Ecological restoration is a crucial nature-based solution for carbon sequestration and biodiversity conservation1. To fulfill targets of the Nationally Determined Contributions, the Bonn Challenge2 and land degradation neutrality3, research has identified areas of high value to restoration across the world based on biophysical characteristics4,5,6. While global restoration studies and prospecting tools enable private and public entities to decide where to focus restoration efforts for maximum biodiversity and carbon sequestration value, they leave people off the map. Designing and siting successful restoration projects requires consideration and integration of socio-economic needs and cultural characteristics of local stakeholders. Although there is an increasing recognition that local people need to be engaged and their interests need recognition in the design and implementation of restoration projects7,8, there are few examples of systematic consideration of people’s livelihoods and interests in restoration at large spatial scales9. Coarse socio-economic datasets cannot replace local consultations and needs assessments to ensure restoration projects provide benefits to local people. However, these data can be used as preliminary filters for different restoration methods. Here, we propose an explicit consideration of people’s socio-economic needs through the combination of biophysical and socio-economic factors to identify people-centric restoration opportunities. We also assess the de jure land tenure system to identify which types of land could be targeted for more tenure-responsive, long-lasting and socially just outcomes10.

We use India as a case study as it has a high biophysical restoration potential5,6 and one of the largest restoration targets of 26 million hectares by 203011. A large proportion (64%) of India’s population is rural and relies on local ecosystems for livelihoods through small-scale agriculture and common pool resources, making a people-centric lens to restoration design and implementation necessary. India’s focus on socio-economic development through programs such as the Aspirational Districts Programme12, emphasizes the need for the environmental agenda to align with the development agenda. For this analysis, we thus consider the living standards component of the multidimensional poverty as our socio-economic metric at the district level (N = 579 districts) to reflect dependence on natural resources. We choose this metric because people more dependent on natural resources for their subsistence and livelihoods are more likely to (a) be vulnerable to decisions made regarding land uses and (b) benefit from improved availability of natural resources in the short term. We compare this metric with the biophysical restoration potential (as quantified in Strassburg et al. 20206) to identify different socio-environmental conditions restoration programs must consider in order to balance environmental and social goals. Furthermore, we classify de jure land tenure regimes by aggregating village-level census data13 to identify prevalent land tenures. Land tenure is important for understanding who may have the authority to change land use. Although the biophysical restoration potential considered in this study refers to restoration without human disturbance6, we argue that such restoration is challenging and socially unjust in a country with high human population densities. Therefore, we define restoration as any activity which restores ecological functionality to degraded landscapes2, ranging from alternative agricultural and pastoral practices to natural ecosystem restoration.

We find that approximately 29% of districts (N = 166) with high biophysical potential are also above average poverty levels in India (above 50th percentile for biophysical potential and poverty of 579 districts; Figs. 1 and 2 quadrant 1). Similarly, 30% (N = 168) of districts have both below average biophysical potential and below average poverty (below 50th percentile for biophysical potential and poverty; Fig. 2, quadrant 3). This overlap indicates the potential and need to pursue restoration in a manner that addresses both ecological and social goals.

Fig. 1: Map of India displaying districts mapped according to variables considered in this study.
figure 1

A Living standards component of the Multidimensional Poverty Index and B Biophysical restoration potential (quantified by Strassburg et al. 20206). The colors represent the percentile range to which the districts belong.

Fig. 2: A comparison of each district’s biophysical potential and poverty level.
figure 2

A Districts plotted in reference to biophysical restoration potential and poverty measured by the living standards component of multidimensional poverty. Each district is presented as a circle. Colors represent the dominant land tenure in the district. Vertical and horizontal dashed lines represent the 50th percentile according to biophysical restoration potential and poverty. The numbers in the corner of each quadrant correspond to districts of the same color in B.

In the majority of the 579 districts considered in this study, private land is the predominant land tenure, followed by non-forest commons, then forest commons (Fig. 3). Although recent restoration efforts have overwhelmingly focused on afforestation14,15, recent evidence indicates a larger climate change mitigation potential in alternative agricultural systems, such as agroforestry and trees outside forests (ToF), than in areas which are likely to be managed as closed-canopy forests16. Furthermore, the disproportionate focus on carbon-centric forest-based projects has led to underrepresentation of projects aimed at reducing emissions of other greenhouse gases (GHGs) such as methane with enormous mitigation potential14. Traditional agroforestry practices and ToF (e.g., live fences, silvi-pastures, horti-pastoral systems) are common in India17 and could lower other GHG emissions. While it may be simpler to facilitate agroforestry among individual land holders with clear land titles; restoring degraded common lands may facilitate broader benefits, particularly among the poorest people who often don’t own land or have a strong culture of common ownership (e.g., pastoralist communities in Gujarat and Rajasthan). However, restoration of the commons can be complex when the source of degradation (e.g., an invasive species), becomes a source of livelihood for a section of the local community18.

Fig. 3: The proportion of each land tenure in the 579 districts belonging to the ten percentiles in ascending order.
figure 3

Districts above 90th percentile are poorer than districts under the 10th percentile.

By analyzing biophysical, socio-economic and land tenure data together, policy makers can devise restoration programs more holistically. For example, ten of the fourteen poorest districts that have very high biophysical restoration potential (above 90th percentile in both restoration potential and poverty), have a predominance (>50%) of non-forest (N = 8) and forest commons (N = 2). In districts above the 80th percentile in terms of both restoration potential and poverty, approximately 40% had a predominant land tenure of forest (N = 9) and non-forest commons (N = 9, total = 45 districts). It may be tempting to situate reforestation and afforestation projects, which are based mainly on plantation models15, in poorer districts with high value for restoration. However, emerging evidence shows that afforestation projects do not always increase forest cover19, sometimes reduce pastoralist access to grazing lands20, and do not contribute much to the local communities’ needs for firewood and fodder19. We argue that in districts with high biophysical restoration potential and high poverty, it could be more effective to (a) encourage traditional agroforestry practices, (b) leverage economic policies and schemes designed to raise living standards21, (c) use alternative restoration practices, such as invasive species management in districts with a high proportion of common land and (d) allow for greater community rights to manage the commons22. For example, approximately 30% of the districts above the 80% percentile of both restoration potential and poverty are in Madhya Pradesh. Managing an invasive species, Lantana camara in forest and non-forest commons in that state increased the local communities’ access to firewood and fodder15. Moreover, recent evidence from some of these districts shows that switching to alternative energy sources for cooking and use of durable housing materials raised living standards, as well as provided a safer cooking fuel option and contributed to forest regeneration near villages21.

Similar evidence of forest regeneration with the adoption of biogas digesters in a district with high poverty but low biophysical restoration potential, such as Chikkaballapur in Karnataka, emphasizes the potential of human well-being policies to have positive ecological outcomes23. In districts with high biophysical restoration potential and low poverty, including Malappuram and Thrissur in Kerala, agroforestry and cash crop plantations, along with other livelihood alternatives, have played a role in alleviating poverty and increasing food security24. These traditional agroforestry systems and private home gardens could continue to be supported and incentivized. Furthermore, novel tools such as Diversity for Restoration (D4R) help people select appropriate species for planting based on the outcomes they are interested in, such as erosion control25. In regions with low poverty and low biophysical potential (both factors below 50th percentile), such as districts in Rajasthan and Gujarat, the predominant land tenure is private. These districts could be targeted for irrigation management to increase drought resistance and agri-pastoral projects which could simultaneously contribute to reductions in methane emissions14,17. With a considerable area of non-forest commons (>33.33% land tenure), pasture and open natural ecosystems (ONEs) restoration could also be beneficial to the numerous indigenous pastoralist communities in these states26,27. Moreover, ONEs would not necessarily store more carbon if afforested28. Thus, preserving these non-forest ecosystems will not only benefit pastoralists but also conserve unique non-forest ecosystem biodiversity27,28. The interventions suggested in the four different socio-environmental conditions were not designed in the context of the relationship between biophysical restoration potential and poverty. Therefore, it is critical to understand the applicability of these interventions in the context of these different conditions, and the cost-effectiveness of these interventions to successfully scale them.

Our analysis has some limitations. First, the district administrative unit is a convenient spatial scale to plan interventions and programs. But we recognize that households are not socio-economically uniform and thus, restoration programs will not have uniform effects in a district. As an example, agroforestry programs can have very different food security outcomes for people who own land and those who do not. Second, the analysis carries inherent uncertainties found in the data sources.

This study attempts to demonstrate a people-centric approach to translating global biophysical restoration potential studies for application to a country-specific context, rather than prescribing restoration priorities. Based on a country’s development and environmental agenda, the variables used to determine the different socio-environmental conditions may be different. An analysis of this nature can help policy makers and an emerging diversity of actors in the field of ecological restoration broadly filter restoration methods best suited for different socio-environmental conditions.


Data sources and preparation

Land uses and de jure land tenure regimes

We aggregated the most recent publicly available census data (2011)13 at the village level to the district level to quantify the de jure land tenure regimes that include private land, common non-forest land and forest land. For this study, we consider 579 districts for which we had a complete dataset, including the data on poverty and biophysical restoration potential. We categorized the land use data available at the census village level into the following de jure land tenures:

Land tenure regime

Land use categories from Census 2011 land records

Private land

1. Net sown area

2. Current fallow land

3. Fallow lands other than current fallows

Common non-forest land

1. Culturable wastelands (grasslands)

2. Area under non-agricultural use

3. Barren or uncultivable land

4. Permanent pastures or grazing lands

5. Land under miscellaneous tree crops (orchards)

Common forest land


In order to only include inhabited census villages, we removed census villages with zero as total population and those explicitly labeled ‘uninhabited’ in the village name. Further, we included only non-state-owned land by filtering out the following categories of census villages:

Type of state-owned land

Terms used in the census village name

Army owned land or firing range

firing range


reserve, beat, block, forest, camp, range, gate, K.M.

In order to report the total hectares of specific land uses and to calculate the proportion of de jure land tenures, we treated any inconsistencies in the original census land use data in the following manner:

Inconsistency in the land use records

Description of the inconsistency

Potential reason for inconsistency

Treatment of inconsistency

No land use records

All land use columns show zero hectares but Total area in hectares has a positive value

The census enumerators did not reach these villages.

These villages appear as ‘No data’

Total areas in hectares reported not equal to total of all land uses

Column from Census 2011 records not equal to actual total hectares of all land uses.

There are two possibilities:

a: Total area in hectares > total of all land uses or

b. Total area in hectares < total of all land uses

Error in addition of land uses by census enumerator or land use is currently disputed.

1. For our analysis, we considered the total of all land uses to calculate the proportion of land tenure for a village.

2. We created a variable ‘Unaccounted land’ = Total area in hectares- Total of all land uses

Total area in hectares is reported as zero but land use records exist

All land use columns have a positive value in hectares but Total area in hectares is zero

Error in addition of land uses by census enumerator.

For our analysis, we considered the total of all land uses to calculate the proportion of land tenure for a village.

Living standards component of the multidimensional poverty index

Our study used one dimension (living standards) of the three dimensions of the multidimensional poverty index (living standards, health and education)29. We chose to only look at the percent contribution of living standards to poverty in a district because education and health services are provided largely by the government and may not necessarily reflect poverty due to the lack of viable livelihood options. For 579 districts, the percent contribution of living standards to multidimensional poverty ranged from 18.2% to 56.7%. We scaled this percentage from 0 to 1 to ensure that we could make a fair comparison with the biophysical potential for restoration taken from Strassburg et al. 20206. We split the districts into 10 percentiles based on their value, with values closer to zero indicating higher living standards and 1 denoting lower living standards or higher levels of poverty (Fig. 1a).

Biophysical potential for restoration

We used the spatial data from Fig. 1e from Strassburg et al. 2020, which considers the ecological restoration potential of countries around the world based on the biodiversity conservation and climate change mitigation potential that a location holds while considering the cost of land. In R computing software, using the packages raster30 and rgdal31, we clipped the map of the restoration potential of the districts in India to compute the mean biophysical restoration potential of a district. The values of the original dataset ranged from 1 to 20, denoting 5% increments in restoration potential. We rescaled the values from 0 to 1 to make a fair comparison with the living standards component of the multidimensional poverty index. We split the 579 districts into 10 percentiles for presentation (Fig. 1b).

All maps in this study were created using QGIS version 3.16.832.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.