ORIGINAL RESEARCH published: 12 March 2021 doi: 10.3389/fevo.2021.621783 Three Histories of Greening and Whiteness in American Cities James J. T. Connolly 1,2* and Isabelle Anguelovski 2,3 1 School of Community and Regional Planning, University of British Columbia, Vancouver, BC, Canada, 2 Barcelona Lab for Urban Environmental Justice and Sustainability, Institute of Environmental Sciences and Technology (ICTA), Universitat Autònoma de Barcelona, Barcelona, Spain, 3 Catalan Institution for Research and Advanced Studies, Barcelona, Spain How has urban greening related to the degree of whiteness in neighborhoods? The answer to this question provides an essential “historical diagnostic” that can be used to develop an approach to urban ecology which integrates racial and ethnic change into the planning for proposed interventions. In this paper we employ state sequence analysis to analyze the historical trend of greening (including the implementation of new parks, greenways, community gardens, green recreation areas, and nature preserves) between 1975 and 2014 in a sample of nine cities in the United States relative to concentrations of white and non-white residents. We divide the nine cities into three common growth trajectories and separately examine the trends for each growth trajectory. We further illustrate these trends by mobilizing qualitative data from field Edited by: work in selected neighborhoods to help explain the processes that generate certain Kitty Connolly, key findings in the quantitative data. We find that the relationship between greening Theodore Payne Foundation, United States and race/ethnicity differs according to city-level growth trajectory. Cities with continuous Reviewed by: high and rapid levels of growth in the postwar period have the strongest link between Lindsay Kathleen Campbell, increased greening and whiter populations. Meanwhile, in cities that contracted or had United States Forest Service (USDA), a punctuated growth pattern, non-white areas had a uniformly low level of greening United States Iryna Dronova, that occurred mostly in recent years. In all, we show how urban growth, greening, University of California, Berkeley, and whiteness are inextricably associated qualities of American cities. We argue that United States understanding this association is essential for development of a race-conscious model *Correspondence: James J. T. Connolly for enhancing urban ecosystems. james.connolly@ubc.ca Keywords: race and greening, green justice, urban greening, green gentrification, social-ecological conflict, urban ecology and sustainability Specialty section: This article was submitted to Urban Ecology, a section of the journal INTRODUCTION Frontiers in Ecology and Evolution Efforts to enhance urban ecosystems, including greening, nature-based solutions, and Received: 27 October 2020 environmental sustainability initiatives, have been shown to have substantial shortcomings with Accepted: 08 February 2021 regard to lack of recognition of the racist and white supremacist histories upon which they Published: 12 March 2021 are built (Pulido, 2000, 2017; Safransky, 2014). These critiques reflect a deeper concern for the Citation: extent to which social injustices persist in the context of an increasingly green urban planning Connolly JJT and Anguelovski I (2021) Three Histories of Greening orthodoxy (Connolly, 2019; Anguelovski et al., 2020). According to this critique, a great deal of the and Whiteness in American Cities. modern environmental movement (Curnow and Helferty, 2018), urban sustainability and resilience Front. Ecol. Evol. 9:621783. planning (Goodling et al., 2015; Walsh, 2018), and green space initiatives (Anguelovski et al., 2018) doi: 10.3389/fevo.2021.621783 proceed as if the racial and social context in which they operate is immaterial to the urban ecological Frontiers in Ecology and Evolution | www.frontiersin.org 1 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness initiatives enacted. Some even find greening efforts to be In this paper, we address this lack of generalized knowledge part of a conscious plan for gentrifying areas, causing by further developing the historical diagnostic needed for a displacement, and social exclusion along racial/ethnic lines more race-conscious urban ecological practice to take hold. (Gould and Lewis, 2016). We employ state sequence analysis to understand the historical In recognition of this critique, cities have already begun to pattern of increases in new green spaces and racial change rethink the approach to greening. Philadelphia’s Green Works between 1975 and 2014 in nine American cities. The nine Equity Index (City of Philadelphia, 2016) and resource allocation cities were chosen to be representative of three predominant strategies around Los Angeles’s Measure A (Hingorani, 2019) growth trajectories (described below). We specifically ask: What have recently proposed equity-centered measures or indicators are the common sequences across all nine cities with regard for green planning. Though these examples remain exceptions for changes in green space and changes in the racial makeup in the American landscape, they point toward a situation where of surrounding areas? Does this pattern vary according to cities are seeing more and more reasons to avoid uncritical the historical growth trajectory of the city? Can a generalized approaches to greening, and these reasons are not confined to a relationship be determined with regard to the relationship vague concern for social justice. Rather, as some scholars have between greening and racial change in these cities? In the sections already argued, to the extent that urban greening initiatives help that follow, we describe the data and methods used to understand retrace racist and white supremacist tendencies onto present the relationship between green spaces and racial change in the urban development (Safransky, 2018), social justice goals are nine cities that we study. Next, we present our results of the undermined, but so too are goals related to generating robust state sequence analysis and illustrate three key trends in our data urban ecosystems (Ernstson, 2013; Langemeyer and Connolly, through case examples drawn from qualitative field work. Finally, 2020; Schell et al., 2020). we discuss the implications of the findings for the development Safransky (2018) points toward the need to understand the of a race-conscious model for enhancing justice-centered urban historical trend of greening and racialized change in cities as ecologies and summarize the key conclusions relative to our a key step toward addressing the drawbacks of de-politicized research questions. green planning. She describes the “historical diagnostic” tools needed for the field of urban ecology to have the capacity to account for racist and ethnocentric effects. These tools, she MATERIALS AND METHODS argues, lead toward modes of enhancing urban ecosystems that are “attuned to the moral claims that historical dispossessions This paper is a sub-study of a large international European make on the present (Bird Rose, 2004, as cited in Safransky, 2018). Research Council funded project (GREENLULUS, 2016–2021) In its pursuit of justice, an historical diagnostic is concerned focused on the conditions under which urban re-naturing with the task of illuminating hidden histories that point toward projects in distressed neighborhoods redistribute (or not) access alternatives, decolonization, and the challenge of recuperation.” to green/blue spaces for historically marginalized groups in mid- Thus, Safransky pushes beyond a theoretical framing of the sized cities in the United States, Canada, and Western Europe. problem and toward a call for development of empirical tools The data and methods for the overall project from which this and information that reshape the goals of urban ecological paper draws were chosen with an eye toward developing a more planning agendas toward a more consciously just model. These generalized notion of greening and social change over time than is tools, we agree, are essential for cities to connect greening and currently available in the literature. For the quantitative analysis social justice goals. of this paper, we selected cities based on a balance between A racially or ethnically focused historical diagnostic useful data availability and representativeness with regard for growth for developing a more just set of interventions in urban trajectories. For our qualitative case examples, we chose a further ecology would provide a thorough understanding of the general sub-set of three neighborhoods that represent theoretically and patterns of association between urban greening and racial change practically important trends shown in the quantitative data and in cities (for some examples of work in this direction see that were included in our qualitative data collection based on Boone et al., 2009; Grove et al., 2018; Schell et al., 2020; see in-person fieldwork in 2018, 2019, and 2020. also work on legacy effects of urban greening, e.g., Roman et al., 2018). More specifically, it would identify the extent, City Selection periodicity, and pattern over time of planned changes in urban We hypothesize that the relationship between urban greening ecosystems and changes in the racial makeup across a wide and racial change differs according to the overall dynamics of variety of urban settings. With regard to urban greening, the urbanization and growth within the city of interest. In order development of such an historical diagnostic remains hindered to test this hypothesis, we selected nine cities representative of by a lack of overall understanding of how commonly greening three common trajectories of growth in the United States during is associated with racial and ethnic change; the type of greening the post-war period (covering the transition to post-industrial and racial/ethnic change that generates an association; and American urbanization that began in the 1950s, but was clearly the conditions under which greening is not associated with evident by the 1970s). We use data from the United States Census racial/ethnic change. That is, we lack generalized knowledge of on changes in population, race, and city-level Gross Domestic the overall race and ethnicity trends that coincide with urban Product (GDP) in order to differentiate cities according to general ecological interventions. trajectories of growth (Manyika et al., 2012). We acknowledge Frontiers in Ecology and Evolution | www.frontiersin.org 2 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness that these are simplified indicators of deeper cultural, social, example, the 2008 bankruptcy of General Motors collapsed the and political dynamics that have played out in the context of regional economy and contributed to the subsequent further a complex shift from “Fordist” production models rooted in estimated loss of 40,000 residents between 2009 and 2019 (US the political economy of nation states to “post-Fordist” models Census Bureau, 2019). These cities have seen the lowest rates rooted in globalized systems (Marcuse, 1997; Walks, 2001) and of increase in GDP since 1950 relative to other large American embodied in the skilled creative, high technology, biotechnology, cities. These cities are also characterized by the highest rates education, or finance sectors (Glaeser and Saiz, 2003; Peck, 2005). of increase in non-white population since 1950. The cities Primarily, we relied on population changes as the key indicator in this study representing those that experienced continuous of a city’s trajectory (see also Park and LaFrombois, 2019), with contraction include Detroit (Michigan), Baltimore (Maryland), race and GDP data as secondary indicators. The GDP data and Cleveland (Ohio). trends largely follow changes in population, with some deviations Punctuated growth cities gained prominence during the early described in the text below, but for simplicity’s sake we visualize Industrial Era in the United States, and then had declining rates only the population data. The nine cities selected are shown in of population and economic growth in the decades following Figure 1 with their population trends between 1950 and 2018 1950, but saw a resurgence between 1980 and 2010. Generally, (the most recent available data). The trend lines are color coded to by the 1980s, large sections of these cities were substantially denote cities with similar growth trends. Note that New York City disinvested after a process commonly known as white flight is shown with the data adjusted down to 25% of the population in to the suburbs. However, between the 1980s and early 2000s, order to be on a similar scale as the other cities. rapid and high reinvestment and economic conversion, with Since the second world war, these nine cities took one of a new prominence gained by the high technology, financial, three growth trajectories. They either experienced a more or less and/or pharmaceutical industries, resulted in GDP increase and steady process of expansion; a more or less continuous process population comeback, including increased white population after of contraction, or a punctuated process wherein population and decades of decreases in this area. For example, companies such GDP declined heavily but then saw a resurgence to new highs. as Merck or Johnson and Johnson contributed to Philadelphia’s As a shorthand, we call these three trajectories “steady growth” resurgence while San Francisco’s new economy was driven by cities (Austin, Portland, Seattle); “continuous contraction” Silicon Valley tech giants Google and Apple as well as newer cities (Cleveland, Detroit, Baltimore); and “punctuated growth” firms, like Airbnb and Uber. Meanwhile, punctuated growth cities (New York, San Francisco, Philadelphia). With distinct cities saw continuous increases in non-white population and population growth, economic growth, and racial change currently have a moderate non-white percentage relative to other trajectories, we expect that the different levels of resources and US cities. GDP also increased at a moderate rate relative to other cultural conditions in these three categories of cities led to American cities between 2000 and 2018. The cities considered different relationships with regard to race/ethnicity and greening. to be representative of the punctuated growth category that The generalized trend line of population growth for these were analyzed in this study include San Francisco (California), categories of cities is shown in Figure 2 and the three trajectories New York (New York), and Philadelphia (Pennsylvania). While are explained in greater detail below. San Francisco and New York had a clear resurgence by the 1990s, Steady growth cities generally were not the most prominent Philadelphia was slower in terms of population and GDP growth, urban centers during the Industrial Era in the United States, but but its trajectory by the mid-2000s put it in this category. became prototypical attractor cities during the latter part of the twentieth century. Often, these cities were home to emergent technology and retail firms (e.g., Microsoft and Starbucks in Tract-Level Population and Green Space Seattle; Dell and Whole Foods in Austin; Intel and Nike in Data Portland) and gained numerous associated economic spin-off The population data used to develop the census tract-level benefits that generated rapid and relatively high levels of growth analysis in this study is derived from the Longitudinal Tract in population and GDP by the early 2000s. Overall, these cities Database (Logan et al., 2014). This database provides a have been mostly white throughout their histories, with a long standardized estimate of numerous demographic variables from history of housing discrimination and urban renewal programs. the United States Census between 1970 and 2010. It was They have seen increases in non-white population since 1950, but developed to be the best possible data for analyzing change remain majority white and are among the whitest of the large across time in American urban neighborhoods. We utilize the American cities. GDP since 1950, and especially since 1990, has race and population estimates standardized to 2010 census tract risen at a relatively fast rate in steady growth cities compared to boundaries, which cover the full period of time used in this study. other American cities. The cities representative of this category The green space data for the nine cities analyzed is derived that were analyzed here include Seattle (Washington), Portland from a specialized dataset we assembled and created that contains (Oregon), and Austin (Texas). the boundaries of all public green spaces in each city as of 2016, Continuous contraction cities were prominent Industrial Era along with the type of space (classified in a standardized way cities that saw a steady decline in population and GDP since by park, greenway, preserve, garden, recreation), the size of the 1950, without any period of substantial resurgence. Generally, space, and the year the space was acquired for public use as these cities are characterized, like most “rustbelt” cities, by a greenspace (if this year was not available, then the year the disinvestment from industrial urban centers and white flight space was opened to the public was used instead). Greenspaces to the suburbs, except in a few neighborhoods. In Detroit, for in the dataset include all formal and informal public spaces Frontiers in Ecology and Evolution | www.frontiersin.org 3 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness FIGURE 1 | Selected city population trends, 1950–2018. Data source: US Census. For population between 1950 and 1990, see Gibson and Jung (2002); for population between 2000 and 2010, see censusviewer.com; for population 2018, see censusreporter.org. FIGURE 2 | Generalized trend line of population growth in three types of American cities. for each of the five types analyzed here (i.e., parks, greenways, to these cities in that time. Please see the methodological note preserves, gardens, and recreation) and are derived from city files in the Supplementary Material for a more specific description augmented by aerial imagery analysis (see methodological note). of the source and procedure used to acquire and develop the For each of the nine cities analyzed in this study, the year the green space data. space was acquired/opened is known for at least 90% of all of the This analysis does not include private residential yards or qualifying green spaces citywide. Thus, we are reasonably sure landscaped areas, street trees, or most greenspaces built as part of that the data we analyze between 1975 and 2014 represents the road infrastructure. Thus, the focus here is not on a generalized full set (or very near to the full set) of public green spaces added measure of “greenness,” but rather on a specific measure of Frontiers in Ecology and Evolution | www.frontiersin.org 4 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness publicly supported urban green spaces designed for active use The area change (ac) value is the degree of change in green or habitat preservation. It is worth acknowledging that a small space area in a tract for a given time period relative to the median number of especially informal spaces likely came and went (e.g., change in all populated tracts that added green space throughout a space that started in 1991 but was defunct by 1998) and thus the city (i.e., a different median value is used for each city and each could not be identified when the dataset was developed. time period). There is also a zero category for tracts that added no green space during the time period (note that zero tracts are not included in the median calculations). The “area median” value Assigning States of Greening and refers to the median percent of area that became green for those Racial/Ethnic Change in Tracts tracts that added greenspace (i.e., all non-zero tracts) in the city In order to calculate the amount of green space in a tract at a given containing the tract. The “area tract” value refers to the percent time, it was first assumed that every green space has a catchment of land area within city limits that became green in the tract (plus area wherein nearby residents are easily drawn to it. While it buffer) for which the ac value is being calculated. For example, if is likely that this catchment area is not uniform across green the median new green space area coverage of all populated tracts spaces, the widely used measure of 400 m as an easily walkable that added green space in a given city between 1975 and 1984 was distance (approximately 10 min) provides a reasonable average 8% and a given tract had 4% new green space coverage during baseline estimate of a standard catchment for a greenspace. For that time, then “ac” for that period is equal to -0.5, indicating the this reason, we calculate the number and area of green spaces tract was 50% below the median. If the given tract has 12% new in a tract by first creating a 400-m buffer around each tract. We green space, then ac is +0.5, indicating the tract was 50% above then use the formal tract boundary plus the 400-m buffer for all the median. These ratings were then divided into four “states of green space calculations by tract. This procedure is important greening” described in Table 1 below. for controlling boundary effects, wherein a tract is directly next In order to measure the state of racial and ethnic mix in a to a green space (making it accessible to residents of the tract). tract, we use data developed by Zhang and Logan (2016), which Without the inclusion of the buffer area, that green space would is freely available in the Longitudinal Tract Database (LTDB). not be included in the tract total. Further, in the cases where green For their project looking at racial and ethnic segregation in US spaces comprise most of a tract area but do not reach beyond, this cities, Zhang and Logan developed a classification for each tract step is required in order to retain that green space in the analysis. that designates the presence by tract of whites (w), blacks (b), Once the catchment areas were defined, new green spaces Hispanics (h), and Asians (a). In order to designate groups as were identified for each tract according to 10-year time periods present or absent from a tract, the authors identify a threshold in order to allow for increases in green space provision to be wherein any group with a share in the tract that is less than one- calculated by tract. Only tracts that had at least some greenspace quarter of their average share in all large and multi-ethnic US added during the study period (1975–2014) were included in metros is considered so underrepresented as to be “absent” (for the analysis below. Green spaces were assigned to analysis years more detail, see Logan and Zhang, 2010; Zhang and Logan, 2016). according to spaces added in the 5 years before and 4 years after The minimum levels used as cutoff points vary from decade each decennial census year. For example, green spaces added to decade, reflecting the changing overall composition of the between 1975 and 1984 were assigned to census data from 1980. country (minimum cutoffs for whites to be considered “present” This approach is meant to capture the “announcement effect” are: 1980:20.7%; 1990:20.2%; 2000:18.9%; 2010:17.8%). Crucially, of green spaces (Immergluck and Balan, 2018), so that all new this approach allows researchers to distinguish differences that spaces that were either built or likely announced during the are inherent to regions’ demographic composition from those 10 years closest to the one being measured in the census data that simply reveal overall demographic shifts. are associated with that year. Time periods to which green spaces were assigned are: TABLE 1 | Greening states: classifications of new green space area in a tract. Period 1. Pre-1975: all green spaces built before 1975 Period 2. 1980 census data: 1975–1984 green space data Green space area Label Description Period 3. 1990 census data: 1985–1994 green space data classification Period 4. 2000 census data: 1995–2004 green space data Zero new green Z There were no new green spaces added to the Period 5. 2010 census data: 2005–2014 green space data space area tract. Low new green A The change in percent of tract area occupied A categorical grouping based on distance from the median was space area by new green space is notably below the developed to reflect the “state” of greening for each time period. median (50% or greater) across all cities. These categories compare the percent of the tract (plus its 400- Moderate new B The change in percent of tract area occupied green space area by new screen space is near the median (less m buffer within city limits) occupied by new green space at a than 50% above or below) across all cities given time with the median percent across all populated tracts High new green C The change in percent of tract area occupied that added green space in the city at that time. The formula is: space area by new green space is notably above the median (50% or greater) across all cities.   area median − area tract area change (ac) = −1 Bold values refer to the relationship that each category has with the median (bolded area median for visual clarity). Frontiers in Ecology and Evolution | www.frontiersin.org 5 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness Using the same threshold as Zhang and Logan (2016), this We developed the state sequence dataset and analysis using study employs a simple two-state designation, wherein tracts are the latest version (2.2.0.1) of the “TraMineR” package in R assigned “w” if a white population is present above the Zhang- software (version 4.03). For the first step, we built a table Logan threshold of one-quarter of the average share of whites in the state sequence (STS) format that included the full in all large and multi-ethnic US metros for the given year and demographic data; the city classification; the categorical state “n” if the white population is not present above the threshold. of greening (zero, low, moderate, high); and the binary (“white While there is good reason to examine all racial/ethnic categories, present” or “non-white”) state of race/ethnicity for each tract this study is primarily focused on the shifting states of whiteness in the nine cities. For the second step, identifying pairwise relative to greening. Thus, for the sake of not generating an overly dissimilarity, we utilized the Dynamic Hammond Distance complex number of possible states (which would muddle the (DHD) method for calculating how close one race/ethnicity- state sequence analysis process), the racial/ethnic classifications greening sequence was to another. We selected the DHD method are simplified to those that, basically, designate which tracts have over the commonly used Optimal Matching (OM) approach a sufficient enough presence of whites as to make the white because DHD is ideal for data where all sequences have the same population “present.” In other words, we are looking at tracts length and difference across each time period is important from where the presence of whites is large and visible enough (roughly theoretical and analytical perspectives (Lesnard, 2009). For the 20% or higher) as to make them an easily perceived part of the third step, we employ the Ward’s clustering technique with a neighborhood makeup. three-cluster solution in order to identify typical race/ethnicity- greening sequences for the tracts included in our study. The Ward’s cluster method is designed to minimize the total within State Sequence Analysis cluster variance, thus generating homogeneous groupings of Neighborhoods are far from static. Rather, current conditions data. In line with the explorative dimension of this method, in a given area of a city are derived from a complex history we selected the three-cluster solution after analysis of several of demographic and physical change (Grove et al., 2018). We solutions between two and eight, and subjectively determining focus on two of these changes—greening of the local environment that the three-cluster approach produced results that best met and racial/ethnic makeup of the local population. Given the the goal of differentiating the key differences among race- above data, we examine the typical sequences followed by the greening pathways. combined state of greening and racial change between 1975 and 2014. We use the eight possible states (four states of greening multiplied by two racial/ethnic states) as the basis Illustrating Quantitative Trends With upon which to undertake this examination. For example, a tract Qualitative Data state for a given time period might be “Aw,” meaning it had We selected neighborhoods in Austin (steady growth), a relatively low amount of new greening and white population San Francisco (punctuated growth), and Cleveland (continuous was present above the Zhang-Logan threshold for that year. We contraction) as three qualitative case examples to illustrate examine, as well, if these typical sequences varied by type of city selected on-the-ground processes that generate one of the three using the three-city growth trajectory classifications described typical sequences of greening and racial change in each city above. Finally, following Borgna and Struffolino (2018) we type. Within the scope of this study, we could not undertake connected the typical sequences with qualitative data (described qualitative analysis of all of the typical sequences, so we selected below) from a neighborhood representing one cluster in each neighborhoods that reflect theoretically and practically important type of city as a way of enriching our understanding of the trends within the quantitative data. All three qualitative cases general trends found. shared a municipal commitment to developing new greening In order to determine the sequence pattern of change in interventions since the 1990s. greening and racial makeup of tracts, we employ state sequence In each city, we conducted between 25 and 30 semi-structured (STS) analysis. Overall, sequence analysis allows sets of coded interviews after pretesting, modifying and selecting a final set categorical data ordered by time to be understood according to of questions for our interview protocol. We spent 1 month in general trends, including identification of typical and deviant 2018 and 2019 conducting fieldwork in each neighborhood/city, sequences, and parsing out of difference between groups (such enrolling a variety of participants to maximize the heterogeneity as cities with different growth trajectories) (Ritschard and Studer, of experiences and expertise on the topics of urban greening; 2018). There are three general steps to sequence analysis, each urban development and racialization; and inequalities; covering of which requires making methodological choices (Rossignon a range of themes related to re-naturalizing interventions; et al., 2018). The first step is to code data according to categorical equity/justice; segregation; and gentrification. We contacted sequence states and time periods (this step is described above). initial respondents through previous research, expert contacts, The second step is to measure the pairwise dissimilarity internet searches, and review of local media articles about between sequences. In this case, that means measuring tract-level each case. We then contacted further respondents through differences in the sequence of greening and racial change. The snowball sampling, until reaching theoretical and stakeholder final step is to summarize the dissimilarity between sequences diversity saturation. Respondents included elected officials; with the aid of a clustering algorithm. This final step allows for urban planning and environmental staff; community-based typical sequences to be identified. organizations; neighborhood resident leaders; and other local Frontiers in Ecology and Evolution | www.frontiersin.org 6 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness stakeholders (such as developers, designers, or local experts in thresholds, respectively. With all cities considered together, the area of urban greening). Although we did not purposely over the entire time period, these tracts were overrepresented seek older respondents, a variety of respondents (long-term with greening according to the simple (lower) but not the neighborhood activists, environmental non-profits, and/or urban Zhang-Logan (higher) threshold. The Zhang-Logan threshold environmental planners) in each city were able to provide tracts were only overrepresented with greening in the 2000 and historical insights and perspectives about greening changes. 2010 periods. Generally, the overrepresentation of greening in Upon the full transcription of 90% of our interviews, we tracts that became whiter between 1980 and 2010 appears to be a created a detailed coding scheme based on the main questions phenomenon that was largely driven by initiatives that may have addressed by the parent research study. Two primary researchers been planned in the 1990s but were rolled out in the 2000s and developed the overarching coding scheme, which was then 2010s. However, importantly, this is not a uniform trend across slightly altered and clarified after extensive team discussions. all three city types. Coding was completed by one researcher per city, with regular By type of city, there are unique patterns with regard to the check-ins and verification between all team members to ensure overall relationship between greening and racial/ethnic change. intercoder reliability. Last, we used gray literature to document First, regardless of the threshold used, steady growth cities show the greening and urban development trajectory and history of an overrepresentation of greening in tracts that became whiter each city, including archival and newspaper articles, non-profit for the later periods of the study. Within steady growth cities, organizations’ reports, and media articles. when looking at the stricter Zhang-Logan threshold, the tracts that got whiter received new green space area that was roughly twice the percent of area covered by the tracts between 1995 RESULTS and 2014 (which corresponds to the 2000 and 2010 census time periods in the table). This trend for steady growth cities This study analyzes the relationship between greening and increases for tracts with any increase in whiteness to an area- changes in race/ethnicity in tracts that had some greenspace based overrepresentation of roughly 2.5. added between 1975 and 2014 across nine American cities Next, continuous contraction cities show no classified according to three growth trajectories. For each of the overrepresentation of greening in tracts that became whiter three growth trajectories (and for all cities together), we report for the period studied. This trend holds regardless of the below a summary of the overall relationship between changes threshold used for designating tracts that became whiter. Two in race/ethnicity and greening and then visualize the typical underlying dynamics are important to consider for these cities. trajectories (state sequence clusters) for tracts that greened. We First, there were very few tracts that became whiter. These cities then mobilize the qualitative data to report relevant on-the- generally experienced a decline in white population overall ground characteristics of neighborhoods containing tracts that for the time period studied. Second, the area of formal green represent a state sequence cluster. spaces in the cities was largely stagnant or declining during this time period due to budget restraints but new informal gardens Summary of Quantitative Findings and local greening initiatives grew in non-white tracts. Put Across All Cities differently, there was an emphasis in continuous contraction Overall, by area, tracts that shifted toward a whiter population cities on smaller, informal greening that tended to occur in were overrepresented with new green initiatives beginning mostly mostly non-white tracts. in the 1990s–2000s (not prior), but there are differences in this Finally, for punctuated growth cities, greening-racial/ethnic trend by city type. Table 2 shows the percent of all tract area change has a longer and more episodic history. In these with population that shifted whiter in two ways. In the top half cities, the Zhang-Logan threshold shows no overrepresentation of the table, we use the Zhang-Logan threshold (which requires of greening in tracts that became whiter, but the simple that white population is present at or above 25% of the average increase threshold offers a different picture. According to the across all large and multi-ethnic American metropolitan areas simple threshold (bottom half of Table 2), punctuated growth at the time—this means the tract had at least 17–21% white cities had overrepresentation of greening in tracts that became population depending on the year), to identify tracts that shifted whiter in the period between 1985 and 1994 (marked as the whiter. Since the threshold (17–21%) needed to meet the Zhang- 1990 census year in the table) and the period between 2005 Logan standard for “present” is somewhat high, the bottom half and 2014 (marked as the 2010 census year in the table). of the table also reports the tracts that saw any increase in white These periods coincide with the efforts of early “frontier” population in order to provide a more complete picture. The gentrifiers of the 1980s in these cities (Smith, 1996) and Zhang-Logan approach is focused on identifying neighborhoods the early push toward local sustainability initiatives of the where white population has a strong and visible foothold and the 1990s, many of which involved urban greening efforts in simple increase approach is focused on identifying areas where large cities. New greening efforts also took place toward the whiteness is increasing to any degree. Both approaches matter for end of the 2000s, especially so in formerly industrial districts different substantive reasons. such as Hunters Point-Bayview in San Francisco or north of Across all nine cities, 3.17 and 8.47% of the populated tract Center City in Philadelphia. In all, during clearly identifiable area (i.e., percent of area covered by tracts—not percent of the periods where punctuated growth cities pushed new greening number of tracts—was used to control for variation in tract initiatives, tracts that became whiter were overrepresented by size) shifted whiter according to the Zhang-Logan and simple area with greening. Frontiers in Ecology and Evolution | www.frontiersin.org 7 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness TABLE 2 | Representativeness of new greening in tracts that shifted toward a whiter population (notable shift above the Zhang and Logan threshold and any shift) between 1980 and 2010. Year All cities (%) Steady growth Punctuated Continuous cities (%) growth cities (%) contraction cities (%) Percent of all tract area that shifted notably white (above Zhang-Logan threshold) 3.17 2.17 4.60 1.18 1980–2010 2.38 2.77 2.19 0.48 1980 2.77 1.92 3.79 0.52 1990 0.13 0.08 0.19 0.00 2000 5.87 4.94 0.83 0.31 2010 3.39 4.16 1.10 0.64 Percent of all tract area with any shift white 8.47 8.62 16.97 8.78 Percent of all new green in tracts with any shift white 1980–2010 11.32 8.02 17.71 4.57 1980 3.23 3.45 2.79 4.87 1990 23.10 2.28 49.15 4.17 2000 6.49 7.34 5.12 3.66 2010 18.76 21.49 17.21 5.08 Bold values highlight the instances where greening was over represented in tracts that shifted white relative to overall area of tracts that shifted white. Steady Growth Cities driving much of the sector, which thrust the city into the heart When considering the data for steady growth cities that had of the global economy. The growth in technology design and a more or less linear upward growth trend over the period production industries was sparked by strategic investments and studied, the three most representative clusters of state sequences a local culture that was increasingly attractive to a highly paid in Figure 3 demonstrate a strong link between greening and white young workforce (Smilor et al., 1989), but the fact that the city populations. For each cluster type, Figures 3–5 show the two had extensively preserved its natural ecosystem and park system, most frequent sequences in black and gray lines (reflecting the with enviable spaces such as Barton Springs Greenbelt, Walnut modal trend) for that cluster. For a more detailed depiction of Creek Metropolitan Park, Red Bud Isle, or Mayfield Park, was the sequences in each cluster see the Supplementary Material. essential to the city’s attractiveness. High income technology The first cluster of tracts (Type 1, n = 63), labeled “high workers could afford to choose between locations based on such green/white,” mostly stayed white over the entire period and intangibles as the local environment, so cities like Austin with a had intense periods of greening. The most common trends rich store of green amenities and attractive local culture had an for “high green/white” tracts were for large areas of greening advantage (Florida, 2004). to occur in these tracts during the period between 1985 and Austin wasn’t designed to accommodate the rapid growth 2005, with lower levels outside of that time period. The second it experienced at the turn of the twentieth century, and so cluster of tracts (Type 2, n = 205), labeled “low green/white” had to plan quickly for where the new development would go. also mostly stayed white over the entire period, but had lower Due to a racist historic land use policy passed in the 1930s levels of greening scattered throughout. The third cluster of tracts that required (through legal mandate) non-white residents to (Type 3, n = 13), labeled “high green/shift white,” mostly shifted move to the east side of town, which was also where most from non-white to white immediately after or immediately heavy industry was concentrated, the most underdeveloped area before intensive periods of greening, and thus demonstrate was also where many people of color lived. In the 1980s, the historical pathway of co-occurring greening and increased community groups in the East Austin neighborhood began whiteness in areas that displace people of color. Interestingly, organizing to clean up and remove many of the industrial the “high green/shift white” cluster of tracts shows that intense sites near their homes. One organizer said of the time, “. . .we greening may come before or after the shift toward whiteness in made history by rezoning over 600 properties. . .you had to steady growth cities. address what was going on but then you had to address what In order to demonstrate how these historical patterns in the was allowing all these polluting and industrial facilities in our relationship between greening and whiteness translate into on- communities, and it was through land use and zoning that the-ground experiences in steady growth cities, we draw on was doing that.” As a result of these changes to land use and qualitative data gathered in the central East Austin neighborhood the greening initiatives that were underway by the mid-1990s, of Austin, Texas, which contains core tracts that fall within the East Austin was seen by many as “ready” to receive the new “high green/shift white” (Type 3) cluster. In the 1990s, a lot development, a crucial condition for its formal designation by the changed for Austin. It transformed from a relatively sleepy college city as a “desired development zone,” which shifted new growth and government town into the fastest growing large US city in that direction. (Kolko, 2016), labeled the second-most overvalued real estate By the early 2000s, a new development push had taken root market in the country and the “most difficult” real estate market in East Austin, which by then was becoming one of the most in Texas (Gross, 2018). Its growth was propelled by the high rapidly gentrifying neighborhoods in the country associated with technology industry, with firms such as Dell, AMD, and others widespread displacement of low-income people of color (Maciag, with secondary offices including Adobe, Amazon, and Apple 2015). Meanwhile, a new round of greening and associated Frontiers in Ecology and Evolution | www.frontiersin.org 8 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness FIGURE 3 | Steady growth cities typical trajectories: three-cluster solution for race/ethnicity-greening trajectories of all tracts that had any greening between 1975 and 2014. The black line represents the most common sequence and the gray line represents the second-most common sequence. Together, these lines reflect the modal sequence, or most common trend, within a cluster. Detailed cluster results can be found in the Supplementary Material. green rhetoric (Garcia-Lamarca et al., 2019) was used as a way Continuous Contraction Cities to further draw development to East Austin. One city official When considering the data for continuous contraction cities, commented, “the emphasis [became]. . .getting some parkland the three most representative clusters of state sequences shown established out ahead of development actually to help lure in Figure 4 demonstrate that, while small-area greening development out toward that way [the east side]. . .We were has widely occurred in non-white tracts, the most intense trying to buy land out there ahead of development and hopefully levels of greening have mostly occurred in white tracts. The draw development out there with parkland. And it was hugely first cluster of tracts (Type 1, n = 136), labeled “variable successful. Now all of the development has followed those green/white,” retained a notable level of white population tracts of land.” (above the Zhang-Logan threshold of at least 25% of the In the end, greening was used to “prepare” East Austin nationwide average for whites across all metros) throughout for development in two ways at two different time periods, the study period with various levels of greening including both of which show up in the Type 3 steady growth cluster. high- and low-area greening concentrated in the 1980s Intensive greening in the early 1990s was focused on cleaning and the 2000s. Most importantly, “variable green/white” up formerly industrial land in such a way as to make the area tracts are the only areas of continuous contraction cities ripe for designation as a “desired development zone.” Then, that saw any high-area greening and are the only cluster intensive greening in the early- to mid-2000s took the form of type that stayed white throughout the study period. The new parks and greenways as a “lure” for development to the second cluster of tracts (Type 2, n = 26), labeled “low area, with acquisitions for parks such as Roy Guerrero Metro green/shift non-white,” changed from white to non-white Park and East Boggy creek Greenbelt, and contributing to green (white population dipped below the threshold to be considered gentrification trends (Long, 2016; Garcia-Lamarca et al., 2019). “present”) in the 2000s with low-area greening carried out That “high green/shift white” clusters are one of the typical in the non-white period, but not prior. The third cluster historical trajectories in steady growth cities shows that the of tracts (Type 3, n = 134), labeled “low green/non-white,” intensive greening associated with a shift from non-white to stayed non-white throughout the study period (or, in some white population seen in East Austin was a defining characteristic cases became non-white very early in the study period) across these cities. with low-area greening concentrated in the latter years. Frontiers in Ecology and Evolution | www.frontiersin.org 9 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness FIGURE 4 | Continuous contraction cities typical trajectories: three-cluster solution for race/ethnicity-greening trajectories of all tracts that had any greening between 1975 and 2014. The black line represents the most common sequence and the gray line represents the second-most common sequence. Together, these lines reflect the modal sequence, or most common trend, within a cluster. Detailed cluster results can be found in the Supplementary Material. There are thus two major, polarized trends in continuous targets for reinvestment or gentrification by private developers contraction cities: limited but intense new greening in and homeowners. white areas only and widespread but small-area greening in Since the late 2000s, the restoration of the Cuyahoga River and non-white areas. Lake Erie, as part of the transformation to rebrand Cleveland In order to demonstrate one of these clusters, we focus on as “a green city on a blue lake” and depict both spaces as qualitative data gathered in the Detroit Shoreway neighborhood assets (Samanta and Kellogg, 2019), has been triggering other of Cleveland, Ohio, an area containing several tracts in the nearby environmental efforts. Neighborhoods like Tremont, “variable green/white” (Type 1) cluster of continuous contraction Ohio City, and Detroit Shoreway are being revived through new cities. Since the 1960s, drastic economic and population housing developments and greening investments. As of 2018, decline have defined Cleveland’s trajectory, with blue-collar the Northern section of Detroit Shoreway has seen new luxury employment plummeting to only 29% in 1986, a population housing built next to the restored Edgewater park. Although the loss of 380,000 inhabitants by 2019 (US Census Bureau, 2019), park is now an attractive amenity with lakeside beaches, boat and thousands of uninhabited homes and vacant lots. From docks and fishing areas, and picnic sites, local black families an environmental justice standpoint, the Cuyahoga River and report being unable to find affordable housing in its vicinity, its adjacent black areas have historically been the symbol and thus to enjoy greater proximity to the new green spaces. of environmental degradation and industrial and other types As one resident shared, “In the 5 years since my son bought his of waste contamination. Since the beginning of the 2010s, house [in the Detroit Shoreway neighborhood], there’s no house however, Cleveland has experienced some cultural, art, and in this [Detroit Shoreway] neighborhood that they [my daughter food-based revival, with new real estate development and and her partner] could find, that’s move-in ready that they could investment, mostly in racially mixed areas of the city that afford. And they’re not, they’re not super low income.” would fall into the Type 1 cluster (whites are present above Developers reported during interviews that the new green the Zhang-Logan threshold of at least 25% of the average revival of part of the neighborhood was a major asset, and across all metros) for continuous contraction urban areas. In particularly highlighted their ability to sell high-end homes contrast, one respondent stated what they saw as an impression by marketing Edgewater Park and another recent park called held widely by local residents that some other areas, like Battery Park as a core asset in Detroit Shoreway. The East Cleveland, are “too poor and too black” to be seen as Detroit Shoreway neighborhood is close to downtown, close Frontiers in Ecology and Evolution | www.frontiersin.org 10 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness to motorized high-speed roads, full of sit-down restaurants In order to demonstrate how these trends translated and shops, and also hosts a sizable white population (49%), into on-the-ground experiences, we draw on qualitative data all making up the “attractiveness” of Detroit Shoreway to gathered in the Bayview-Hunters Point neighborhood of outsiders. One of them highlights the green value of Battery San Francisco, which contains core tracts that fall within the Park for residents looking for a more dense, proximate- Type 3 cluster for Punctuated Growth cities. San Francisco to-downtown “sustainable lifestyle”: “From Battery Park you has been a node in the globalized, tech-driven economy can walk to the Gordon Square area, you can walk to since the 1980s, with its green identity being a front-and- restaurants, you can walk to Capitol Theater, there are theaters, center attractor of new high-income and majority white bars, restaurants, retail and people are really attracted to residents. Until recently, some heavily industrial communities the walkability and accessibility. To walk to Edgewater is a such as Bayview-Hunters Point were considered hostile to beautiful thing as well because there are the walking paths new developments and to migration by upper income or over here, so you can actually get to Edgewater Beach.” Detroit white residents. Historically the site of intensive municipal, Shoreway shows that “variable green/white” (Type 1) tracts regional, and military industrial facilities, including the former in Continuous Contraction cities, which have sustained white PG&E power plant; the Southeast Sewage Treatment plant; populations above the Zhang-Logan threshold and experienced the naval shipyard; waste sites; diesel freight transportation; concentrated periods of high-area greening, are key areas and two freeway lines, Bayview-Hunters Point residents have for revitalizing declining economies and population growth, continuously been exposed to toxic smells, waste, and acute but also for strengthening the link between higher income negative health outcomes (Dillon, 2014; Liu and Shatford, 2019). whiter neighborhoods and greening in cities—a process which Local residents and organizers almost universally describe a undermines racial green justice. pattern of municipal neglect of the area as a key factor in exacerbating these conditions. Punctuated Growth Cities However, since becoming a federal Superfund site in 1989 and When considering the data for punctuated growth cities that benefiting from remediation, 500 acres of the neighborhood and recovered from prior declines, the trends associated with high its waterfront have been or are being redeveloped into 13,000 and low growth in other cities can both be seen. In other new homes and more accessible and integrated green and blue words, there are similarities with steady growth and continuous spaces (e.g., Hunters Point Shoreline, Candlestick Point, and contraction cities. The three most representative clusters of state India Basin). Today, the neighborhood includes one of the few sequences shown in Figure 5 demonstrate an historic association natural areas within the city adjacent to the Bay (India Basin between tracts with a high level of greening that sustained white Neighborhood Association, 2019) and it has unique natural and populations and tracts with a low level of greening that sustained recreational features such as a recreational trail on a segment of or shifted toward non-white populations. the Bay Trail, shoreline access for recreation, one of only a few The first cluster of tracts (Type 1, n = 281), labeled “spiky tidal salt marsh wetlands in San Francisco, and restored habitats green/white,” represents the bulk of the areas that greened. for shore birds. These areas retained white population above the Zhang-Logan Yet, everyone we interviewed (regardless of background) threshold of at least 25% of the average across all metros was acutely aware that the area is seen as one of the last throughout the period studied and had spikes of concentrated development frontiers in San Francisco. The southeastern greening in two periods—the late 1980s and the early 2000s. waterfront including Bayview-Hunters Point now sits adjacent Most of the early (1980s) intensive greening in these cities to and is planned for new luxury development and retail occurred in Type 1 tracts, implying that these actions were gentrification through new restaurants, shops, and services. For associated with sustained whiteness in the following decades. example, the India Basin mixed-use project is envisioned as The second cluster of tracts (Type 2, n = 32), labeled “low “a pedestrian-first, human-scaled, bicycle-friendly, residential green/shift non-white,” had the lowest occurrence and showed a village where neighborhood amenities are within a short walk,” shift from white to non-white in the 1980s or early 1990s with including 15.5 acres of open space. While historically populated greening scattered throughout. These tracts represent the tail-end by working-class African American families and other ethnic of a long prior period of “white flight” away from these large minorities (Chitewere et al., 2017), an increasing number of cities. Philadelphia was especially still experiencing this trend white and Asian residents with higher incomes have recently in the 1980s and, thus, has a concentration of Type 2 tracts. purchased homes in the area. As an example of the housing Greening in these tracts was mostly relatively low early-on and boom characterizing the area, recently homes have gone for then non-existent in more recent years. The third cluster of sale for more than $800,000 for a 2-bedroom—well above the tracts (Type 3, n = 170), labeled “low green/non-white,” mostly financial capacity of existing residents. This shift has created a stayed non-white (with a few exceptions) and had greening new sense of acute insecurity, wherein the benefits of the new concentrated around the early 2000s. Type 3 tracts commonly environmental cleanups are almost lost on existing residents had low or moderate levels of greening by area, implying that of color. In the words of one resident, “everyone is trying later and lower levels of greening were not associated with to stabilize housing. If you don’t have a place to stay, you substantial racial or ethnic shifts in these cities (however, see don’t care that the sea level is going to rise, you don’t care qualitative case below for an example of the precarious nature that your air is bad, you don’t care about the quality of of this trend). the environment.” Frontiers in Ecology and Evolution | www.frontiersin.org 11 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness FIGURE 5 | Punctuated growth cities typical trajectories: three-cluster solution for race/ethnicity-greening trajectories of all tracts that had any greening between 1975 and 2014. The black line represents the most common sequence and the gray line represents the second-most common sequence. Together, these lines reflect the modal sequence, or most common trend, within a cluster. Detailed cluster results can be found in the Supplementary Material. The sense that demographic shifts generate a more acute more likely to be white residents—high profile green projects like threat than environmental conditions is sensible given recent those in Bayview-Hunters Point and their associated branding census data. While in 1980, 72% of residents self-identified as seem to accompany shifts toward a whiter population. African American, today only 27% identify as black (of any sort). This decline is summarized by the San Francisco Bayview Trends Across All Cities Newspaper as “the smallest black population and the fastest While each city growth trajectory has a unique register of exodus of any big city in America” (Epps, 2020). In contrast, association between greening and the presence of white according to the 2009–2013 American Community Survey, 19.5% population over time, the recurrence of similar cluster of all neighborhood residents are now white. This has led to characteristics across different trajectories shows that there a deep suspicion among Bayview-Hunters Point residents of are some overall commonalities. Steady growth cities have the the motivations behind greening efforts, and they have started strongest and most uniform tendency toward a trend expressed to resist greening. As one citywide greening group said, “We across all cities: clusters of tracts that greened and sustained have encountered some real resistance to [greening] in this or shifted toward white populations were the only ones that part of town. . .If we look at a map of where we have been were defined by high-area, concentrated greening. We do also active over the last four years, you know, this is the one area see a cluster of low levels of greening in white tracts in steady where we have been unsuccessful. . .and this is primarily about growth cities, but this cluster is mostly a consequence of the fact social justice concerns.” Bayview-Hunters Point demonstrates that these cities have a high percentage of white population so that, while many Type 3 tracts have not yet seen a substantial nearly all greening occurred in tracts where white population was shift toward whiter populations, the people living in these tracts present. Thus, Figure 6 shows high and low greening occurred are quite aware of the trends seen in Type 1 (spiky green/white) in tracts that sustained or shifted toward a white population for areas of their city. People of color living in recently greened steady growth cities, but not for any other type of city. The key areas expressed feeling highly vulnerable to displacement in trend for high area greening, though, is that it occurred mostly in punctuated growth cities, even if those are still contaminated white tracts across all three types of cities. and industrial areas. Our quantitative data analysis supports their Meanwhile, continuous contraction cities have the strongest suspicion that, as these areas become extensively greened, those and most uniform tendency toward another common trend: who will benefit from new restored and cleaned-up grounds are all clusters of tracts that sustained or shifted toward non-white Frontiers in Ecology and Evolution | www.frontiersin.org 12 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness guaranteed by the government (Squires and Woodruff, 2019). Predictably, redlined neighborhoods in cities such as many of those in our sample—Cleveland or Baltimore in particular— became severely disinvested and lost population during a roughly 50-year period of white flight to suburban areas. This mass exodus left many underinvested urban neighborhoods in those cities with under-maintained green spaces. By the mid-1980s, though, two processes were taking shape that would alter the fate of many redlined neighborhoods and that shapes the findings we report. First, banks developed lucrative and predatory high-interest loan products targeted toward high- risk areas and began making predatory loans widely accessible. Second, a renewed interest on the part of middle-class white Americans and an emerging awareness of “rent gaps” made these neighborhoods a target for gentrification that caused widespread residential displacement of non-white residents. As a result, many neighborhoods that were redlined in the 1930s started to be gentrified by the 1980s (Wyly and Hammel, 1999), and our results for Austin, San Francisco, New York, Philadelphia, Seattle, and Portland especially demonstrate this trend. For example, 87% of San Francisco’s gentrified neighborhoods in 2018 were also redlined in prior periods (Urban Displacement FIGURE 6 | Greening trends in white and non-white tracts. and Project, 2020). This effect was common in cities throughout the country, especially those that experienced a post-industrial populations were defined by low-area greening. Overall, economic boom around the turn of the twentieth century combined trends show clusters with an association between (Wyly and Hammel, 2004). high-area greening occurring in tracts with notable white Meanwhile, by the 1990s, our analysis illustrates the outcomes populations and low-area greening concentrated in non-white of a process by which the selection of which formerly redlined areas (with the exception of steady growth cities where there neighborhoods would be targeted by gentrification investments was little to no greening in tracts that did not, at some point, was increasingly driven by actions undertaken within urban have white representation). Figure 6 visualizes this overall sustainability and greening circles. By the early 2000s, this link commonality across all growth trajectories. between gentrification of formerly disinvested neighborhoods and the global push toward greener cities came to be seen by some as a consolidated process of urban eco, green, or environmental DISCUSSION gentrification (Dooling, 2009; Checker, 2011; Anguelovski et al., 2019), an argument that has been especially applied to cities such In order to serve as an effective historical diagnostic for as New York, Seattle or Portland in our sample. As parks, gardens, urban greening, our findings need to be filtered through an and greenways started to be seen as investment strategies that understanding of the underlying processes of urban land use would prepare areas for increased property value due to greater planning that caused them (Heynen et al., 2006; Safransky, 2014; attractiveness (Immergluck and Balan, 2018), greening in those Anguelovski, 2016; Schell et al., 2020). In general, the associations cities became linked with the racialized history of gentrification. we find make sense when considering the intertwined legacies of Urban greening proponents could no longer claim to be innocent housing segregation, gentrification, and urban greening (Grove bystanders as cities became more unequal and elite (Connolly, et al., 2018; Roman et al., 2018). Our findings reflect an 2018; McClintock, 2018; Garcia-Lamarca et al., 2019) and thus underlying process wherein racial difference has been used to centers of green privilege for upper-class white residents. generate monetary value through home ownership (Gotham, In booming housing markets such as New York and 2000; Fluri et al., 2020) and been demarcated by policy (Fields, San Francisco, our results show that especially the push toward 2018). Urban greening is wrapped up in this underlying process. presumably more visible concentrated greening was associated In the US, housing finance has facilitated benefits to flow with sustained whiteness or a shift from non-white to white— to white homes more readily than non-white (Aaronson et al., a consequence in these cities of gentrification processes. As 2017; Rothstein, 2017), with an extreme example being what one organizer we interviewed in Austin put it, the interests is referred to as “redlining.” From the 1930s through several shifted toward a “different shade of green”—referring at once to decades past World War II, the federal agency responsible for the green of American money and the continued push toward guaranteeing home loans formally reinforced—by drawing red a whiter cultural norm that came with increased monetary lines on maps—the widely held racist notion that non-white value. Safransky (2014) described this as a process of greening neighborhoods were “high risk,” thus effectively excluding becoming a tool for the settler colonial approach to urban space. non-white areas from access to low-cost mortgage lending Going back to Marcuse (1997), racialized residents in the new Frontiers in Ecology and Evolution | www.frontiersin.org 13 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness green “citadels” of fast growing post-Fordist cities have been the detailed historic patterns. We can start recognizing how the increasingly excluded and often have had to move to “outcast history of greening—a seemingly apolitical “urban project” sold [gray] ghettos.” as a win-win, universal benefit with ample health, economic, and environmental qualities—has deep links to the exclusionary and racist roots of urbanization that manifest at different points in From Historical Diagnostic to time and space across and within cities. Social-Ecological Justice What are alternatives to the racialized green histories we see in these nine American cities? One response to the racial CONCLUSION blinders worn by earlier urban greening agendas was articulated by Heynen (2016) as “abolition ecology.” In this alternative, Our goal in this paper was to analyze the co-evolution of greening outcomes of recent decades of efforts to re-nature cities point and whiteness in nine American cities with different growth toward a need to rethink the study of urban ecology by examining trajectories. We propose here a novel quantitative, spatial, and urban natures first and foremost through the lens of antiracist, qualitative mixed methods analysis of trends and advance existing postcolonial and indigenous theory. This approach points toward research at the intersection of urban ecology, green justice, and what some scholars have started to call “green reparations” racialization of urban space. Specifically, we sought to answer (Draus et al., 2019). The green reparations approach explicitly three questions. First, we asked: What are the common sequences links urban greening initiatives with efforts to atone for historical across all nine cities with regard for changes in green space violence committed against racial or ethnic groups in a given city. and changes in the racial makeup of surrounding areas? We For example, in a recent study comparing the use of greening find that the combined trend within the state sequence analysis in Berlin (Germany) and Detroit (United States), the authors shown in Table 2 and Figure 6 demonstrates a basic underlying found examples of green spaces that served as curative bridges relationship wherein higher area greening is more associated with with explicit recognition of historical racial violence embedded tracts that either sustained or shifted toward whiter populations in them. Another case is the planned 11th street bridge park in in American cities. There are, though important variations across Washington DC, which, despite staying shy of returning land time and type of city. to displaced black residents, aims at allowing local control over Second, we asked: Does this pattern vary according to the development in the neighborhood through a linked land trust historical growth trajectory of the city? While there is an (Avni and Fischler, 2019). underlying narrative of greening and whiteness, our hypothesis Abolition ecology and green reparations represent ways of that there are three sub-histories to this narrative was affirmed. thinking and acting from an urban ecology perspective that The tight interlock between race/ethnicity and socio-economic allows the historical diagnostic developed here to be mobilized opportunities in American society appears to be visible within for what Steil (2018; see also Steil and Delgado, 2019) refers to urban greening outcomes and growth seems to shape that as “anti-subordination planning.” Anti-subordination planning relationship to a substantial degree. Each of the three growth calls for the evaluation of all interventions in urban space trajectory histories point toward different relationships between from the perspective of effect, rather than intent. If the effect greening and whiteness. For the most rapidly growing cities— of an action is to perpetuate the subordinate position of the steady growth cities—greening was either closely associated a disadvantaged group, then it violates the tenets of anti- with white areas (Type 1 and Type 2 clusters) or linked to subordination planning, regardless of intent. It would be hard a process of shifting areas from non-white to white (Type 3 to argue against the assertion that urban ecology and urban cluster). For the cities that lost population continuously and saw environmental planning are at times focused only on process a declining GDP—the continuous contraction cities—there was a and intent, while abolition ecology, green reparations, and anti- trend for high-area greening to be associated with whiter areas subordination planning seek to force an accounting for effect, (Type 1 cluster) and low-area greening to be associated with which is also the primary purpose of the historical diagnostic non-white areas (Type 2 and 3 clusters). For the largest cities presented in this paper. that saw a resurgence during the study period—the punctuated While it is not easy to conceive of exactly what just growth cities—intensive greening in the 1980s and early 1990s greening looks like when it embodies abolition ecology principles was associated with a sustained white population (Type 1 cluster), by enacting green reparations through an anti-subordination while low or moderate levels of greening in the 2000s was planning mode that respects the lived experience of marginalized more associated with sustained non-white populations (Type 2 populations (Anguelovski et al., 2020), history provides a place to and 3 clusters). start. When we know the pattern of relations between greening Finally, we asked: Can a generalized relationship be and whiteness and the social-ecological outcomes of a city, we can determined with regard to the relationship between greening and ask how and where that pattern needs to be changed. We can use racial change in these cities? To answer this question, we look the patterns alongside nuanced qualitative understanding of the at the subset of clusters where racial change occurred (steady context to develop strategies for creating a new history that does growth Type 3; continuous contraction Type 2; punctuated not passively accept the social effect of greening as an unmovable growth Type 2). The trend of co-occurring greening and a shift historical artifact. Rather, an historical diagnostic brings into from non-white to white was limited for tracts that greened to relief the contours of effect that prior actions had by outlining steady growth cities (Type 3), and this trend was characterized Frontiers in Ecology and Evolution | www.frontiersin.org 14 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness by high levels of greening. This finding does not mean that such analysis, and primarily drafted the text. IA co-developed the a process cannot be found in other cities, but rather that it was qualitative analysis, contributed to writing sections of the only a common trend in cities with intense growth. Meanwhile, text, and edited the full text. Both authors contributed to the trend of co-occurring greening and a shift from white to the article and approved the submitted version. The authors non-white only occurred in continuous contraction (Type 2) would also like to acknowledge that some of the greenspace and punctuated growth (Type 2) cities. In both cases, this trend and qualitative data used in this paper are based on datasets involved zero early greening and low levels of recent greening, developed by several members of the Barcelona Lab for Urban demonstrating the recent push toward small-scale green projects, Environmental Justice and Sustainability. In particular, we often led by residents or community activists, in these areas. would like to acknowledge that the qualitative analysis of Overall, the answer to this final question—and the general Cleveland was based on the draft of a book chapter co- takeaway of this research—is that greening, growth, and race are authored by Margarita Triguero-Mas and Wendy A. Kellogg, tightly interlocked qualities of unequal urban space in American which will be published in full form soon after this article cities. Only in considering this interlock through an integrated in the forthcoming book published by Routledge and edited historical diagnostic can urban ecological interventions hope to by Isabelle Anguelovski and James J. T. Connolly, The work toward racial justice. Green City and Social Injustice: 21 tales from North America and Europe. DATA AVAILABILITY STATEMENT The raw data supporting the conclusions of this article will be FUNDING made available by the authors, without undue reservation. This research was supported by the EU H2020 ERC project ETHICS STATEMENT GreenLULUs (GA678034) and faculty support from the University of British Columbia and the “María de Maeztu” The studies involving human participants were reviewed and Programme for Units of Excellence of the Spanish Ministry of approved by the Universitat Autonoma de Barcelona. The Science and Innovation, CEX2019-000940-M. patients/participants provided their written informed consent to participate in this study. SUPPLEMENTARY MATERIAL AUTHOR CONTRIBUTIONS The Supplementary Material for this article can be found JC primarily conceived the manuscript, completed all online at: https://www.frontiersin.org/articles/10.3389/fevo.2021. quantitative and spatial analysis, co-developed the qualitative 621783/full#supplementary-material REFERENCES Boone, C. G., Buckley, G. L., Grove, J. M., and Sister, C. (2009). Parks and people: an environmental justice inquiry in Baltimore, Maryland. Ann. Assoc. Am. Aaronson, D., Hartley, D., and Mazumder, B. (2017). “The effects of the 1930s Geograph. 99, 767–787. doi: 10.1080/00045600903102949 HOLC“ redlining” maps (No. 2017-12),” in Working Paper Series WP-2017-12, Borgna, C., and Struffolino, E. (2018). “Unpacking configurational dynamics: (Chicago, IL: Federal Reserve Bank of Chicago). sequence analysis and qualitative comparative analysis as a mixed-method Anguelovski, I. (2016). From toxic sites to parks as (green) LULUs? New challenges design,” in Sequence Analysis and Related Approaches: Innovative Methods and of inequity, privilege, gentrification, and exclusion for urban environmental Applications, eds G. Ritschard and M. Studer (Basingstoke: Springer Nature), justice. J. Plann. Literat. 31, 23–36. doi: 10.1177/0885412215610491 298. doi: 10.1007/978-3-319-95420-2_10 Anguelovski, I., Brand, A. L., Connolly, J. J., Corbera, E., Kotsila, P., Steil, J., Checker, M. (2011). Wiped out by the “greenwave”: environmental gentrification et al. (2020). Emancipatory, antisubordination, intersectional, and relational and the paradoxical politics of urban sustainability. City Soc. 23, 210–229. approach. Ann. Am. Assoc. Geograph. 110, 1743–1769. doi: 10.1080/24694452. doi: 10.1111/j.1548-744x.2011.01063.x 2020.1740579 Chitewere, T., Shim, J. K., et al. (2017). How neighborhoods influence health: Anguelovski, I., Connolly, J., and Brand, A. L. (2018). From landscapes lessons to be learned from the application of political ecology. Health Place 45, of utopia to the margins of the green urban life: for whom is the 117–123. doi: 10.1016/j.healthplace.2017.03.009 new green city? City 22, 417–436. doi: 10.1080/13604813.2018.14 City of Philadelphia (2016). Greenworks: A vision for Sustainable 73126 Philadelphia. City of Philadelphia Office of Sustainability. Available Anguelovski, I., Connolly, J. J., Garcia-Lamarca, M., Cole, H., and Pearsall, H. online at: https://www.phila.gov/media/20161101174249/2016- (2019). New scholarly pathways on green gentrification: what does the urban Greenworks-Vision_Office-of-Sustainability.pdf (accessed October 2, ‘green turn’mean and where is it going? Prog. Human Geography 43, 1064–1086. 2020). doi: 10.1177/0309132518803799 Connolly, J. J. (2018). From systems thinking to systemic action: social vulnerability Avni, N., and Fischler, R. (2019). Social and Environmental Justice in Waterfront and the institutional challenge of urban resilience. City Community 17, 8–11. Redevelopment. The Anacostia River, Washington, DC: Urban Affairs Review. doi: 10.1111/cico.12282 Bird Rose, D. (2004). Reports from a Wild Country: Ethics for Decolonisation. Connolly, J. J. (2019). From jacobs to the just city: a foundation for challenging the Sydney NSW: UNSW Press. green planning orthodoxy. Cities 91, 64–70. doi: 10.1016/j.cities.2018.05.011 Frontiers in Ecology and Evolution | www.frontiersin.org 15 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness Curnow, J., and Helferty, A. (2018). Contradictions of solidarity: whiteness, settler Kolko. (2016). 2015 US Population Winners: The Suburbs and the Sunbelt. coloniality, and the mainstream environmental movement. Environ. Soc. 9, Ambalangoda: Citylab. 145–163. doi: 10.3167/ares.2018.090110 Langemeyer, J., and Connolly, J. J. (2020). Weaving notions of justice into urban Dillon, L. (2014). Race, waste, and space: brownfield redevelopment and ecosystem services research and practice. Environ. Sci. Policy 109, 1–14. doi: environmental justice at the hunters point shipyard: waste, race and space. 10.1016/j.envsci.2020.03.021 Antipode 46, 1205–1221. doi: 10.1111/anti.12009 Lesnard, L. (2009). Cost Setting in Optimal Matching to Uncover Contemporaneous Dooling, S. (2009). Ecological gentrification: a research agenda exploring justice Socio-Temporal Patterns. Paris, OSC: Sciences Po/CNRS. doi: 10.1177/ in the city. Int. J. Urban Reg. Res. 33, 621–639. doi: 10.1111/j.1468-2427.2009. 0049124110362526 00860.x Liu, X., and Shatford, S. (2019). “Affordable housing and residential segregation Draus, P., Haase, D., Napieralski, J., Roddy, J., and Qureshi, S. (2019). Wounds, in san francisco: a spatial analysis between 2000 and 2015,”in Working Paper ghosts and gardens: historical trauma and green reparations in Berlin and 2019-2 San Francisco, CA: Applied Housing research initiative. Detroit. Cities 93, 153–163. doi: 10.1016/j.cities.2019.05.002 Logan, J. R., Xu, Z. W., and Stults, B. J. (2014). Interpolating US decennial census Epps, K. (2020). San Francisco Don’t Like Black People. SF Bayview. February 11. tract data from as early as 1970 to 2010: a longitudinal tract database. Prof. Available online at: https://sfbayview.com/2020/02/san-francisco-dont-like- Geographer 66, 412–420. doi: 10.1080/00330124.2014.905156 black-people/. (accessed October 12, 2020). Logan, J. R., and Zhang, C. (2010). Global neighborhoods: new pathways to Ernstson, H. (2013). The social production of ecosystem services: a framework diversity and separation. Am. J. Soc. 115, 1069–1109. doi: 10.1086/649498 for studying environmental justice and ecological complexity in urbanized Long, J. (2016). Constructing the narrative of the sustainability fix: Sustainability, landscapes. Landscape Urban Plann. 109, 7–17. doi: 10.1016/j.landurbplan. social justice and representation in Austin, TX. Urban Studies 53, 149–172. 2012.10.005 doi: 10.1177/0042098014560501 Fields, D. (2018). Constructing a new asset class: property-led financial Maciag, M. (2015). Gentrification in America Report. Governing. Available online accumulation after the crisis. Econ. Geography 94, 118–140. doi: 10.1080/ at: http://www.governing.com/gov-data/gentrification-in-cities-governing- 00130095.2017.1397492 report.html (accessed January 9, 2018). Florida, R. (2004). The Rise of the Creative Class and how it’s Transforming Manyika, J., Remes, J., Dobbs, R., Orellana, J., and Schaer, F. (2012). Urban Work, Leisure, Community and Everyday Life (Paperback Ed.). New York, NY: America: US Cities in the Global Economy. Washington, DC: McKinsey Global Basic Books. Institute. Fluri, J. L., Hickcox, A., Frydenlund, S., and Zackary, R. (2020). Accessing Racial Marcuse, P. (1997). The enclave, the citadel, and the ghetto: what has changed Privilege Through Property: Geographies of Racial Capitalism. Amsterdam: in the post-Fordist US city. Urban Affairs Rev. 33, 228–264. doi: 10.1177/ Elsevier. doi: 10.1016/j.geoforum.2020.06.013 107808749703300206 Garcia-Lamarca, M., Anguelovski, I., Cole, H., Connolly, J. J., Argüelles, L., Baró, McClintock, N. (2018). Cultivating (a) sustainability capital: urban agriculture, F., et al. (2019). Urban green boosterism and city affordability: for whom is the ecogentrification, and the uneven valorization of social reproduction. ‘branded’green city? Urban Stud. 58:0042098019885330. Ann. Am. Assoc. Geograph. 108, 579–590. doi: 10.1080/24694452.2017.13 Gibson, C., and Jung, K. (2002). Historical Census Statistics on Population Totals by 65582 Race, 1790 to 1990, and by Hispanic Origin, 1790 to 1990, for the United States, Park, Y., and LaFrombois, M. E. H. (2019). Planning for growth in depopulating Regions, Divisions, and States. Washington, DC: US Census Bureau. cities: an analysis of population projections and population change in Glaeser, E. L., and Saiz, A. (2003). The Rise of the Skilled City (No. w10191). depopulating and populating US cities. Cities 90, 237–248. doi: 10.1016/j.cities. Cambridge, MA: National Bureau of Economic Research. 2019.02.016 Goodling, E., Green, J., and McClintock, N. (2015). Uneven development of the Peck, J. (2005). Struggling with the creative class. Int. J. Urban Reg. Res. 29, sustainable city: shifting capital in Portland. Oregon. Urban Geography 36, 740–770. doi: 10.1111/j.1468-2427.2005.00620.x 504–527. doi: 10.1080/02723638.2015.1010791 Pulido, L. (2000). Rethinking environmental racism: white privilege and urban Gotham, K. F. (2000). Racialization and the state: the housing act of 1934 and development in Southern California. Ann. Assoc. Am. Geograph. 90, 12–40. the creation of the federal housing administration. Soc. Perspect. 43, 291–317. doi: 10.1111/0004-5608.00182 doi: 10.2307/1389798 Pulido, L. (2017). Geographies of race and ethnicity II: environmental racism, racial Gould, K. A., and Lewis, T. L. (2016). Green Gentrification: Urban Sustainability capitalism and state-sanctioned violence. Prog. Hum. Geography 41, 524–533. and the Struggle for Environmental Justice. Milton Park: Routledge. doi: 10.1177/0309132516646495 Gross, J. (2018). Austin Might be the ‘Most Difficult’ Real Estate Market in Texas.: Ritschard, G., and Studer, M. (2018). “Sequence analysis: where are we, where are https://www.austin360.com/business/real-estate/austin-might-the-most- we going?,” in Sequence Analysis and Related Approaches: Innovative Methods difficult-real-estate-market-texas/to6G2Ku5d7PKHfHbBYuMwI/ (accessed and Applications, eds G. Ritschard and M. Studer (Basingstoke: Springer August 1, 2019). Nature). doi: 10.1007/978-3-319-95420-2 Grove, M., Ogden, L., Pickett, S., Boone, C., Buckley, G., Locke, D. H., et al. (2018). Roman, L. A., Pearsall, H., Eisenman, T. S., Conway, T. M., Fahey, R., Landry, The legacy effect: understanding how segregation and environmental injustice S., et al. (2018). Human and biophysical legacies shape contemporary urban unfold over time in Baltimore. Ann. Am. Assoc. Geograph. 108, 524–537. doi: forests: a literature synthesis. Urban Forestry Urban Green. 31, 157–168. doi: 10.1080/24694452.2017.1365585 10.1016/j.ufug.2018.03.004 Heynen, N. (2016). Urban political ecology II: The abolitionist century. Prog. Hum. Rossignon, F., Studer, M., Gauthier, J., and Le-Goff, J. (2018). “Sequence History Geogr. 40, 839–845. doi: 10.1177/0309132515617394 Analysis (SHA): estimating the effect of past trajectories on an upcoming Heynen, N., Perkins, H. A., and Roy, P. (2006). The political ecology of uneven event,” in Sequence Analysis and Related Approaches: Innovative Methods and urban green space: the impact of political economy on race and ethnicity in Applications, eds G. Ritschard and M. Studer (Basingstoke: Springer Nature). producing environmental inequality in Milwaukee. Urban Affairs Rev. 42, 3–25. doi: 10.1007/978-3-319-95420-2_6 doi: 10.1177/1078087406290729 Rothstein, R. (2017). The Color of Law: a Forgotten History of how Our Government Hingorani, A. (2019). LA County Park Equity Groups Fight to Turn Park-Poor Segregated America. New York, NY: Liveright Publishing. Communities ‘Red to Green’. Advancement Project Website. Available online at: Safransky, S. (2014). Greening the urban frontier: race, property, and resettlement https://www.advancementprojectca.org/blog/la-county-park-equity-groups- in Detroit. Geoforum 56, 237–248. doi: 10.1016/j.geoforum.2014.06.003 fight-to-turn-park-poor-communities-red-to-green (accessed October 18, Safransky, S. (2018). Land justice as a historical diagnostic: thinking with detroit. 2020). Ann. Am. Assoc. Geograph. 108, 499–512. doi: 10.1080/24694452.2017.138 Immergluck, D., and Balan, T. (2018). Sustainable for whom? Green urban 5380 development, environmental gentrification, and the Atlanta Beltline. Urban Samanta, A., and Kellogg, W. (2019). “Back to the beginning: the resurgence of the Geography 39, 546–562. doi: 10.1080/02723638.2017.1360041 cuyahoga river and cleveland,” in Legacy Cities: Continuity and Change amid India Neighborhood Association (2019). India Basin History. San Francisco, CA: Decline and Revival, eds S. Webster and R. Tighe (Pittsburgh, PA: University of India Basin Neighborhood Assoc. Pittsburgh Press). Frontiers in Ecology and Evolution | www.frontiersin.org 16 March 2021 | Volume 9 | Article 621783 Connolly and Anguelovski Histories of Greening and Whiteness Schell, C. J., Dyson, K., Fuentes, T. L., Des Roches, S., Harris, N. C., Miller, Walsh, E. A. (2018). White fragility as an obstacle to anti-racist resilience D. S., et al. (2020). The ecological and evolutionary consequences of systemic planning: opportunities for equity-conscious partnerships. J. Urban Manag. 7, racism in urban environments. Science 369:eaay4497. doi: 10.1126/science. 181–189. aay4497 Wyly, E. K., and Hammel, D. J. (1999). Islands of decay in seas of renewal: housing Smilor, R. W., Gibson, D. V., and Kozmetsky, G. (1989). Creating the Technopolis: policy and the resurgence of gentrification. Hous. Policy Debate 10, 711–771. High-technology Development in Austin, Texas. Austin, TX: IC2 Institute doi: 10.1080/10511482.1999.9521348 Articles. doi: 10.1016/0883-9026(89)90033-5 Wyly, E. K., and Hammel, D. J. (2004). Gentrification, segregation, and Smith, N. (1996). The New Urban Frontier: Gentrification and the Revanchist City. discrimination in the American urban system. Environ. Plann. 36, 1215–1241. Palmeira Mansion: Psychology Press. doi: 10.1068/a3610 Squires, G. D., and Woodruff, F. (2019). Redlining. Hoboken, NJ: Wiley Online Zhang, W., and Logan, J. R. (2016). Global neighborhoods: beyond the multiethnic Library, 1–8. doi: 10.1002/9781118568446.eurs0260 metropolis. Demography 53, 1933–1953. doi: 10.1007/s13524-016-0516-4 Steil, J. P., and Delgado, L. H. (2019). Limits of diversity: jane jacobs, the just city, and anti-subordination. Cities 91, 39–48. doi: 10.1016/j.cities.2018.06.017 Conflict of Interest: The authors declare that the research was conducted in the Steil, J. (2018). Antisubordination planning. J. Plann. Educ. Res. doi: 10.1177/ absence of any commercial or financial relationships that could be construed as a 0739456X18815739 [Epub ahead of print]. potential conflict of interest. Urban Displacement and Project (2020). Online Resource. Available online at: https://www.urbandisplacement.org/redlining (accessed April 15, 2020). Copyright © 2021 Connolly and Anguelovski. This is an open-access article US Census Bureau (2019). Detroit, Michigan Quickfacts. Available online distributed under the terms of the Creative Commons Attribution License (CC BY). at: https://www.census.gov/quickfacts/fact/table/detroitcitymichigan, The use, distribution or reproduction in other forums is permitted, provided the MI/PST045219 (accessed October 13, 2020). original author(s) and the copyright owner(s) are credited and that the original Walks, R. A. (2001). The social ecology of the post-Fordist/global city? economic publication in this journal is cited, in accordance with accepted academic practice. restructuring and socio-spatial polarisation in the Toronto urban region. Urban No use, distribution or reproduction is permitted which does not comply with Stud. 38, 407–447. doi: 10.1080/00420980120027438 these terms. Frontiers in Ecology and Evolution | www.frontiersin.org 17 March 2021 | Volume 9 | Article 621783