Figure 3.

Bivariate choropleth maps of global ecosystem restoration priorities (Strassburg et al. 2020) compared with (a) human population density in 2018 (Rose et al. 2019) and (b) Human Development Index data (HDI) for the year 2015 (Kummu et al. 2018) on a global hexagonal grid (10 × 10 kilometer resolution) in Mollweide projection. (a) We classified areas as having no restoration potential (identified as lacking potential by Strassburg et al. 2020), intermediate (identified as the lower 80% of the areas identified as having restoration potential) or high restoration potential (identified as the areas with the top 20% restoration potential). We classified areas with less than 5 people per square kilometer (km²) as low population density areas, areas with 5–50 people per km² as medium population density areas, and areas with more than 50 people per km² as high population density areas. The bright areas show high restoration priority and high populations, highlighting the strongest need for restoration approaches that place people at the center. Important to note is that even areas with low population may still be under indigenous or local community management, which should be accounted for in restoration planning. (b) We considered areas with an HDI of less than 0.6 as low, areas with an HDI of 0.6–0.8 as medium, and areas with an HDI of more than 0.8 as high. The dark areas indicate high HDI and low restoration priority, bright areas show low HDI (i.e., highest socioeconomic vulnerability) and high restoration priority. (c) Bar plot showing the total number of people living in each of 10 restoration priority classes based on Strassburg and colleagues (2020). The number of people in areas without restoration potential are highlighted in blue, the top 20% priority classes in yellow. (d) Violin plot showing HDI within pixels grouped by restoration priority class. White dots show medians, black bars show quantiles and colored envelopes indicate the distribution of the data. We have created this map for illustrative purposes. It should not be taken as a data product itself and is limited by the underlying data sets on which it is based. A more comprehensive method is provided in the supplemental material.