Equity in Urban Forest Management:
An assessment of street tree condition, maintenance, and neighborhood income levels in
Seattle, Washington

by:

Ara Kaufer Erickson

A thesis submitted in partial fulfillment of the requirements for the degree of

Master of Science

University of Washington

2004

Program Authorized to Offer Degree: College of Forest Resources

Table of Contents

Click here for a PDF copy

List of Figures

List of Tables

Acknowledgements

I wish to express my deepest gratitude to all of the people and groups who made this project possible.

To Witney, Lauren, Kimiko, Catherine, and Mackenzie, who made the entire project so meaningful. Your enthusiasm and positive responses to this endeavor, made me so grateful that we were able to spend the summer together and proud of what you added to this project.

To my committee chair, Gordon Bradley, for your immense support, compassion, and guidance throughout the entirety of my graduate school adventure. Your encouragement made me continue when I was ready to surrender all hope. To Rowan Rowntree, for guiding me towards the field of urban forestry and providing me with an endless flow of compassionate- and wisdom-filled words. You are the mentor I have always dreamed of having. To Kathy Wolf, for teaching me the real workings of urban forestry and for sharing your amazing insight and expertise. I admire your dedication to this field and am grateful for your excellent edits, counsel, and suggestions throughout this project.

To the Community Forestry Research Fellowship program, for providing financial and intellectual support. Your program is one of the most worthwhile programs I have encountered–and I feel extremely fortunate to be part of it.

To my parents, Ben and Gilda Wheeler, for enduring my long study weekends and providing excellent edits, advice, and support (as well as a great laundry facility). You truly carried me through this process.

To my friends and classmates, who never failed to lift my spirits and provide an amazing support system. I am in awe of your motivation, passion, and outstanding ability in all that you undertake.

Thank you.

Introduction

It is widely accepted that benefits of urban forests include energy savings, reduced stormwater runoff, clean air, reduced levels of violence, increased levels of community involvement and interaction, higher property values, aesthetic values and more. These benefits greatly enhance a neighborhood’s social, ecological, and economic well-being. Because of the myriad of benefits provided by urban forests, it can be argued that the urban forest should be considered part of the basic infrastructure of urban cities and towns where people reside.
It is vital that the benefits of urban forests be equally distributed across communities, not just to select groups of individuals who have the financial, intellectual, and political resources to acquire increased levels of urban forest care and attention. Unfortunately, for an assortment of reasons, many cities appear to have varying degrees of street tree condition in different neighborhoods. Tree maintenance from homeowners, pollution levels, soil type, age and species of trees, and damage from automobiles are just a few explanations. One additional reason may be the socioeconomic level of a neighborhood; there is an unspoken assumption that the urban trees may be indirectly (or directly) affected by a neighborhood’s economic status.
This study explored the relationships between neighborhood income levels, the condition of the neighborhood street trees, and the level of tree maintenance provided by the City of Seattle, Washington. A sample of tree condition, census data, and street tree maintenance records were used to investigate the above relationships. In addition, this study incorporated participatory research methods into the majority of the steps by including high-school students in the data preparation and data collection stages.
Chapter One begins with a brief background on urban forest benefits and the important roles trees play in the quality and health of life in a city—and the importance of equal access to these benefits. Additional information is presented regarding the documented need for good urban forest management and problems associated with the delivery of tree maintenance across an entire city. The background concludes with a review of studies on socioeconomic factors with regard to urban trees. The final sections of the first chapter are dedicated to research objectives, presentation of the hypotheses, and a description of the study site.
Chapter Two describes the methods used in the study: data sources, data manipulation, and the subsequent analysis. Chapter Three presents results from the tree condition assessment and tree maintenance records in relation to median household income. Discussion of the results and the larger context of the findings are presented in Chapter Four. The thesis concludes with a brief look at the limitations of this study, implications of the findings, and possibilities for future research. The community participation aspects of the project are presented throughout the paper.

Chapter 1: Background and Literature

URBAN FORESTS AS PART OF THE URBAN INFRASTRUCTURE
Urban forests are known to provide many benefits – ecologically and socially – to cities and their residents. It is gradually being accepted that urban forests are an integral part of the basic infrastructure of urban cities and urbanizing areas. Just as sewers direct wastewater, streetlights provide safety, and road systems provide safe, reliable transportation, people have begun to recognize the wealth of benefits provided by urban forests. Although these benefits often commingle, they can be described in three categories: ecological, social, and economic.

Ecological Benefits of Urban Forests
Reduced levels of storm water runoff
In cities with large amounts of rain, or infrequent yet heavy rainstorms, storm water management is costly and extremely important. With an ever-increasing amount of impervious surfaces in our expanding cities, storm water hits the ground and rapidly finds its way into storm drains and holding tanks, rather than seeping into the ground, resulting in large amounts of pollutants entering our waterways. Urban trees afford an opportunity for cities to significantly reduce the costs of storm water management and act as nature’s water storage system. In Dayton, Ohio, Sanders calculated that the city tree canopy lowered potential water runoff by approximately 7% for a 6-h, 1-year storm event (Sanders 1986).

Removal of air pollutants
Air quality in cities continues to be extremely important as industry, vehicles, and other sources of pollution continue to emit harmful pollutants into the atmosphere. It is widely recognized that trees intercept some airborne pollutants from the air and remove some gaseous pollutants via stomata uptake. Depending on their size and species, urban trees can remove large amounts of pollutants from the air. From a computer simulation, it was estimated that trees in New York, Philadelphia, Baltimore, and Boston removed, respectively, 1821 metric tons, 1031 metric tons, 499 metric tons, and 278 metric tons from the air in 1994 (Nowak et al. 1998). In an earlier study, Chicago’s trees removed an estimated 5575 metric tons of air pollutants (Nowak 1994).

Carbon storage
While the debate surrounding global climate change still wages, there is an accepted need to mitigate the large amounts of carbon being emitted into the atmosphere. It was estimated that the national urban tree cover plays a critical role in reducing the effects of carbon emissions as it stores between 350 and 750 million metric tons of carbon and approximately 27 tons/acre – a substantial carbon sink (Rowntree and Nowak 1991, Nowak 1993, McPherson et al. 1997).

Social Benefits of Urban Forests
Community ties
In a study conducted at a large public housing site in Chicago, researchers found that increased levels of vegetation and presences of trees resulted in stronger neighborhood social ties. People living near the green spaces had increased levels of interactions with neighbors, stronger senses of belonging, and were more apt to spend time outdoors than people living adjacent to barren spaces (Kuo et al. 1998). Urban forests also provide an opportunity for community interactions through tree-planting programs. Dwyer and others wrote, “Active involvement in tree-planting programs has been shown to enhance a community’s sense of social identity, self-esteem, and territoriality…(Dwyer et al. 1992).”

Reduced levels of violence
Kuo and Sullivan also found that individuals living near the greener common spaces felt safer in their homes and buildings than the residents living near relatively barren spaces (Kuo et al. 1998). In another study, the researchers found that as greenness of the landscape increased the number of police reported crimes per building decreased (Kuo and Sullivan 2001).

Mental health and well-being
The body of literature connecting personal well being and health with urban nature is extraordinary and strong. Kaplan has shown that job satisfaction and a person’s well-being can be drastically improved merely by nature viewed through a window (Kaplan 1993). Faster recover rates in hospital patients and reduced levels of stress in college students have also been associated with the presence of urban trees and vegetation (Ulrich 1984, 1986)

Economic Benefits of Urban Forests
Energy Conservation
The proven savings in heating and cooling costs for buildings near urban trees are well documented. McPherson and Rowntree showed that heating and cooling costs for a typical resident were reduced by 8% - 12% with an appropriately located single 25-ft tree (McPherson and Rowntree 1993). Another team of scientist showed that approximately three mature trees per building lot saved an estimated $50 to $90 per dwelling in heating and cooling costs in Chicago, Illinois (McPherson et al. 1997).

Property values and retail
In addition to energy savings, increased levels of property values are an economic benefit provided by urban trees. In a study conducted in Athens, Georgia, single-family homes were found to sell for 3 to 5% more than other homes when trees were located in their listings’ photographs (Anderson and Cordell 1985). Recently, a study showed that consumers are willing to pay, on average, 11.95% more for equivalent good purchased in retail areas with high quality landscapes, including trees, compared to areas with little or no landscaping (Wolf 2003). This furthers the argument that urban trees provide a variety of economic benefits to both residents and commercial owners.

BENEFITS OF STREET TREES
The benefits discussed above are generated by diverse urban forest types – parks, private trees in yards, greenways, vacant lots, and street trees. Since this thesis’s research is focused on street trees, it is important to note some of the science that focuses on street trees. The Center for Urban Forest Research, a unit of the USDA Forest Service’s Pacific Southwest Research Station, is a leading group on quantifying the benefits and costs of urban forests, including both private and public trees. Here is a brief overview of information related solely to street trees’ benefits.
In a study done in Modesto, California, energy savings (heating and cooling costs) from street trees were due to a mix of shade provision and climate effects (15% and 85% respectively). Compared to Residential trees in front and back yards, street trees provided much more savings from shade than from climate effects because of the proximity to the actual homes (McPherson et al. 1999). From a report focusing on urban forests in Western Washington and Oregon, estimates of the value of street trees on property values, rather than front yard trees presented in the Anderson and Cordell’s Georgia study, showed that a typical large street tree increased property values by $0.118 per square foot (McPherson et al. 2002). Additionally, the average net benefits (calculated by adding the annual values of net energy savings, air quality improvements, carbon dioxide reductions, stormwater runoff reductions, and aesthetics and other benefits) of small, medium, and large street/park trees were, respectively, $1, $19, and $48 annually (McPherson et al. 2002).

DISTRIBUTION OF URBAN FOREST SERVICES
Researchers and professionals in the urban forestry field agree that, unfortunately, municipal decision makers and budgets often place urban tree care and maintenance low on the priority list. The following quote is a common theme being expressed in the current literature:

Far from being an amenity, then, it appears that trees play multiple fundamental roles in the continued health of urban communities and should be regarded in the same light as other urban infrastructural elements…the fact remains that few urban politicians view these issues as central to their agendas (Kuo 2003).

Once acknowledged that urban forests are integral to the public infrastructure, managers, in turn, should distribute the benefits of the urban forests equally across the entire city. Unfortunately, urban forestry programs often receive scant funds from municipal budgets or limited attention from the decision makers due to the financial demands of more pressing political and social problems facing cities, such as crime and education (Parker 1995, Tate 2000). This lack of attention results in understaffed tree crews (for routine pruning and necessary tree care) and a scarcity of funding for community involvement and education projects. This is currently the case in Seattle, Washington – a large city proud of its trees, but plagued with a disproportionate number of older trees requiring maintenance and scarce resources (i.e. funding for additional tree crews) to care for both the young and old trees (Rundquist 2003). Nolan Rundquist, the Seattle City Arborist, claims that without additional tree crews the City cannot do much more than respond to emergency clean-ups and hazard trees – a “crisis management” situation. A routine management schedule is needed to manage a green infrastructure program.
If tree programs do not have sufficient resources to begin with, how can they best guarantee that the benefits of trees are equally distributed among all of the residents, regardless of socioeconomic status? In the case of Seattle, the city has made many street trees the responsibility of the adjacent property owner (Seattle Tree Ordinance #90047). This situation could lead to either an increase in attention towards urban trees if more private residents take responsibility of the trees in front of their homes, or a decrease in overall health of the urban forest if the city reduces maintenance and the trees are ignored. Some public policy experts believe that higher-income residents contract to the private sector to fill in when the city no longer provides infrastructure services, leaving the lower-income areas to fend for themselves, with limited resources (Merget 1979).

INPUTS AND OUTPUTS IN URBAN FOREST SERVICE DELIVERY
Continuing with the premise that the urban forest is part of the public infrastructure, a possible way to measure distribution of the public service is to measure the condition of trees and the level of tree maintenance across the entire city. The literature shows that maintained trees are, on average, in better condition than trees that are not maintained. Good maintenance practices include pruning, watering young trees, and protecting trees from construction and other potential damages (Miller and Sylvester 1981, Achinelli et al. 1997, Luley et al. 2002). Miller expands on the need for street tree maintenance in his book Urban Forestry: Planning and Managing Urban Greenspaces: “Of all municipal tree management activities pruning is the most essential for long-term safety and survival(Miller 1997, page 263).” Without proper and adequate levels of maintenance, tree survival can be drastically reduced and tree growth can suffer (Miller 1997). The same body of literature shows that trees in better condition provide greater levels of benefits, especially trees that are older and more established (Abbott et al. 1991, McPherson 1995, Miller 1997). Since maintained trees are most likely in better condition, it can be argued that trees that are maintained result in greater levels of the known benefits of urban forests (Tate 2000).

Inputs and Outputs
In the study of urban-service distribution, there are many ways to compare goods and services. One interesting method is to measure the level of service as the “input” and the quality of the purpose for the service as the “output” (Lineberry 1977, Rundquist 2003). In urban forestry terminology, tree maintenance can be the input and tree condition can be the output (with the goal being a desired level of tree condition). Often, research is focused on the level of input, rather than output; unlike in many service-distribution situations, the urban forestry field can measure the output – tree condition. In order to maintain equal levels of benefits across an entire city, we cannot look solely at the services provided, but also at the benefits realized. In the literature surrounding urban-service distribution, it is well understood that output equality often requires unequal levels of the input (Lineberry 1977, Lucy 1981, Rich 1982).
The literature on distribution of public services is extensive. There appears, however, to be a general acceptance of the difficulty of measuring how services are distributed. In a case in Brooklyn, New York, the residents of a lower-income, predominately Puerto Rican and African-American neighborhood felt that the park in their neighborhood was under-funded compared to other parks in the area. The court ruled in favor of the city, after the city showed they spent proportionately more on the park in question than on other parks. Regrettably, the city was not able to show that their proportionally higher spending actually resulted in equal levels and numbers of park facilities (Merget and Wolff 1976). If the goal of the city was to spend the same amount of money on each park, and not be concerned with the quality of the park, then the ruling correctly dealt with the legal distribution of money spent. However, if the goal was to provide parks of equal quality to all residents, then the ruling may have not dealt with the ethical responsibility to provide equal levels of park services. It is a complex and difficult dilemma: should resources or services be distributed equally or equitably?

Service Delivery and Environmental Justice
One challenge facing many cities is in regard to environmental injustices in urban planning and management. Although the claims of environmental injustice are still debatable (Cutter et al. 2001), there is a general understanding that it is unjust when proposed sites for toxic waste areas, power plants, and other activities harmful to the health and well-being of nearby residents are more likely to be near disadvantaged populations, or when environmental management options are discriminatory in intent. One paper suggests, “the intersection of environmental management and the environmental injustice movement occurs when a governing body fails to manage resources effectively, resulting in all or part of a community unable to attain a minimum environmental equality (Macey and Her 2001).” Keeping this in mind, if, by providing equitable levels of tree maintenance, the benefits of urban trees are not equally distributed across an entire city, then the claim could be made that a minimum environmental equality is not being realized for all residents. Albeit unintentional, this would be a possible environmental injustice.

RELATED STUDIES
There is a small, yet growing, body of literature regarding the levels of tree cover, distribution of forest structure, and participation in the care and management of urban trees amongst different socioeconomic groups. Currently, however, there is little research in the urban forestry field regarding the distribution of the benefits of urban trees across different socioeconomic groups. If trees are in worse condition in lower-income neighborhoods, the benefits of urban forests may not be available to the residents of those neighborhoods.

Making the Connection between Tree Condition and Socioeconomic Variables
Only one piece of research, to date, has been published discussing a direct link between socioeconomic variables and tree condition. A Ph.D. dissertation found that census tracts in Boston experienced a positive relationship between average income and good forest structure (both street and park trees) (Welch 1991). Welch’s research used in-depth factor-analysis to sample two Boston neighborhoods, Roxbury and North Dorchester, in relation to forest structure. She classified forest structure by quality and quantity, meaning that good forest structure included both healthy trees and a certain quantity of trees. She then calculated the average median household income for all areas that had “good forest structure” and found that areas with above average income had higher percentages of good forest structure. The income aspect of Welch’s research was just a small part of a much larger research design, and little interpretation was presented about why this correlation existed.
Grey and Deneke state in their 1986 book, Urban Forestry, “In areas of low resident income, the urban forest is often composed of declining older trees remaining from times of greater prosperity…In [the] areas of middle and upper income the urban forest is generally well planted and well tended, reflecting the options of affluence (Grey and Deneke 1986).” However, this appears to be a continuation of the anecdotal evidence regarding tree condition and socioeconomic factors. Additionally, in a study on newly planted street trees in Berkeley and Oakland, California, it was found that higher tree mortality was associated with lower socioeconomic status (Nowak 1990).

Differences in Urban Forest Structure
In the New Orleans study done in the late 1980’s, Talarchek found that tree cover appeared to be more highly concentrated in neighborhoods with high-income and high-status white residents (Talarchek 1990). A 2000 study on the distribution of forest cover in the Chicago region found that wealthier areas tended to have higher tree cover than poorer areas (Iverson and Cook 2000). The study area, however, included both the highly urbanized city interior as well as the rural suburbs, offering comparisons of the urban poor residents and the wealthy rural residents.
Nowak concluded that the lower-income areas in Oakland, California, had more utilitarian trees (with intrinsic value of producing nuts or fruits) and fewer broadleaf, ornamental, and coniferous trees than the higher-income areas (Nowak 1991). Whitney and Adams (1980) performed a detailed analysis of plant communities in and around Akron, Ohio, and found that different complexes were distinguishable by socioeconomic patterns. Distribution ranged from the inner city maple complex to the conifer complex in the working-class neighborhoods, and finally to the old oak mixed suburban complex further outside of the city.

Residents’ Ability and Involvement in Urban Forestry Issues
Many would argue that higher-income residents are more able to take part in community tree planting efforts and maintenance of right-of-way trees (Grey and Deneke 1978, Johnston 1985, Lorenzo et al. 2000), yet confirmatory evidence is scarce and hard to find. In a recent study, researchers found that the willingness of individuals in Mandeville, Louisiana, to pay higher prices for tree preservation and protection was directly related to income levels. While there was no notable difference in income among the respondents willing to pay between $6 and $12 per year, the percentage of respondents willing to pay more than $12 per year increased from 15.6% in the lowest income category to 37.8% in the highest income category (Lorenzo et al. 2000). Iles (1998) made this observation regarding the demographics of where urban forestry efforts are directed: “Generally speaking, traditional targets for urban and community forestry programs in the United States include financially comfortable, if not affluent, well-educated, predominantly Caucasian members of the community.”

RESEARCH OBJECTIVES
As the literature review shows, there is sparse information regarding the condition of urban trees in relation to neighborhood socioeconomics. There is ample, and highly beneficial, science on urban tree cover and benefits; however, high percentages of tree cover may neither equate to healthy and well-maintained trees, nor allow for comparison of age or species diversity. With the intention of adding to this small body of literature, this study attempted to investigate if an association existed between neighborhood median household income, street tree condition, and levels of tree maintenance in Seattle, Washington.
Another objective of this project was to incorporate participatory research methods, allowing for a greater connection to the community and an increased awareness of urban forests. Made possible by funding from the Community Forestry Research Fellowship Program, a group formed by the Ford Foundation in 1996 to increase awareness of community forestry issues in the United States, high-school students were invited to participate in the project—from beginning to end. This participation allowed students to learn about a new, and often unknown field, while providing valuable insight as young residents of Seattle and future urban tree planters and caretakers. The Community Forestry Research Fellowship Program believes that enhancing research with local knowledge and participation, healthy and sustainable communities can thrive; thus, the high-school students’ involvement in this project benefited both the research and the sustainability of Seattle’s urban forest.
This research problem was sparked by observations while working as a summer intern for the Center for Urban Forest Research – a research unit in the USDA Forest Service’s Pacific Southwest Research Station in Davis, CA. During the data collection phase in San Francisco, a disparity of tree condition was informally observed between high and low income neighborhoods; the trees in Pacific Heights appeared to be much healthier and vigorous than the trees in Bay View/Hunter’s Point. The disparity could not simply be attributed to varied management efforts, as the trees are all maintained by the City of San Francisco. These initial observations inspired the thesis that inherent social inequities may exist in current urban forest management allocations.
This research is valuable in that it adds to the small body of literature surrounding urban forests and socioeconomic information. It is important to continue investigating the relationships that may exist between urban green resources and the residents of those urban areas. As the importance of environmental justice continues to become evident, it is pertinent that all fields address the issues of equal distribution of environmental benefits and the inherent inequalities that may exist in our current management systems, including urban forestry.

HYPOTHESES
The questions posed below, and the diagram in Figure 1, illustrate the hypothesized relationships tested in this study.

1. Is average street tree condition different between low and high-income areas—specifically, is average street tree condition higher in higher-income areas?
2. Is street tree maintenance, performed by the City’s Tree Crew, different between low and high-income areas?
3. Is average street tree condition related to street tree maintenance?
   The null hypotheses were that there was neither a difference in average tree condition between areas with different median household incomes nor a difference in public tree maintenance in the same areas.
   

Figure 1. Relationships between income and street tree condition and maintenance

Chapter 2: Research Methods

This chapter provides an overview of the methodology used in this study. The first section describes the selection of the study area and how data was acquired from outside sources. The second section describes the tree sampling method–using Seattle’s GIS-based street tree inventory–and the tree condition assessment used for data collection. The third section explains the processing of Seattle’s tree crew work records, and the final section presents an overview of the analysis used to examine relationships in the data.

STUDY SITE
The City of Seattle, Washington was chosen as the study site for this project due to the immense interest in the management and continuation of the city’s urban forest resource - as reflected by the development of the transportation department’s urban forestry program, the city’s 16-year Tree City USA status, and the Mayor’s Urban Forest Coalition. Seattle boasts a fairly extensive urban forest presence, with more than 80,000 street trees and countless park, greenbelt, and backyard trees.

Figure 2. Map of study site

Seattle’s street trees are scattered throughout the entire city. Many streets have planting strips, while the trees in higher density areas are planted in square or rectangular planting areas, usually quite small. Trees in Seattle have long growing cycles, due to the high levels of moisture during the rainy season and brilliant sunshine during the summer.
Seattle is a medium-sized city, with a total population close to 570,000. The city is topographically diverse, with many steep hills and valleys. The city is long and narrow, approximately 4 kilometers across at its most narrow, 14 kilometers at its widest, and 27 kilometers from north to south. Seattle is bordered by Lake Washington to the east and Puget Sound to the west. Seattle is well known for its small neighborhoods, each with their own unique characteristics and demographics. The areas in the northernmost and southernmost parts of the city limits are more residential, with lower-density residential living, while the areas in the mid-area are more developed and dense.

COMMUNITY PARTICIPATION: RECRUITMENT
In order to satisfy the requirements of participatory research methods, high school students were recruited to participate in the project. Incorporating high school students would introduce a group of young people to the often unfamiliar field of urban forestry, provide a fresh insight into community resources and social justice issues, and enhance the project with the perspective of young Seattle residents. Students were recruited through visits and e-mails to high school environmental clubs, science and humanities classes, and various environmental educational organizations across Seattle. Figure 3 shows the flyer that was used to attract students to the project. After a handful of visits and organizational meeting, six students committed to work on the project for the duration of the summer–after an initial interest of forty-six students. The students were pertinent to the project, and their involvement was a testament to the benefits of including community members in projects related to the health of cities and urban residents.

Figure 3. Recruitment flyer

DATA SETS
This project included substantial data reconnaissance and transformation. Census information, from both 1990 and 2000, were combined and synchronized to classify Seattle’s high- and low-income census tracts. The tree inventory map data–used to select trees–was adjusted to work with a manageable number of species and trees, and was aligned with census information for accurate location and selection of the trees. The City of Seattle’s street tree maintenance records involved an enormous amount of data entering and recoding. In order to integrate the three types of data, months of data transformation work was necessary. The steps taken to manage the data are described, in further detail, in the following three sections. Table 1 lists this project’s type and source of data¹.

¹Full citations located at end of reference section, in the same order presented here.

STUDY AREA DATA CONTENT
Boundaries
Work began with locating 1990 and 2000 Census information for the Seattle area. Census boundary files and associated median household income, population, and ethnic make-up for each census tract in Seattle were downloaded via the U.S. Census Bureau’s American FactFinder and the Washington State Geospatial Data Archive (WAGDA). Census tracts are defined as small, relatively homogeneous areas, with respect to economic status, and living conditions, into which large cities and counties are divided (U.S. Census Bureau 1997). The GIS files were projected in Geographic Projection, Decimal Degrees for Map Units, and NAD 83 Datum. Using 1990 and 2000 data, a check for consistency of Census tract boundaries showed only five tracts dividing into two or more tracts between the selected years. Therefore, it was decided that using the 2000 boundaries would still provide accurate assessments of income for the areas from 1990 to 2000.
Areas that made up Seattle’s primary business and industrial land use areas (i.e., downtown, Safeco Field, Seahawk Stadium, and Harbor Island) were excluded from site selection. It was not appropriate to include non-residential areas, as this study was aimed at exploring the differences in tree condition between areas of diverse household, not commercial, income.
Comparison of income level between 1990 and 2000
Since tree condition is often a function of time, it was necessary to have fairly consistent levels of income over a set-period of time. To ensure that selected areas did not have above-average income increases or decreases in the recent past, change in income level between 1990 and 2000 was calculated. Using the following Consumer Price Index (CPI)² formula, 1990 dollars were converted to 2000 constant dollars:

2000 MHI = 1900 MHI * (2000 CPI/1900 CPI)
where CPI for 1990=130.7 and 2000=172.2

On average, Seattle’s census tracts’ median household income increased by 24.76% (+/- 20.98) between 1990 and 2000. Fourteen tracts were excluded because their change in median household income from 1990 to 2000 was greater than 46% (one standard deviation from the mean)³.

²CPI is a measurement that explains the change in the prices paid by urban consumers for a representative basket of goods and services from one year to another (reference). This means that the same basket of goods that an urban consumer could by for $130.70 in 1990 would have to be purchased for $172.20 in 2000.
³Percent change in median household income between 1990 and 2000 for Seattle’s census tracts are listed in Appendix A.

Distribution of Median Household Income
Seattle’s census tracts showed a normal distribution of median household income, with the majority of the tracts falling between $30,000 and $60,000 with an average median household income of $47,416 (+/- $16,000). Census tracts were divided into “high” and “low” income groups: greater than $47,416 and less than $47,416. Sixty-nine tracts were considered high-income, while forty-nine tracts were considered low-income, equaling a total of 108 acceptable tracts to select study sites from.

Selected Tracts
Six census tracts were randomly selected from the 108 acceptable tracts in the city– three of each income group. Median household incomes for each of the six areas were compared against each other to ensure an adequate distribution of income conditions. The selected tracts–from lowest income to highest income–were matched with the following approximate neighborhoods: the Central District, Fremont, Northgate, North Capitol Hill, Seward Park, and Madrona areas. Table 2 provides demographic information and total tree population for the six selected study areas. Additional information for all census tracts within Seattle’s city boundaries, not just the selected tracts, is presented in Appendix A.

A map of where the tracts are located in Seattle is shown in Figure 4. Northgate and Fremont are in the north, Seward Park is furthest south, and North Capitol Hill, Madrona, and the Central District are in the middle of the city and closest to downtown.

Figure 4. Map of study sites

TREE SAMPLE DATA CONTENT
Street Tree Inventory
The City of Seattle’s GIS-based Street Tree Inventory, generously provided by the City Arborist’s Office, was used as the data source for the tree sample. Although this inventory could be considered rather static (not used on a daily basis), the City Arborist’s Office makes relatively frequent updates and has added more than 10,000 trees to the database since the original inventory done in 1990 (Rundquist 2003). Each street tree is geocoded to a particular address and location in Seattle and includes information about species, size, planting location, planting date, management responsibility, and more. The full database includes more than 75,000 street trees.

Species
Seattle is known as the “Emerald City”–a testament to the abundant green vegetation that adorns the parks and hills of the city. Having close to 500 different tree species, Seattle’s streets are no exceptions to the diversity of the city’s natural landscape. Due to the enormous task of training the high-school students in tree identification and condition assessment for such a large number of species, it was necessary to limit the number of tree species that would be part of this study. Therefore, tree species that amounted to less than 0.5% of the total street tree population were excluded from the tree sample. This reduced the number of tree species from 500 to 41–a much more manageable number of species.
Using ArcView, the tree inventory shapefile was “clipped”–similar to using a cookie cutter to cut out pieces of dough–to include only those trees that made up more than 0.5% of the total street tree population. The inventory was further clipped to include only the trees that fell within the six selected tracts, resulting in a total of 4107 trees. Systematic random sampling was used to select forty trees from each of the six tracts, for a total tree sample of 240.
Table 3 shows the fifteen most common tree species on Seattle’s streets, and their composition of the total street tree population. Forty-one additional species, not listed in the table, made up the remaining trees that represented more than 0.5% of the total street tree population.

Tree Condition Rating
The Council of Landscape and Tree Appraisers Tree Assessment Guide includes a tree condition rating as part of the total assessment method (Council of Tree and Landscape Appraisers 1983). The condition assessment separates a tree into five factors: roots, trunk, scaffolding branches, smaller branches and twigs, and foliage. Figure 5 depicts a tree with lists of the characteristics used to gauge the level of problems associated with each of the five factors. When assessing the condition of a tree, each factor is looked at separately; using the characteristics in Figure 5 as guidance, each factor is assigned a score based on if the characteristics present on the factor are positive or negative. The following scores match the appropriate description of the level of problems (or lack of problems) associate with each score:

1 = extreme problem
2 = major problem
3 = minor problem
4 = no apparent problem

Figure 5. Tree condition assessment criteria

The five scores are then summed together to get one final condition score, with five (5) being the worst possible score and twenty (20) being the best. Consistent with the method used by the USDA Forest Service, Center for Urban Forest Research, and many other urban forestry professionals, scores were divided into four categories: dead or dying, poor, fair, and good. Table 4 lists the condition categories and their associated sums of the factor score and characteristics.

Reliability Test of Condition Rating
The tree condition assessment technique relies on professional judgment, is highly reliable, and other experts would repeat very similar condition levels using the same method. Additionally, the technique is not “sensitive,” meaning that it does not give variable results depending on species or age. The assessed condition of twelve trees was compared with a Certified Arborist’s assessments of the same trees, using the same described method, and the results showed a high level of reliability (Number of Cases = 12, F= 2.20, Reliability coefficient Alpha = 0.9242). The sample did not stratify by tree age or species because condition assessment does not compare older trees to younger trees or one different species to another species; it looks solely at a particular species at a particular age. However, the assessor must be familiar enough with the species to know the characteristics of a “perfect” specimen as well as common problems associated with that particular species.

Tree Condition Assessment – Data Collection
Data collection was performed between June 15 and August 1, 2003. Spreadsheets were created with a list of the selected trees for each area and maps, created in ArcView GIS, assisted in locating the exact location of a specific tree. Thirty-six trees were missing, recently removed, or incorrectly listed. Diameter (DBH), planting width, planting area type, and planting area ground cover were also recorded. If more than one tree of the same species was present at the specified address, a tree was chosen at random. If the specified tree was not found at the address provided, either a nearby tree of the same species was selected or it was noted that the tree was missing.

COMMUNITY PARTICIPATION: DATA COLLECTION
The students participating in the project were trained in tree identification and basic tree condition assessment. Using tree identification books and printouts of each tree species, the selected trees’ species were verified in the field. For tree condition, “cheat sheets” and diagrams (similar to Table 4 and Figure 5) were used to guide the students in the correct direction when gauging tree condition. In order to maintain consistency and accuracy throughout the entire project, the primary researcher was always part of the assessment team and assured that the appropriate condition scores were assigned. The experience was valuable for both the students and the project; the students gained new skills and the project was rewarded with knowing that students enjoyed the experience, as evident by the submission of the following poem from one of the students:

Standing tall and majestically,
Deep olive and light green hues contrasting against
The azure blue sky with white wisps,
Summer smells still in the air,
Sun burning down upon leaves,
Shadows cast down,
Stopping
Notepad in hand
To meticulously write measurements,
Side note to self: unusual number of tumors on bottom trunk
As you move on to the next tree.

TREE MAINTENANCE DATA CONTENT

Work Records
Beginning in March 1998, the City’s tree crew reports weekly “accomplishments” to the City Arborist. The weekly reports describes where the tree crew went, what type of tree care activity was performed, and how many of what tree species were worked on. Seattle’s Arborist Office generously provided all of the work records through December 2002. The total number of weekly accomplishments equaled 250 individual files, approximately one for each week between March 1998 and December 2002.
The “accomplishments” from each file were consolidated into a single spreadsheet and each entry included a unique ID, date, address (or intersection), type of tree work, number of trees worked on, and type of tree(s) worked on. The end result was a database with 1236 individual entries of tree maintenance performed between 1998 and 2002. Examples of these entries are shown in Table 5.

Each entry was matched to a location on the City’s Street Network shapefile, using with ArcView’s “geocode addresses” function. The “batch match” method resulted in 563 individual “events” matched with a 75-100% accuracy and 511 events matched with a 50-70% accuracy rate. An interactive match (not automatically done by the software) located 97 additional events and 64 events were unable to be matched at all. The end result was a GIS shapefile with 1171 separate work events, spanning the entire time period of recorded weekly accomplishments. The total of the work events were matched with the six areas and a new file was created to include the work events that fell within the boundaries of the areas.

ANALYSIS PROCEDURES

Tree Condition and Income
Tree condition was analyzed using the Statistical Package for Social Sciences (SPSS), Microsoft Excel, and ArcView GIS. Average tree condition was calculated for each of the six tracts, using the condition number (1-4) associated with each condition category (poor, fair, etc.). A OneWay ANOVA test was used to compare the mean condition across all tracts and an Independent Sample T-Test was used to compare the mean condition between the aggregated high and low-income areas (the six tracts were collapsed into one high-income variable and one low-income variable, resulting in a comparison of two variables).
The mapping of tree condition and work records across the city was deemed an important visual display, enabling city managers and decision makers, unfamiliar with urban forestry terms, to use visually displayed data to assess current patterns in management and make appropriate decisions. Maps displaying tree condition in the different areas were also produced. Relevant to the goals of this study, one author writes:

As a basis for planmaking, revelations about who benefits, presented in visual, spatially-oriented terms, enable communities to evaluate their distributional preferences and see whether or not they are in line with broader community goals and with notions of fairness (Talen 1998).

Work Records and Income
The distribution of tree maintenance (total visits and total number of trees worked on) was calculated for the entire city (See Appendix B), for each of the six tracts, and for the high and low-income areas. Total number of visits and number of trees worked on (from 1999-2002) were compared between the six areas and the aggregate low and high-income areas. Additionally, the type of activity performed at each location (pruning, removing, etc.) was calculated for the entire city (see Appendix B).
Work records from 1998 began in March and were slightly incomplete, thus were eliminated from analysis. Additionally, the number of visits and trees worked on were weighted by total street tree population and total city-maintained trees, to ensure an accurate comparison of tree maintenance in relation to the actual number of trees present and the number of trees under City responsibility⁴.

Work Records and Condition
Average tree condition among each area and between the high and low-income areas was graphed in relation to the number of visits and number of trees worked on per visit. Although this information is not adequate enough for robust analysis, a basic relationship between maintenance and average tree condition of the sampled trees can be detected with graphs and presented in tables. This information is presented in the Results section alongside the tree maintenance and income findings.

⁴Trees were weighted by total-city maintained trees due to an interesting, yet common, practice of tree responsibility in cities strapped for funding. The City Arborist Office, part of the Department of Transportation, manages Seattle’s street trees with help from Seattle City Light, the utility company, and Parks and Recreation. All city trees are considered publicly owned, but responsibility of the trees is often in the hands of the adjacent property owner. Unless the city physically planted the tree, the property owner is responsible for any pruning, removal, or costs associated with damage of the tree. However, the property owner still does not “own” the tree; they must apply for a tree work permit through the City Arborist, pay a $265 refundable fee, and have the final work inspected by the City. If the work is not up to standard, the fee is not refunded.
City planted trees, on the other hand, are the responsibility of the City Arborist’s Office – who only has one permanent tree crew, made up of two or three people. Most of the City maintained trees are located on the main arterial streets and commercial districts throughout Seattle. The one exception is a large part of the Central District; in the 1970’s the Federal Government funded a program called “Forward Thrust,” aimed at encouraging development and improvement in low-income areas. Seattle used a large part of that money to plant a large number of street trees in one of its poorer neighborhoods. Thus, the City is responsible for the maintenance of those trees. Although the City is not responsible for maintaining trees other than the ones they plant, the City tree crew will respond to occasional pruning requests, tree removal, and clean-up/clearing needs in neighborhoods throughout Seattle.

Other Factors
In order to ensure the validity of the findings, it was necessary to investigate if other factors were contributing to the average condition in the different areas. Although soil quality and type, air pollution levels, and precipitation levels could contribute to different tree conditions, they were outside the scope of the project. Nevertheless, the effect of tree diameter, genera, and planting strip width and cover on tree condition was analyzed.

Chapter 3: Results

This chapter presents the results from the tree condition assessment and the tree work records analysis. First, it compares the tree condition assessed in this study to the condition assessed in the City’s 1994 inventory. The second section reports average tree condition as it related to income and the third section reports the level of tree care as it related to income and tree condition.

COMPARISON OF SAMPLED TREES’ CONDITION AND CITY INVENTORY
With more than 98,000 street trees, Seattle boasts many tree-lined streets. However, the condition of the street trees varies. As shown in the Figure 6, this study found that approximately 23% of the sampled street trees were in good condition and 65% were in fair condition, with 12% of the trees in poor and dead or dying condition. Similar to the above percentages, calculations of tree condition from the entire inventory (N=70,023) showed that condition results from this study did not vary widely from the condition assessments performed by the City in their 1994 inventory⁵. This similar pattern of street tree condition between the two assessments was important to the reliability of this project’s findings as it increased the possibility of extending to the rest of Seattle.

5 In the original inventory, the City Arborist assigned condition values from 1-5 for each tree. For comparison to this study, the numbers were translated as follows: 1 = Dead or dying; 2 = Poor; 3 or 4 = Fair; 5 = Good.

Figure 6. Tree condition distribution

IS TREE CONDITION RELATED TO INCOME LEVELS?

Tree Condition in Six Tracts
Surprisingly, as Figure 7 shows, the six areas shared very similar mean street tree condition levels. Moreover, mean street tree condition showed no significant relationship to income; tree condition neither increased nor decreased with median household income⁶. Figure 7 shows the absence of any statistical relationship between median household income and average tree condition; the line is relatively flat across all income levels and the average condition scores do not vary by median household income.

⁶ One-Way ANOVA results = Sum of Squares = 1.617, df = 5, F=0.7598, p > 0.5

Figure 7. Average tree condition in sampled tracts compared by income levels

Condition Categories in Six Tracts
Figure 8 shows the percent of trees in the different condition categories for the six areas. The majority of the trees were in fair condition, with good being the second most common condition of trees. A Chi-Square Test performed on the distribution of tree condition categories showed no significant difference in distribution among the six tracts (Pearson Chi-Square Value = 15.129, df = 15, p > 0.4).

Figure 8. Distribution of tree condition categories in sampled tracts

Tree Condition in Low and High-Income Tracts
Similar to the results from the tract-by-tract analysis, average tree condition was not significantly different between the aggregated high and low-income areas (Figure 9). The high-income areas’ average tree condition was 3.09 (standard deviation = 0.6783), while the low-income areas’ average tree condition was 3.06 (standard deviation = 0.6269). A t-test performed on these means showed no significant difference between the two income areas, thus the null hypothesis could not be rejected⁷.

Figure 9. Average tree condition for high and low-income areasCondition Categories in Low and High-Income Tracts

CONDITION CATEGORIES IN LOW AND HIGH-INCOME TRACTS
Figure 10 shows that there was little to no difference in the patterns of tree condition distribution of between the high and low-income areas⁸. Although the high-income areas had slightly more trees in good and poor condition and slightly less trees in fair condition, the difference between the income groups was insignificant.

⁷ t = -0.4001, df = 195.88, p > 0.6
⁸ (Pearson Chi-Square Value = 0.9853, df = 3, p > 0.8)

Figure 10. Distribution of tree condition categories for high and low-income areas

Visual Display of Tree Condition
Figure 11 is a map displaying tree condition across all sampled tracts. Tree icons in four different colors represent the tree condition categories: dead or dying, poor, fair, and good. The tracts on the right-hand side, with purple backgrounds, are the three high-income areas and the tracts on the left-hand side, with green backgrounds, are the three low-income areas. At first glance, this visual representation of tree condition does not show any obvious distinctions in color patterns – nor tree condition - among the different areas. There is a large presence of yellow and green trees scattered throughout all of the areas, with no particular differences in location. On the other hand, there are a larger number of red and orange colored trees in Seward Park – one of the lower average tree condition tracts. This tool can be useful for managers to notice subtle patterns of tree condition throughout a city, rather than needing to decipher complicated numbers and charts.