Aim

Street trees have been shown to provide many benefits to urban dwellers, including ecosystem services such as carbon sequestration, heat-island mitigation, and storm water retention (Roy, Byrne, and Pickering, 2012; Mullaney, Lucke, and Truman, 2015). Street trees have also been shown to provide health benefits, like a reduction in childhood asthma (Lovasi, Quinn, Neckerman, Perzanowski, and Rundle, 2008). Street trees are, therefore, an amenity for residents of urban areas. However, this resource is not always equitably distributed (Landry and Chakraborty, 2009; Pham et al. 2012). This study aims to examine the spatial distribution of street trees, including factors that may correlate with street tree distribution, such as income.

Question and problem

Are street trees equitably distributed in major cities?

Location and grain

This study will examine the distribution of street trees in Philadelphia, PA at the census tract level.

Methods

Data on street trees was obtained from the Philadelphia Department of Parks and Recreation (2016). Additional data was obtained from the iEcoLab (unpublished) and the Delaware Valley Regional Planning Comission (2018).

To determine which factors strongly correlate with street tree distribution, a multiple linear regression model was run with street tree density per census tract as the response variable.

In the full model, the following explanatory variables were included: percent impervious surface cover, percent green space, particulate matter concentration, ozone concentraton, rate of people with high blood pressure, rate of people with asthma, rate of people with bad mental health, corrected property values, youth population perccentile, older adult population percentile, female population percentile, racial minority population percentile,ethnic minority population percentile, foreign born percentile, limited english proficiency population percentile, disabled population percentile, and low income population percentile.

However, there was a high degree of collinearity in the full model. The final model included only the following explanatory variables: percent impervious cover, particulate matter concentration, ozone concentration, asthma rate, racial minority percentile, ethnic minority percentile, and property values. These variables were included in the final model because they were of interest and because they did not display collinearity.

Model formula: tree density ~ percent impervious surface + fine particulate matter + ozone concentration + asthma rate + corrected property values + racial minority percentile + ethnic minority percentile

Results

Multiple factors correlate with street tree density, including property value (a proxy for income).

Figure 1: Added variable plots of street tree density

Street trees are not equitably distributed in Philadelphia, PA.

Figure 2: Spatial distribution of (a) street tree density, (b) property values, (c) percent impervious surface, (d) particulate matter concentration, (e) ozone concentration, and (f) percentile of ethnic minority population