The following data files, R scripts, and Supporting Documents are found in Gallo_etal_FEE_2026.zip

Authors

Travis Gallo¹, Mason Fidino², Chris Schell³, Marccus Hendricks¹, Seth Magle²

1. University of Maryland
2. Lincoln Park Zoo, Chicago
3. University of California, Berkley

Abstract

We investigated how a discriminatory housing policy – redlining – has shaped the spatial patterns and configurations of greenspaces throughout 177 U.S. cities. Housing segregation has been a long-term development practice that has sequestered communities of color to areas with heightened environmental and public health risks. While the lasting environmental, social, and economic impacts of redlining are clear, the impact of redlining on landscapes are still unfolding. We found that neighborhoods that were historically redlined have less greenspace and that individual greenspaces were smaller and less connected. We also found that residents living in these neighborhoods with less greenspace were predominantly communities of color and/or had lower income. Thus, the legacy of redlining can be seen in the modern spatial patterns of urban greenspaces, and ecosystem services provided by greenspaces have been systematically absent from redlined communities for decades

File Descriptions

To conduct this analysis you will have to download the publicly available shapefile from the Mapping Inequality project (https://dsl.richmond.edu/panorama/redlining/data).
The shapefile that we used for the entire U.S. is no longer available as a single shapefile. The full U.S. data can be downloaded as a GeoPackage or GeoJSON file and processed or
the shapefiles can be downloaded city by city and combined at the link provided above.

To use 2026-01-12_data_processing.R you will also need to download all of the publicly available GIS data.
Links to the url that we used to download each publicly available data layer is provided within the script that they are called.

Data File:

2023-03-02_processed_dataframe.rds: this dataframe is the results of 2026-01-12_data_processing.R and can be called into 2026-01-12_model_fit.R to fit models.

This data has 43 variables and 10588 observations. Each row is a polygon (feature) from the Mapping Inequality HOLC shapefile used in this analysis.

R Scripts:

2026-01-12_data_processing.R: will process the Mapping Inequality HOLC shapefile and prepare the data for 2023-03-01_model_fit.R

2026-01-12_model_fit.R: will use the results form 2023-03-01_data_processing.R to fit Bayesian models to each response variable using NIMBLE.

2026-01-12_Plotting_Landscape_Results.R: is used to make the first panel in Figure 2 plotting the landscape metric results

2026-0-12_PlottingDemographyResults.R: is used to create the second panel in Figur 2 plotting the demographic results

DBDA2E-utilities.R and package_load.R are utility scripts with various functions used throughout the analyses

SummaryScripts.R: script to calculate various summary statistic

Supporting Documents:

CitySpecificModelResults.pdf: this supporting document plots the 95% credible intervals of city-specific posterior distributions for each response variable
as a function of HOLC category. Each page contains a city with a 5-panel plot containing the model results
for each response variable.

CitySpecificModelResults-colorblindFriendly.pdf: this is a colorblind friendly version of CitySpecificModelResults.pdf