This dataset was selected to represent lands that are relatively free from human modification (i.e. natural) and thus supporting areas important for maintaining biodiversity and providing ecosystem services. The degree of human modification models known stressors to ecological systems due to land use and human activities. Major stressor types include: built-up (urban and low-density residential), agriculture, oil and gas production, mining/extraction, power generation, roads/railways, powerlines, logging/harvesting, human intrusions, and air pollution. Both the area (footprint) and intensity of use are represented, and stressors are combined using rigorous mathematical equations to reduce collinearity. The raw values were binned into five classes: very low (i.e. ‘wild’; <0.01), low (0.01-0.1), moderate (0.1-0.4), high (0.4-0.6), and very high (>0.6). The wild category value is based on the average human modification value of all IUCN protected areas 1a, 1b, and 2. The ‘low’ category represents the next lowest level of human modification. The 0.4 and 0.6 cutoffs for the moderate and high human modification categories are based on thresholds derived from percolation theory that relates habitat loss to fragmentation. In addition, these thresholds are consistent with values of "working" landscapes (e.g., intensive agriculture has a value of 0.5). For cartographic purposes, the classes were smoothed using a moving window radius of 1 mile. These data provide the basis to measure relative intactness or ecological integrity for species and ecological processes sensitive to human activities. Important additional human activities missing from these data include fire suppression activities and visitation/recreation. Larger areas with low human modification likely require lower relative energy inputs to support native biodiversity than areas that are more modified, except where intensive fire suppression has occurred, invasive species have been introduced, or other uncaptured influences (e.g., drought, erosion) are at play. These data are often also used as a primary factor to develop a resistance surface used for connectivity modeling (see Belote connectivity data), though doing so requires specific constraints and interpretations of these data. More information: https://essd.copernicus.org/articles/12/1953/2020/essd-12-1953-2020.html