) 82 . The HDI is a multi-dimensional index that describes human well-being with respect to life expectancy, education attainment, and standard of living. Higher HDI scores suggest the availability of adequate human, social, and financial capital within the population at-large to prioritize and make such local investments. They also suggest a context that allows for more latitude to be selective about the types of development accepted; whereas lower HDI scores suggest a vulnerability to accepting any form of potential development, and a diminished incentive to pursue competing priorities (e.g., biodiversity conservation vs. GDP). These relationships are supported by others that have found that countries with rapidly expanding economies and higher HDI scores exhibit subtle improvements in environmental condition 80 . Finally, we included country-level investments made in support of the United Nation’s SDGs 83 , which totaled nearly $1.52 trillion dollars between 2000 to 2013. Country recipients of higher SDG investments are assumed to have greater potential to support IP in achieving Indigenous-led conservation and sustainable development. Yet, with this indicator, we cannot determine the distribution of national-level investments or differentiate specific beneficiaries, though several of the 17 SDGs include Indigenous-specific indicators (Supplemental Table 5).
We constructed the ACS Index as a hierarchical, geometric mean of the three sub-indices (authority, capacity, and support) and their two indicators (as described above), according to the formula:
ACSi = Geomean (Authority Geomean (Ind1, Ind2); Capacity Geomean (Ind1, Ind2);
Support Geomean (Ind1, Ind2))
where Ind1 and Ind2 refers to the two indicators included in each of the sub-indices. Similar to other global metrics 69, we used a geometric mean given the non-compensatory nature of indices and domains and to control for any unequal or skewed values across indicators, or residual correlations across domains. We also excluded within-domain indicators that were closely correlated (Supplementary Table 1), and equally weighted sub-indices due to lack of justification for differential weighting. Prior to calculating the sub-indices, we transformed individual indicators with skewed values, reflected their values to ensure consistent valence, and rescaled (normalized) their range of values from 0 − 1, such that a value of 0 indicated the lowest authority, capacity or support score and a value of 1 indicated the highest (Fig. 2, Supplementary Figure 3). Lastly, an ACS score was only calculated for countries where all three sub-indices were present (n=199). We assigned ranks to countries based on their ACS Index scores, such that a rank of 1 indicated a country with the most favorable context for IP to effectively govern and continue to steward their lands and resources. See Supplementary Methods for further details on pre-processing of indicator datasets, criteria for inclusion/exclusion of indicator datasets, and statistical results on indicator reliability and inter-correlations.
Development Risk to Indigenous Lands. We defined development risk to Indigenous lands to be a function of a country’s development threat and its authority, capacity, and support score (Fig. 3). For each country, we calculated development risk to Indigenous lands by multiplying its mean development threat score by its inverse ACS Index score, such that higher values were consistent with highest risk, representing places of high development threat coupled with low authority, capacity, and support. The resulting risk scores obtained for each country were then rescaled so that values ranged from 0 (country with lowest development threat and highest ACS scores) to 1 (country with highest development threat and lowest ACS scores). For display purposes, all country maps were assigned to one of five categories of very high to very low based on continuous scores using Jenks Natural Breaks.
We then examined the influence of national contexts by evaluating 1) changes in country ranks for development threat vs. development risk (Fig. 4) and 2) how ranks and median scores varied across countries with respect to each of the Authority-Capacity-Support sub-indices (Table 1). A total of 88 countries had development threat scores and 79 countries had development risk scores.