The relationships between racial composition and garden proximity remain relatively unchanged with interaction terms added to the Philadelphia model: an increase in the share of Black or Hispanic residents predicts that the nearest garden would be closer, while the share of Asian and Pacific Islander residents is not significantly associated with garden proximity. The interaction between percent Black and year also has a significant and positive coefficient, suggesting that like housing costs, the relationship is gradually disappearing over time. The model estimates that in 1980, a 1% increase in the share of Black residents would be associated with a decrease of about 19 meters to the nearest garden, all other factors being equal. However, every additional year would see the predicted distance to the nearest garden increase by about 0.3 meters for every 1% increase in the share of Black residents. Taken together, these coefficients suggest that the greater proximity of gardens to neighborhoods with more Black residents will disappear after about 65 years, or around 2044. The model cannot determine whether this change in the relationship results from more garden attrition in Black neighborhoods than other neighborhoods, or from the proportion of Black residents decreasing in neighborhoods in which gardens have more durability , cannabis equipment but either scenario seems plausible given the shift in organizational priorities at the Pennsylvania Horticultural Society.
With little grant funding to help maintain existing gardens, many of the projects that were developed through Philadelphia Green did decline over time. At the same time, the Neighborhood Gardens Trust has begun to target its garden preservation efforts at rapidly gentrifying neighborhoods where gardens appear to be the most threatened; however, preserving the gardens does not prevent the demographic change and possible displacement that are associated with gentrification. Results for Seattle provide a contrasting example of what can happen when garden preservation becomes the rule rather than the exception, and when organizational priorities shift over time toward developing gardens closer to people who ostensibly need them more. As in Philadelphia, with interaction terms added to the Seattle model, a higher poverty rate is associated with the nearest P-Patch being further away . However, unlike Philadelphia, the interaction term is also significant, and it moves in the opposite direction. That is, for every passing year, a 1% increase in the poverty rate is associated with a decrease of about 0.7 meters to the nearest P-Patch, with all other factors being equal. Taken together, these coefficients suggest that Seattle’s gardens were originally distributed further away from high-poverty neighborhoods and closer to low-poverty ones, but the relationship reversed after about 22 years, and from 2002 onwards gardens are increasingly likely to be found closer to high-poverty neighborhoods than low-poverty ones. As explained in chapter 3, the leaders of the P-Patch program in the 1990s were responsive to the public concern that the gardens were a private use of public space and to city officials’ appreciation for evidence showing how the gardens benefitted low-income and other marginalized residents. The program leaders undertook a concerted effort to expand the program in neighborhoods with greater socioeconomic need, an effort which gained traction especially after the 2000 Pro-Parks Levy infused the program with $2 million.
This effort included working with the Seattle Housing Authority to build gardens in low-income housing developments specifically for use by their residents. The changing organizational priorities in the 1990s and influx of resources in 2000 would logically explain why the interaction model shows gardens’ proximity to poor neighborhoods equalizing around 2002 and growing gradually closer since then. The interaction model’s results suggest that P-Patch gardens have become more accessible to poor neighborhoods over time, but also that they have become less accessible to immigrants. Similar to the pattern observed with percent Black residents in Philadelphia, the coefficient for percent foreign born in Seattle is negative and significant while the coefficient for interaction between year and percent foreign born is positive and significant—in fact, it is the largest coefficient of any interaction term across the three models, suggesting a relatively fast pace of change. The model estimates that in 1980, the nearest garden would be about 60 meters closer for every 1% increase in the immigrant population, with all other factors held constant. With every additional year, a 1% increase in the immigrant population relative to otherwise identical tracts would predict 1.4 meters further to the nearest garden, suggesting that after about 42 years the percent foreign born in a tract will have no impact on garden proximity, and ultimately after 2022 gardens will be further away from communities with higher shares of immigrants. As in Philadelphia, the gradual attenuation of garden accessibility for immigrants in Seattle may be linked to gentrification, as higher housing costs push vulnerable groups further from the more desirable areas, but this explanation cannot be verified from the model alone. What the model can tell us is the relationship between housing costs and garden locations in Seattle, as well as how this relationship changed over time. The coefficient for housing costs is significant and positive, suggesting that gardens were originally built further from high-demand real estate. With the coefficient for interaction between year and housing costs being negative, this relationship appears to be gradually weakening over time. Chapter 3 describes the widespread garden preservation that P-Patch advocates accomplished with the passage of Initiative 42, which offers a plausible explanation for why this pattern would be seen in Seattle: gardens were initially built where more land was available, and most of them have not been removed as property values in the surrounding neighborhoods have increased.
Spatial analysis indicates that the citywide programs in Milwaukee, Philadelphia, and Seattle have generally developed gardens closer to marginalized communities than to more privileged ones. That said, significant historical trends and a few deviations from the overall pattern are important to note, especially given their apparent relation to organizational decisions and political-economic factors described in previous chapters. First, while the models suggest that community gardens in all three cities have generally been closer to neighborhoods with more Black and Hispanic residents, their accessibility for Asian and Pacific Islander residents and for immigrants is not as consistent. On the one hand, vertical grow shelf studies of urban food access indicate that Black and Hispanic communities are the ones most impacted by lack of healthy, affordable food options , so if organizations are prioritizing the food-security benefits of urban agriculture, then building access for Black and Hispanic residents more than for Asian and Pacific Islander populations may genuinely reflect understandings of local need and equitable use of the organizations’ resources. On the other hand, food insecurity is just as acute in some Asian American and immigrant communities, and there is a chance these communities are being overlooked. Furthermore, food access isn’t the only benefit that community gardens bring; the organizations in this study have also emphasized social, cultural, and economic benefits of urban agriculture. Advocates for Seattle’s P-Patch program were the most explicit in touting the ability of gardens to build community among diverse people and to provide cultural continuity for immigrants from agrarian backgrounds. Perhaps because of this recognition, Seattle’s gardens have been the most accessible to immigrants according to the spatial error models. However, the P-Patches’ proximity to immigrants is eroding over time. In Milwaukee, gardens appear to be further away from communities with higher foreign-born populations, and in Philadelphia the relationship does not register as significant in the spatial error models. According to my interviews and review of organizational documents, in all three cities, immigrants—and in particular Southeast Asian immigrants—have been heavily involved in building gardens and organizing the labor required to keep them going. Yet it appears that immigrant gardeners may have to travel further than others to reach their sites, and they may lose access altogether if they or their garden is displaced when a neighborhood gentrifies. Qualitative researchers have drawn attention to some ways that immigrant gardeners may be undervalued by urban agriculture organizations and media accounts . My research suggests that this oversight may influence organizational priorities in development and preservation efforts, extending inequity to the physical siting of gardens. When community garden organizations do identify the benefits they want their spaces to provide and identify neighborhoods to prioritize in receiving those benefits, they can achieve desired outcomes over time. One example is the expansion of Seattle’s P-Patch network through the 1990s and 2000s, which was undertaken with conscious attention to increasing garden access for the city’s low-income residents. The spatial error model with interaction terms shows that initially, P-Patches were less accessible for communities with higher poverty rates, but this relationship flipped over time such that communities with higher poverty rates are now likely to be closer to the nearest garden than otherwise similar communities with lower poverty rates. Philadelphia Green provides another example of how programs can achieve clear outcomes by prioritizing a certain benefit that they want urban agriculture to provide in their city. In this case, the benefit has been economic.
As explained in Chapter 2, Pennsylvania Horticultural Society secured grant funding for its Philadelphia Green program to undertake concentrated neighborhood greening initiatives in the 1980s and 1990s; these initiatives included tracking how the greening affected the target neighborhoods, which helped the Society to make a broader case for public investment in their greening services. Scholarship based on the greening initiatives and organizational publications from the time highlight how the program’s community gardens and greening intervention improved neighborhood attractiveness and increased local property values. One organizational brochure includes a map showing, for different neighborhoods, the percentage of vacant lots involved in the program which had subsequently been sold and developed. Building and preserving green space for the benefit of disadvantaged neighborhoods was not the goal, and it has not been the primary outcome. Compared to Milwaukee and Seattle, Philadelphia has seen the highest rate of garden attrition, so a longitudinal analysis based on the distribution of existing gardens at given points in time misses some of the story. Still, even by examining the spatial error models and maps of garden locations over time, we can see distributional outcomes that are likely related to Philadelphia Green’s prioritization of economic benefits and limited efforts toward long-term garden preservation. The program’s gardens tended to be developed in neighborhoods with higher housing costs and lower poverty rates, but this relationship with housing costs has gradually diminished over time. Maps of garden locations in successive decades show that gardens have disappeared in neighborhoods where housing costs have increased and poverty rates have decreased. This pattern reflects the program’s overall weak commitment to maintaining gardens for the long term, allowing market forces to displace gardens from more desirable areas. Meanwhile, garden proximity to neighborhoods with a higher share of Black residents has decreased over time, which suggests that either gardens are disappearing at higher rates in neighborhoods with more Black residents, or that as neighborhoods themselves are changing through gentrification, those that keep their gardens are nevertheless seeing decreases in the proportion of Black residents. The pattern of garden distributions over time in Milwaukee demonstrates an outcome likely to result from program management without the resources or a clear strategy to direct garden development toward specific communities. The model with interaction terms did not yield any significant interactions with year, suggesting that garden distributions over time have not moved toward or away from communities with any of the characteristics analyzed . Instead, the static model shows that the nearest garden is likely to be closer to neighborhoods with higher poverty rates, higher percentages of Black, Hispanic, and/or Asian and Pacific Islander residents, and lower percentages of immigrants. Given what we know about the potential benefits of urban gardens and the communities most in need of those benefits, the distribution of Milwaukee’s gardens seems to produce equitable outcomes other than the lower proximity to neighborhoods with more immigrants. However, the historical analysis in preceding chapters demonstrates the ongoing vulnerability of most gardens in Milwaukee to potential removal in the face of development pressure. In other words, the gardens are close to the populations where they are needed because that is where land is available and development pressure is low; the process of displacement and disappointment that unfolded in Philadelphia is likely to be repeated in Milwaukee if and when market conditions change.