Connecting the Dots: Making tree inventory analysis smarter

Page 1

CONNECTING THE DOTS: Making tree inventory analysis smarter A case study of street trees in Eugene, Oregon By Joel R. Grogan June 2014

Zoning Residential Parks & Public Land Commercial& Industrial Agricultural Other Not Zoned

Land Use Created 3/10/2014

Tree Canopy over Public and Private Property

Tree Defects

en t

Tre e Inventory

A ge

pm o l e of Dev

Tree Size

Created 3/9/2014

Submitted in partial fulfillment for the Master of Landscape Architecture Department of Landscape Architecture, University of Oregon

ÂŻ

0 0.5 1

2

3

Miles 4


Approval Page

Student: Joel R. Grogan Title: Connecting the Dots: Making tree inventory analysis smarter, a case study of street trees in Eugene, Oregon

Project Chair

Rob Ribe, PhD

Project Committee

Roxi Thoren

Chris Enright, PhD

Deni Ruggeri, PhD

2

Rob Ribe, PhD

DATE

Roxi Thoren

DATE

Chris Enright, PhD

DATE

Deni Ruggeri, PhD

DATE


Acknowledgments This project was inspired by my work for the City of Eugene, where I assisted in preparation of the City’s first two neighborhood tree inventory reports, collected tree inventory data, and enjoyed every moment of it. I would like to thank: Rob Ribe, for pushing me to fully pursue this challenge; and for generously sharing his time and expertise; the City of Eugene’s Urban Forestry Office, especially Scott Altenhoff and Eric Cariaga, for sharing their knowledge, inspiration and passion for the urban forest; my brother, Tristan, who voluntarily served as my teacher and consultant in performing the statistical analyses; my family for their constant support and for encouraging me to pursue this degree; and my wife for her continued patience, understanding, and love.

3


4


Contents Acknowledgments 3 Index of Figures

6

Index of Tables

7

Abstract 8 Introduction 9

INTRODUCTION TO URBAN FORESTRY

11

History of American Urban Forestry

12

Contemporary Urban Forestry

16

Urban Forests and Land Use

29

EUGENE’S URBAN FOREST

31

Eugene’s Trees

32

Eugene’s Street Tree Inventory

36

RESEARCH DESIGN

41

Introduction

42

Simplified Research Process

42

Detailed Research Process

44

Methods Employed in This Analysis

45

FINDINGS 49 Results of One-Way ANOVA

50

Results of Regression Analysis

57

Results of descriptive analysis

60

DISCUSSION 63 Interpreting Results for Mean DBH

65

Interpreting Results for Structural Defects

66

Interpreting the Attributes of the Urban Environment

67

Implications for Managers

68

Implications for Scientists

68

Conclusion 69

APPENDIX 73

5


Index of Figures

6

Figure

1

Historic map of Philadelphia street trees

p. 13

Figure

2

Diagram of several benefits of urban trees

p. 17

Figure

3

Typology of publicly-owned urban trees

p. 18

Figure

4

Common data fields recorded in a tree inventory

p. 22

Figure

5

Typical DBH distribution chart from Davidson, NC

p. 23

Figure

6

Relationship between tree lifespan, costs, and benefits

p. 24

Figure

7

Diagram of urban soil conditions

p. 25

Figure

8

Screenshot from San Francisco’s urban forest map

p. 28

Figure

9

Eugene, Oregon context map

p. 32

Figure

10

Eugene zoning and tree canopy per zone

p. 33

Figure

11

Eugene’s tree canopy

p. 34

Figure

12

Eugene’s street tree density

p. 35

Figure

13

Screenshot from JWN inventory map

p. 37

Figure

14

Structural defects in urban trees

p. 38

Figure

15

Simplified research process

p. 43

Figure

16

Means plot: DBH by zoning classification

p. 52

Figure

17

Means plot: DBH by land use

p. 55

Figure

18

Scatterplot of DBH by annex year

p. 56

Figure

19

Structural defects by zoning classification

p. 61

Figure

20

Structural defects by land use classification

p. 61

Figure

21

Structural defects by planter width

p. 62

Figure

22

Structural defects by iTree size codes

p. 62


Index of Tables Table

1

Comparison of urban forest assessment methods

p. 21

Table

2

Common elements of a tree inventory report

p. 27

Table

3

Portland’s Urban Land Environments

p. 29

Table

4

Summary results of Jefferson Westside Tree Inventory Report

p. 37

Table

5

Detailed research process

p. 43

Table

6

Identification and selection of variables

p. 44

Table

7

Table showing simplification of land use categories for this study

p. 45

Table

8

Results of one-way ANOVA for predictor variable sets

p. 51

Table

9

Homogeneous subsets of DBH by zoning

p. 52

Table

10

Homogeneous subsets of DBH by land use

p. 55

Table

11

Zoning mean DBH and annexation year

p. 56

Table

12

Land use mean DBH and annexation year

p. 56

Table

13

Multiple regression for DBH: Size codes

p. 57

Table

14

Simple regression for DBH: Age of development

p. 57

Table

15

Simple regression for DBH: Planter width

p. 57

Table

16

Multiple regression for DBH: Age of development, size codes, and planter width

p. 58

Table

17

Multiple regression DBH: Zoning indicators

p. 58

Table

18

Multiple regression DBH: Land use indicators

p. 59

7


Abstract Urban forests’ contributions to climate mitigation, air and water quality, property value, and human health and well-being are increasingly recognized and appreciated; however, these benefits are increasingly at risk. Shrinking municipal budgets, global climate change, and urbanization have increased the need for efficient prioritization of urban forest management, maintenance, and planning activities. Though urban foresters have long recognized the influence of land use type and conditions on urban forest characteristics, their ability to examine this relationship at the local level has been limited by available data and a lack of simple and efficient means for analysis. The growing use and accessibility of GIS-based tree inventories offer new opportunities for urban foresters to better understand these influences. Street tree inventory data from Eugene, Oregon was analyzed to quantify relationships of individual tree attributes with adjacent land uses and conditions. Results indicate significant differences among land uses and site conditions in predicting diameter at breast height (DBH) and structural defects. This study suggests that urban forest managers with access to tree inventory data have opportunities to enhance their understanding of factors that influence tree characteristics and health in their cities, and direct management accordingly. Results from Eugene indicate that age of development, species characteristics, and planter width have the greatest influence on DBH. Basal and structural defects are disproportionately high in industrial areas, suggesting that design interventions to protect tree roots and trunks could reduce mortality in these areas.

8


Introduction

T

REES IN THE CITY is a concept that has been embraced by urban dwellers for millennia. Collectively, these trees are known as the urban forest, made up of all the trees within and surrounding a city, regardless of ownership - including trees along streets and backyards, as well as stands of remnant forest (Nowak et al., 2001). Urban forests are almost entirely the product of human activities (or lack thereof). For this reason, long term planning and management plays a critical role in preserving and enhancing the social, economic, and environmental benefits that urban forests provide. The goal of urban forestry should be to maintain and expand the function of the urban forest, even while its form and distribution is constantly in flux. In recent decades, much research has been devoted to understanding and quantifying the benefits of urban forests. In Oregon, many municipalities have developed comprehensive plans and programs that attempt to maximize these benefits, including Eugene, Corvallis, Gresham, and Portland. Tree inventories are a key component of comprehensive urban forest management, containing detailed information such as location, species, size, and maintenance needs. Reductions in municipal budgets, combined with advances in calculating benefits of urban trees, has put additional pressure on managers to devote resources to collection of tree inventory data, in order to justify their programs. Aside from the resources required to conduct an initial inventory, there

remains the problem of how to keep the information current. A variety of urban forest assessment methods are available. Sample inventories, which use data from a stratified sample of trees to extrapolate findings to the whole tree population, are one method of addressing this challenge. Another strategy is the use of aerial or remotely-sensed imagery to analyze urban tree canopy cover. Though useful, neither of these alternatives has the same ability to directly inform urban forest management and maintenance activities. The reality is that, though expensive, there is no substitute for a complete, on the ground inventory. Typically, tree inventory data is analyzed by city or neighborhood boundaries. Though there are logical reasons for this approach, these attributes have been shown to have relatively little influence on urban forest structure and function. Without a significant return on investment, it is unlikely that this rich source of data will continue to be collected. Tree inventory data should serve to finetune management of urban forests to the unique biophysical and sociopolitical context of a particular city. The question is, how can tree inventory data be better analyzed to inform and improve urban forest management and maintenance activities? I have developed a process that enhances the ability for tree inventory data to inform management and maintenance activities. This process can be conducted using a variety of urban environmental attributes.

“City plants are a natural resource, but human function and fashion are often more influential than natural processes in determining the location and arrangement of plants.� (Sprin, 1985, p. 172)

9


10


INTRODUCTION TO URBAN FORESTRY

1


Introduction to Urban Forestry URBAN FORESTRY is the art,

lined streets, with commercial areas and

science, and technology of managing

wealthier residential neighborhoods being

trees and forest resources in and around

more likely to have trees (Lawrence).

urban community ecosystems for the

Figure 1 is a portion of the first Philadelphia

physiological,

economic,

map to show the location of its city’s trees.

and aesthetic benefits trees provide

Some early municipal tree ordinances

(Konijnendijk

sociological, et.

With

limited tree-planting by citizens, while

continuing population growth and the

others encouraged it. In 1789, New York

global migration from rural to urban areas,

City adopted an ordinance preventing

urban forests are becoming an increasingly

the planting of street trees south of the

valuable tool in the attempt to create

common, “except before churches or other

healthy cities. Collectively, these trees have

public buildings…,” (Lawrence, 2006, pp.

demonstrated a dramatic effect in reducing

164-165) a rule which prohibited street

air and water pollution, moderating

trees in over eighty percent of the city’s

temperature, sequestering carbon, and

current land area. In 1799, it adopted

reducing

psychological

a policy permitting trees on all streets

ailments in urban dwellers (Dwyer et al.,

over forty feet wide, and in 1806 even

2000; Nowak et al., 2002). These benefits

encouraged residents to plant trees in

are likely to become increasingly valued in

front of their houses. The planting of

the face of global climate change.

trees in cities was becoming a common

physical

al.,

and

2006).

History of American Urban Forestry The beginning of public natural resource management in the United States is as old as its first colonies, where 17th century New England towns appointed tree

wardens

to

govern

communal

woodlots that were used for fuel and construction materials (Forest History Society, N.D.). By the early 19th century, citizens of many social classes were planting trees in most cities and towns in Europe and North America (Lawrence, 2006). Around this time, cities such as New York, Philadelphia, and Baltimore were noted for several prominent tree-

12

practice in many western countries. The role of average citizens in creating a treed social landscape of streets and parks was especially prominent in the United States (Lawrence). By the 1850s, the industrial revolution had

transformed

many

cities

into

manufacturing centers – which had a dramatic influence on their economic base, population and demographics, and their physical size and complexity (Lawrence, 2006). Commuter railroad lines enabled development of new suburban areas of detached and semi-detached housing outside of many large cities – including Boston, New York City, Pittsburgh, Cincinnati, and Chicago. Though suburbs


Figure 1: This 1796 map of Philadelphia was the first to show the location of city trees. Map by John Hills. The Library Company of Philadelphia, as cited in Lawrence, 2006, p. 160

13


varied in their density, they were defined

Frederick Law Olmsted, H.W.S. Cleveland,

by their availability of private garden space

and Charles Eliot began to see city parks

– as they still are (Lawrence). Citizens

as part of a larger network of open space

leaving the city centers brought their

systems (Lawrence, 2006). Though urban

practice of planting street and park trees

open space was beginning to be viewed as a

with them, creating expansive new areas of

larger system, comprehensive management

green urban landscapes.

of the urban forest didn’t emerge until the

In the second half of the 19 century, th

urban forests in the United States were

The new science of individual tree

distinguished from European examples

care, arboriculture, was formally described

by the abundance and distribution of tree-

in a 1911 book by Bernard Fernow

lined streets and by the significant role of

that specifically addressed street tree

private citizens in tree planting. Lawrence

maintenance (Johnston, 1996). Though

elaborates upon this difference,

pressure toward holistic management

“While a large boulevard in a European city was often the only tree-lined street in a particular neighborhood, in America almost all residential streets might be planted with trees. By the third quarter of the nineteenth century, civic improvement associations in cities and towns were calling for civic authorities or volunteers to beautify city centers and to plant trees along city streets. Individuals were exhorted to plant trees in front of their houses.”

(Lawrence, 2006, p. 246) Further evidence of the wide support for tree planting is the founding of Arbor Day in 1872, a national day of tree planting in towns and rural areas that is celebrated to this day. Parallel to the increasing interest in city trees was a growing ecological awareness in urban design in America. Prominent landscape planners and park designers of the time, such as

14

mid-1960s (Johnston, 1996).

of the urban forest began in the 1930s, municipal vegetation management in the United States was mainly focused on tree planting, tree maintenance, and landscape architecture (Deneke, 1978). Dutch Elm Disease caused one of the most significant events in the history of urban forestry, the death of over 40 million American elm trees (Ulmus americana) across America between the 1930s and the early 1970s (D’arcy, C.J., 2000). Many cities from the Atlantic to the Rockies had used elm trees almost exclusively since the 19th century; trees that were suddenly gone (McCombs, 2001). By the 1950s, this heartbreaking loss of so much of the nation’s urban forest helped galvanize the nation’s burgeoning municipal forestry movement. At

the

same

time,

two

major

developments had diluted the zeal of citizen involvement in urban forestry: refinement of air conditioning technology and mass production of the automobile.


These technologies virtually eliminated the most obvious social benefit of urban trees – by providing easy and instant relief from heat and the ability to escape the city to more verdant suburban environments (Bartenstein, 1981). Despite disciplinary

this

setback,

approach

a

toward

multiman-

agement of the whole urban forest resource began to emerge in the early 1970s when academic, municipal, federal, and private-sector efforts coalesced into the founding of a new discipline, urban forestry. Undoubtedly, the rise of the environmental movement marked by the first earth day in 1970 contributed to this phenomenon. Though the term “urban forestry� first appeared in the context of a university thesis in Canada, the subject grew rapidly into a national movement in the United States. This was mostly due to widespread positive attitudes toward urban forestry adopted by many professional organizations concerned with urban trees (Johnston, 1996).

15


Contemporary Urban Forestry Forestry

Though the majority of urban trees

Conference, held in 1978, marked the

occur on private property (Mincey, 2013)

transition into the modern practice of

most municipal urban forestry programs

urban forestry. The USDA Forest Service

are focused on management of publicly-

sponsored the conference, along with the

owned trees (Nowak et. al., 2010). The

State University of New York’s College

primary reason for this is that property

of Environmental Science and Forestry.

owner’s rights usually limit management

This was the first national conference

options on private land.

The

National

Urban

on the topic to explicitly use the term

Urban trees on public land are often

“urban forestry,” and it drew over 450

divided into three basic categories, based

professionals from around the country

on their urban context: trees in natural

(Johnston, 1996).

areas, trees in parks, and street trees

Between 1981 and 1984, TreePeople’s

(illustrated in Figure 3). These categories

successfully

describe adjacent land use, and reflect

motivated the public to plant over one

differences in how trees are established,

million trees in Los Angeles, California

their growing conditions, and the intensity

(City of Los Angeles, 2006). The effort

of maintenance activities.

Million

Tree

Campaign

was partly marketed as a run-up to the

Though all trees provide considerable

upcoming 1984 Olympics. City planners

benefits to livability, the local economy, and

had calculated the air quality benefits of

ecosystem services, the location and health

an additional million trees and hatched

of a tree can dramatically influence the type

the idea for the campaign (Johnston,

of benefits it provides. For example, trees

1996). This was one of the first large-

that are closer to air pollution sources, such

scale municipal tree planting programs

as automobile traffic, can provide greater

that focused on planting trees for their

air quality benefits than those further away

estimated contribution to urban air

(Nowak, 2002; Morani et al., 2011).

quality. Figure 2 illustrates several of the benefits of urban trees recognized and managed in contemporary urban forestry. Urban forest management plans began to emerge in the 1990s, with cities such as Eugene, Oregon (1992); Sacramento, CA (1994); and Portland, Oregon (1995) completing some of the first plans on the West Coast. These plans recognized and responded to the need for clear, quantified goals in steering urban forest management.

16

Urban forest distribution

Urban forest assessment Comprehensive,

efficient,

and

successful management of a resource depends on accurate information. The concept of a tree census or inventory was a direct response to this need. Historic plans and maps of cities dating from the late 18th century, such as the 1796 map of Philadelphia shown in Figure 1, noted the current or desired location of trees, but did


Figure 2: Several important benefits of urban trees

based on Phillips, 2012 and iTree, 2014

Stormwater runoff reduction Energy savings & avoided emissions due to shading

Aesthetic value & Increased real estate value

Air pollutant reduction CO2 Sequestration

not note individual characteristics of each

geographical databases which combined

tree, such as its species, size, or condition.

a tree’s locational information with its

In contrast, early tree inventories, such as

inventory attributes. For the first time, the

Portland’s 1938 street tree census, a project

map of tree locations and the tree’s recorded

of the Works Progress Administration,

attributes were combined into one entity,

collected both locational and descriptive

allowing spatial data to be queried based

information about individual trees (City of

on attributes of individual trees or their

Portland Parks & Recreation, 2004).

locations (Maggio, 1986).

Until

the

advent

of

personal

Accessibility

of

relatively

robust,

computers in the 1980s, municipal tree

affordable, and user-friendly interface

inventories required, “. . . index cards,

for tree inventory data collection and

ledger books, and paper maps, which

management is only one reason for the

made data-searching, manipulation, and

dramatic increase in municipal tree

analysis difficult.” (Bond and Kotwicka,

inventories since the 1990s. Reductions in

2013, p.4). Increased access to Geographic

municipal budgets have placed increasing

Information Systems (GIS) software and

pressure on urban foresters to quantify

personal computers gradually resulted in

the economic benefits of the urban forest

a shift from paper-based inventories to

in order to justify investments in urban

17


Typology of publicly-owned urban trees Trees in Natural Areas: General description

Origin

Occur in remnant or restored areas within or surrounding cities. Are typically woodland or forest landscapes but occasionally in more open areas.

Occur within developed parks (parks with site improvements such as lawns, benches, playgrounds, restrooms, and other infrastructure).

May have preceded development of the adjacent city.

Are often planted intentionally, selected by landscape architect or parks planning staff.

Management activities, or a lack thereof, have allowed regeneration of these wild populations.

Are sometimes remnant trees which occurred prior to the park’s construction are retained

Invasive, non-native trees and other plants can have dramatic influence on these plant communities.

Most ‘volunteer’ trees are removed before establishment in accordance with the park’s maintenance regime.

Are often in their native environment, growing conditions are generally good. Growing conditions

Maintenance

Park trees:

Often have plenty of growing space, with few overhead obstructions and relatively good soils and rooting space. Often irrigated along with adjacent lawns or other vegetation.

In fire-adapted landscapes, lack of disturbance can facilitate growth of underbrush and competition from non-fire-adapted species.

These trees are often allowed to grow well into maturity.

Mostly vegetation management, often to control undesirable invasive (non-tree) species.

Typically receive only occasional maintenance after establishment.

Trees are generally left untouched except in cases of hazards to people and property.

Pruning is often limited to hazard mitigation. Lack of ‘targets’ under park trees allows them to be retained even after they start to decline in health. When removed, park trees are typically replaced.

18


Street trees: Are trees growing within the city-owned right-of-way adjacent to city streets. Are often selected by the adjacent property owner from an approved street tree species list. Are sometimes planted by developers in accordance with City permit requirements/ City code

Figure 3: A typology of publicly-owned trees in an urban area. Synthesized from: City of Gresham, 2011; City of Portland, 2004; City of Seattle Urban Forest Coalition, 2007; Duh & Flanagan, 2009.

In residential areas, many street trees are planted by adjacent property owners, with or without support from community tree advocacy groups. The right-of-way may include existing naturalized trees which were not planted intentionally. Though street trees face some of the toughest growing conditions in cities, these conditions can vary widely. Are often very limited by rooting space and overhead utility lines, vehicle sight lines, and street lights. In low-density residential areas, street trees are generally less limited by soils and overhead obstructions. Historically, many cities performed cyclical maintenance pruning of street trees . Most cities have passed this responsibility onto adjacent property owners, limiting city maintenance to hazard mitigation and abatement. This has left most street trees without maintenance pruning, resulting in declining tree health and condition.

19


forests and street trees (American Public Works Association, N.D.).

Methods of urban forest assessment. There are two primary methods for

The growing importance of urban

assessment of urban trees: aerial (uses

forest assessment is driven by the need to

remote sensing and GIS) and ground-based

quantify benefits of urban trees, prioritize

(uses on-the-ground data collection). Such

maintenance and management activities,

data collection can be done either through

and measure change over time. Over the

a representative sample or a complete

past 25 years, technological advances

inventory. Table 1 compares these methods.

in data collection and analysis have

The strength of aerial assessments is that

contributed to municipalities’ increased

they capture entire urban forests, regardless

interest

inventories

of land ownership. Since private lands

(Cumming, Twardus, and Nowak 2008).

often dominate a city’s area, assessing only

The detailed analysis that once required a

public trees provides an incomplete picture

large and dedicated planning staff can now

of the urban forest resource; however, the

be easily carried out by field technicians,

role of urban forest managers is primarily

temporary staff or trained volunteers. This

the management of publicly-owned trees

is affording small and medium-sized cities

– for reasons discussed previously under

the ability to integrate local tree data into

“Urban Forest Distribution,” p. 16.

in

urban

forest

management decisions (Latimer, 2010).

According to Nowak (2008), most on-

Another factor driving increased efforts

the-ground data collection is conducted on

to assess and quantify the urban forest is the

street and/or park trees due to their “public

growing interest of professionals in related

ownership, high visibility, and relatively

fields such as landscape architecture, urban

easy

planning, and ecology. Current emphasis

provide structural data for individual

on green infrastructure, green streets, and

trees – making them more applicable to

low impact development all share a desire

urban forest management activities than

to maximize the quantity and quality of

aerial assessments. He suggests that a mix

ecosystem services in urban areas - and

of on-the-ground and remotely sensed

justify their costs. Since trees are one of the

assessment methods provide the most

largest contributors to ecosystem services

comprehensive data for improving urban

in urban areas there has been growing

forest management.

accessibility.”

These

inventories

interest in the quantification of these

For purposes of this paper, assessments

services. The USDA Forest Service’s iTree

derived from ground-based data will

suite is a state-of-the-art, peer-reviewed

be referred to as ‘tree inventories’ while

tool for urban forest assessment and

aerial-based assessments will be referred

calculation of tree benefits (i-Tree, N.D.).

to as ‘tree canopy assessments’. My research suggests that most municipal tree inventories focus exclusively on street

20


Urban Forest Assessment Methods Top Down (Aerial-based)

Bottom-up (Ground-based)

Cost Necessary equipment

Relatively low cost

Relatively expensive

GIS data, imagery, and software

GIS data, paper forms or handheld PDAs, DBH tape

Necessary skills

Advanced GIS experience

Tree ID experience, can be done with trained volunteers

Accuracy

Varies depending on data

Relatively precise, depends on data collectors

Strengths

Best tool for assessing total urban forest canopy (includes private property)

Detailed structural data for individual trees

Weaknesses

Does not provide structural Time and labor intensive data needed for urban forest (costly) management (Species, number of trees, trunk diameter, tree health)

trees, rather than other publicly owned

height and width of canopy, likelihood of

trees like those in parks and open space

failure, maintenance history, work priority,

(City of Portland Parks & Recreation,

and site conditions may also be collected

2004; City of Eugene, 2014). This may be

(Bond & Kotwica, 2013).

because the responsibility to care for public

Miller (1997) defines the two main

trees has often been assigned to municipal

components of tree inventory data as tree

public works departments, whose primary

descriptors and site descriptors; these are

job is construction and maintenance of

illustrated in Figure 4. Tree descriptors

streets and other infrastructure. Another

are characteristics of trees themselves,

potential explanation is that street trees

as opposed to site descriptors, which are

typically receive the largest share of tree-

characteristics of the tree’s surrounding

related expenditures in the municipal

environment. Some of the most common

budget (Kielbaso, 1990).

tree descriptors are size, species, condition,

Tree inventories: The foundation of any tree inventory is the extent of its study site and the attributes recorded for each tree. The International Society of Arboriculture’s best management practices for tree inventories suggest the following basic fields: location, tree species, diameter at breast height, and condition. Depending on resources and desired outcomes, additional fields such as

Table 1: Comparison of urban forestry assessment methods. Based on Nowak, et al., 2008

and defects. Site descriptors include planter width, presence of overhead utilities, and adjacent land use.

Tree descriptors and their management implications. Size Though trunk diameter, tree height, canopy spread, and biomass volume are all used to describe the size of street

21


Tree descriptors Crown Width

Figure 4: Common data fields recorded in a tree inventory.

Site descriptors

Based on Miller, 1997

Overhead wires

Crown Height

Diameter at breast height (DBH)

Species, Defects, Condition

Pollution problems Street width

Land use Planter width

Location

Underground utilities

Soil type

trees in inventories, diameter at breast

since there are few young trees (less than

height (DBH) is the most frequently used

12 inches in DBH).

measurement of tree size. It is commonly

Steenberg

al.

(2013),

further

used to measure tree management costs

describes management implications of

and benefits due to the relative ease and

DBH distribution,

accuracy of its collection (Miller, 1992; Vogt et al., 2013). DBH is the diameter of a tree’s trunk taken at “breast height,” which is 4.5 feet from the base of a tree. DBH is also commonly used as a proxy for both tree size and tree age, although this substitution does not account for species-variation (Toronto and Region Conservation Authority, 2011; City of Austin, 2013). When employed as a proxy for tree age or size, DBH can describe the age and size structure of the urban forest. Many urban forest managers would interpret the DBH distribution in Figure 5 as a critical need to increase tree planting,

22

et

“Structural diversity in urban forests is important for avoiding even-aged conditions where the mature trees of entire streets or neighborhoods reach the end of their functional lifecycle at the same time, abruptly halting the services they provide . . . and commanding high removal and maintenance costs”. (p.141) Figure 6 provides a conceptual diagram of the of relationship between tree lifespan, environmental benefits, and maintenance costs. The cost of pruning increases with a tree’s size, with a sudden and significant


i. Davidson Stand Dynamics

Diameter at Breast Hieght(DBH) Distribution

Diameter at breast height (DBH) distribution

600

Figure 5: Bar graph illustrating DBH distribution for a street tree inventory in Davidson, North Carolina. A universal element of any street tree inventory report.

553

Number of Trees

500 386

400 300

215

208

200 118

110

16-30"

31 - 35"

From Anderson et al., 2007

100 22 0 < 12"

12 - 15"

16 - 20"

21 - 25"

36" <

Diameter in Inches

increase in cost once pruning can no longer

stressor.

A

general,

widely-accepted

be conducted from the ground.

guideline for diversity of urban forests,

often credited to Santamour (1990), is the Species Top 5 Most Common Tree Species 10 – 20 – 30 rule, meaning that no species SPECIES Count % Dist A tree’s growth rate and potential Common Name should make up more than 10% of the size are largely dependent on species Quercus phellos willowpopulation, oak 429than 20%, 26.61% no genus more and characteristics. Street trees are often loblloly pine Pinus taeda 125 7.75% no family more than 30%. Acer rubrumvarieties (cultivars) of species, red maple 97 6.02% cultivated Juniperus virginiana eastern red cedar 78 4.84% which are selected based on their Condition & Defects Liquidambar styraciflua sweetgum 72 4.47% predictable form, size, aesthetics, or for Many tree inventories ascribe other qualities that make them well-suited condition values to each tree, based on a to urban conditions, such as tolerance of general rating of a tree’s health, vigor, and drought or limited root space. defects. Though there is no standardized Urban forest managers employ species system for rating tree condition, most use distribution and abundance in setting either percentages or a qualitative scheme priorities for increased species diversity, that essentially rates the tree as “good, fair, determining susceptibility to current and poor, or dead” (Bond & Kotwicka, 2013). future pests, and projecting maintenance Although tree defects are often recorded needs (Bond & Kotwicka, 2013). Species separately from tree condition, the diversity is desired for similar reasons number and type of defects often directly as structural diversity; a more diverse influences the tree’s condition rating. population of trees is less like to succumb Defects may be separated into descriptive to the same type of pest or environmental categories based on their causes, or the

23


Figure 6: Conceptual diagram of tree lifespan, maintenance costs, and environmental benefits. From Mincey et al., 2013

type of management intervention they

sidewalk and the curb of the street,

require, such as structural defects (tree

surrounded on either side by impervious

form), biotic defects (pest or disease), or

surfaces. Often, narrower planter widths

cultural defects (damage by people).

mean less soil is available for rooting, due

The primary concern of an urban

to the relatively shallow depth of gravel

forester is safety and hazard abatement

and engineered subgrade installed under

(Bond & Buchanan, 2006). Condition

and around these impervious surfaces.

ratings and defects can provide means for

Tree grates and tree pits, often found

targeting and scheduling maintenance or

in highly-urbanized areas where space

risk-abatement activities. Spatial patterns

for trees is scarce, are often less than four

in the occurrence of a particular defect

feet wide in each direction. The only soil

can help urban foresters target areas for

accessible to roots in these types of planters

investigation or maintenance.

is whatever material was installed in the

Site descriptors and their management implications. Planter width Planter width is the width of the unpaved area in which a street tree is planted (See Figure 4). Though planter width measures the width of soil on the surface, it is also used as a relative measure of a tree’s available root space. In many cases, this is a strip of land between the

24

basin during construction. Urban foresters have long been concerned with the lack of available rooting space in the city, and poor design practices continue to contribute to inadequate conditions for healthy trees. James Urban, a certified arborist and licensed landscape architect, has written a book entitled Up By Roots (2008), that is intended to educate landscape professionals on the best ways


to accommodate tree growth in urban

Because of this variability, available soil

areas with challenging growing conditions.

data such as the soil maps provided

From an urban forestry perspective, planter

by

width is one of the greatest limitations to

Service are not adequate for urban sites;

planting of large canopy trees, which need

instead, urban soil assessments should be

a large rooting area.

conducted to determine soil conditions at

Soils The physical properties of soils such as texture, structure, and nutrient-holding capacity are important in determining their suitability for plant growth. For plants, pH is likely the most significant aspect of soil chemistry, since many plants tolerate a limited range of pH and it is difficult to permanently change (Urban, 2008). Urban soils, which occur anywhere soils have been moved, graded, compacted, or contaminated, illustrated in Figure 7, are

National

Resource

Conservation

a specific location (Urban). One of the most common tree-related maintenance issues for urban foresters is sidewalk damage, which is often caused by tree roots and a combination of poorly drained soils, inadequate rooting space, and species characteristics. Matching appropriate tree species to available soil conditions and volume is another role of the urban forester.

Age of development The

relationship

between

urban

often manufactured or haphazard mixtures

vegetation and the time since initial

of sand, clay, silt, or other materials (Urban,

development of its surroundings has been

2008). Soil disturbance in the urban

well-documented (Miller, 1997; Heynen

environment results in highly variable soil

& Lindsey, 2003; Conway & Bourne,

conditions, even across small distances.

2013; Steenberg et al., 2013). Following Figure 7: Diagram of urban soil conditions. A, B, C, R. Remnant and buried topsoil may be encountered. C1-C5. Fill soils of various types, bulk density levels, and consistencies from heavy clays to sands. Expect soil interfaces between different soils. 1. Existing impervious surfaces. 2. Buried impervious surfaces and structures. 3. Buried trash and debris; may be unstable organic trash or compacted rock and gravel. 4. Eroded topsoil layer and exposed fills. 5. Exposed original subsoil with low organic content. From Urban, 2008.

25


the city beautiful movement of the early 20

Land use

century, many municipal forestry

Past and present urban land use

programs in the United States flourished

shapes urban conditions and vegetation

with strong public support and adequate

(Pauleit & Duhme, 2000; Wilson et al.,

funding (Miller 1997). Bartenstein (1982)

2003). Residential areas typically have less

describes two events that eventually diluted

impervious surface and more vegetation,

the constituency for urban forestry:

while commercial and industrial areas

th

“Air conditioning reduced the importance of trees in modifying urban microclimates, and the automobile provided city dwellers with a means of escaping to suburban environments where larger house lots provided private green space. Other activities contributing to the erosion of city tree populations include urban renewal, street widening, absentee ownership of housing, budget constraints, safety concerns by municipal government, and Dutch elm disease.”

generally have more impervious surface

(Cited in Miller, 1997 p. 207).

or size class – and assessing their quality

Though

age

of

development

and less vegetation. Different land uses present different opportunities and challenges for tree planting, maintenance, and public/private partnership; and trees in different land uses often fall under the jurisdiction of different parts of the municipal government (City of Portland Parks and Recreation, 2004). The relationship between land use and urban forest management is addressed on p. 29.

Tree inventory analysis. Tree inventory reports summarize data by grouping categories – such as species is

rarely a primary concern for urban forest managers, it seems to have a large influence on many site descriptors, such as planter width, soil conditions, species selection, defects, and tree size. Generally, older areas of town have larger trees, possibly due to better soil conditions and wider planter strips or survival of the fittest trees.

and distribution. Common elements of basic tree inventory reports include textual descriptions, charts, tables, and graphics that are intended produce a clearer picture of urban forest structure and composition of the study area, such as the DBH chart in Figure 5 (Bond & Kotwica, 2013). See Table 2 for a description of common elements of a tree inventory report. These reports serve as a snapshot of current conditions, and can sometimes provide insight into the current trajectory of the urban forest. Some cities such as Seattle, San Francisco, and Salt Lake City have used tree inventory data to create interactive web-based maps to engage and inform the public (Seattle

26


Department of Transportation, 2014;

the neighborhood. Another is the street

Urban Forest Map, 2014; Salt Lake City

typology approach described in Portland’s

Corporation, 2014). Figure 8 is a screenshot

assessment of the costs of managing street

from San Francisco’s entirely crowdsourced

trees as a public asset (Davey Resource

interactive tree inventory.

Group, 2009).

While

urban

tree

canopy

and

ecosystem assessments based on remotelysensed data often employ land use characteristics (Mincey, 2013; Wilson et al., 2003) and other geospatial data in their analyses, analysis of ground-based tree inventory data often focuses primarily on the data fields collected as part of the inventory (Frank, 2006; City of Cambridge, 2011; Portland Parks & Recreation, 2013). This ignores the potential value of using other geospatial data as a part of the

Neighborhood tree inventory approach. Portland, Oregon has developed a model neighborhood tree inventory program which focuses on a communitybased approach toward street tree inventory data collection. Focusing on the neighborhood unit is attractive and appropriate since it helps maximize public buy-in and provides information for managers; however, there are other

analysis. One notable exception is provided by Steenberg et al., (2013) who propose subdividing neighborhood tree inventory data into categories of land uses within

Components of a comprehensive tree inventory report Category

Example

Presentation of data

Tree distribution and diversity

Distribution of species, size and age

Text, tables, charts

Table 2: Common elements of a tree inventory report. Based on Nowak, et al., 2008

Deciduous vs. evergreen

Economic valuation

Estimated dollar value for annual environmental benefits (iTree)

Text, tables, charts

Estimated replacement value (iTree) Problematic species Current issues/Urban forest condition Stocking levels

Urban forest structure & condition Ratio of trees to available planting sites

Text, photographs, tables and maps Text, tables, maps

Sidewalk damage Utility conflicts

Future priorities

Large trees under primary electric lines Goals for planting, maintenance, and education/outreach

Text, tables

Text, lists, tables

27


Add a tree

Login or Sign up

Figure 8: Screenshot from San Francisco’s crowdsourced urban tree inventory project, a collaboration between government, nonprofit, and private sectors, which typifies a modern, citizenbased approach to urban forest assessment.

near

All trees

List

88,328 trees selected

List

Examples: Monterey Pine or Cupressus macrocarpa Examples: 210 Columbus Ave, North Beach, 94133 Show advanced filters

Search San Francisco

San Francisco, CA

Export options: KML CSV Shapefile

Add a tree!

Yearly Eco Impact

View Satellite

Selected trees in the region

Quick view

Total Benefits

$3,447,453 saved

From Urban Forest Map, 2014.

Greenhouse Gas Benefits

26,890,503 lbs CO2 reduced $537,810 saved Water Benefits

138,986,959 gallons conserved $555,947 saved

View all details Tree's profile page

Species name

Platanus acerifolia

Common name

London planetree

Edit details

Tree number

#104007

Tree's edit page

Nearby address

360

Trunk diameter

50.0

Last updated

July 11, 2010

Alerts

No Alerts

Yearly eco impact

$103.98

Energy Benefits

13,620,898 kWh conserved $1,802,044 saved Air Quality Benefits

-92,995 lbs pollutants reduced $551,651 saved

Map data ©2014 Google, a map Sanborn error (c)Report UrbanForestMap.org

Recent updates

units types ofpublicanalysis which could Our databaseor of trees comes from records and citizen foresters like you. Your help will make it better.

trees increases stakeholder involvement

provide important information to urban Neighborhood Tree species Updated by

in urban forestry at an appropriate and

forest plans and For example, Neighborhood Treemanagers. species Updated by

relevant scale.

Guadalupe palm

mdt

Southern magnolia

floraplusfauna

Tree species Southern magnolia

Updated by floraplusfauna

other studies have determined that Neighborhood

The growing desire and importance

land use, zoning, age of development,

for public involvement in urban natural

and other spatial variables are more

resource management, and specifically

significant in determining urban forest

urban forestry, has been well-summarized

characteristics

neighborhood

trees around you and (2007). update when The by Janse and the Konijnendijk

boundaries (Heynen & Lindsey, 2003;

first-hand experience of residents in

Steenberg, 2013).

assessment of their urban forest should

How you can help

1

Find a tree: Search for the trees near you. Find trees that bear edible fruits and nuts or those with beautiful flowers. Search for the biggest trees in your neighborhood -- then go visit them! Go »

The

28

Recent photos from users

than

neighborhood

2

Add a tree: The Urban Forest Map grows as citizen foresters like you add trees. Show a tree's location by putting a dot on the map, then provide as much information as you can. Go »

3

Edit a tree: Check out the facts about

you can. Don't forget to add alerts! With your help, we'll track changes in the urban forest and watch it grow. Go »

inventory

not be discounted – some of the largest

approach has the potential to promote

victories in urban forestry and urban

stewardship by residents, since it is based

natural resource management, such as

on existing sociopolitical boundaries and

the 1981 Million Trees Campaign in Los

organizations. Given the relatively high

Angeles, have resulted from public/private

cost of keeping municipal tree inventories

partnerships which were spearheaded by

up to date (Nowak, et al., 2008; Bond &

grass-roots efforts. A concrete example

Kotwicka, 2013), this method is also likely

is the documented tree canopy increase

to be the most cost-effective. Empowering

in Portland’s Southeast Neighborhoods,

residents and emphasizing the collective

partially attributed to tree planting efforts

ownership of a neighborhood’s street

by Friends of Trees, a community-based


Urban Land Environments

Acres

% of city land

Residential

50,000

57%

Commercial/Industrial/Institutional

20,000

23%

Natural Areas and Stream Corridors

14,500

16%

Transportation Corridors and ROW

8,700 (paved portion)

10%

Developed Parks and Open Spaces

4,000

5%

tree advocacy organization (Poracsky &

Rights-of-way’ ULE occurs within each

Lackner, 2004).

of the other four categories, since roads

Land Use and Urban Forest Management As early as 1978, urban foresters advocated for distinct management plans for the urban forest based on divisions between parks, natural areas, and street trees – even suggesting that street trees may require another level of division based on cover, species, age structure, and neighborhood characteristics (Byrne 1978, as cited in Miller, 1992). Rowntree’s (1984) analysis of canopy cover and land

Table 3: Urban Land Environments described in Portland’s 2004 Urban Forest Management Plan City of Portland, 2004

either transverse or border all of these environments. In Portland’s case, different city bureaus have jurisdiction over these ULEs, further justifying their management as discrete units of the urban forest. Gresham and Corvallis, Oregon’s urban forest management plans also propose implementation of land usespecific guidelines for appropriate tree species and prioritization of tree planting activities (City of Gresham, 2011; Duh & Flanagan, 2009).

use in four Eastern cities found dramatic differences in mean canopy cover across seven land use categories, with 1-2 family residential having the highest mean canopy cover (28.7%), and industrial having the lowest mean canopy cover (4.1%). Many urban forest management plans set different goals and actions based on land use. For example, Portland, Oregon’s 2004 Urban Forestry Management Plan (City of Portland, 2004) divides the urban forest into five Urban Land Environments (ULEs), stating, “Each ULE has particular physical characteristics and issues, provides various benefits, and services different needs.” See Table 3. It is noted in the plan that the ‘Transportation Corridors and

29


30


EUGENE’S URBAN FOREST

2


Eugene’s Urban Forest Eugene, Oregon is a Pacific Northwest city located at the south end of the Willamette Valley near the confluence of the McKenzie and Willamette rivers (See Figure 9). It is the second largest city in Oregon, with an estimated 2012 population of 157,986 (U.S. Census Bureau, 2013). Eugene’s founder settled the area in 1846, but it was not incorporated as a city until 1862. The city grew rapidly through the 20th century until the decline in the logging industry in the 1980s caused high unemployment. The economy recovered, and a 2010 report indicated top employers are in the education, healthcare, and civil services industries. The City’s 2010 land area was 43.72 square miles (U.S. Census Bureau, 2014), divided among 23 neighborhood associations. Eugene lies in the Marine West Coast climate zone, and experiences a Mediterranean climate, with cool, wet winters and warm, dry summers. Spring and fall are moist, with long periods of light precipitation. Winter snowfall is sporadic and rarely accumulates in large amounts. The average annual temperature is 52.1 F Figure 9: Eugene, Oregon lies at the southern edge of the Willamette Valley, approximately 100 miles south of Portland, OR. Google Earth, 2014

and average annual precipitation is 50.9 inches (NOAA, 2014).

Eugene’s Trees Though the Willamette valley supports a wide range of native and introduced tree species, 1850s land survey data from the General Land Office indicates much of the area around present-day Eugene was prairie and oak savanna – in other words, it was much less tree-dominated than it is today (Christy & Alverson, 2011). One of the most widely-accepted explanations for the open character of the historic landscape is the regular burning of the prairies by Native Americans, which shaped and maintained the prairie and savanna. Eugene’s favorable climate supports one of the widest ranges of trees species anywhere in the country - as noted in Trees of Greater Portland, “Between 150 and 175 different species of trees grow well in the Willamette Valley” (Reynolds and Dimon, 1993 p. 15). A 2014 tree canopy analysis by the author calculated tree canopy cover in Eugene at nearly 27%. Figure 10 allows comparison of total land area to the percent of city’s total tree and percent of the city’s total street trees per zoning category. As Figure 11 illustrates, most of the City’s tree canopy is over private property, and in the forested slopes of Eugene’s south hills. In contrast, Figure 12 shows that the highest density of street trees is surrounding the downtown area. Residential

zoning is

the most abundant in terms of land area

32


3% 9% 3% 5% 2%

tia l en

r

id

e th

City zoning and tree distribution

Pa

rk s

Re s

O

& I n du Pu s t b r C lic ia l om La Ag me nd ric rci ul a l tu ra l

79%

% of total canopy per zone

Percent of City area per zone 61% City area per zone

16%

3% 9% 3% 5% 2%

e th

O

Re Re s id s id en en tia tia l l

In du rk st s& ria l & Indu Pub Pu Cst lic b oria L a C lic m om LAa lme nd Ag me gnrdic rcia ric rci ult l ul al O ura tu ra th l l er

79%

% of total canopy per zone

Figure 10: City of Eugene’s zoning map and comparisons of land area, percent total tree canopy, and percent total street trees per zone.

10% 6% 5% 2%

Pa

r

rk s

Grogan, 2014

79%

% of total street trees % total canopy per perofzone zone City area per zone

78%

Pa

Percent of City’s total tree canopy per zone 61%

4% 7% 8% 0% 2% 3% 9% 3% 5% 2% 10% 6% 5% 2%

Pa

In rk P Pa dus s &ar rk tri kPs s I aIln ub& &nPd du C licPub uubsltr str om LCali icia ia C noc l l Ag moem dmLmanLan ricA rAcm ed d ulgri igarleicrc rcia tu cu uila l r O alltu tlur th ra al er l

r erhe th t OO

Re Re s id s id en R en tiaesi tia l de l nt ia l

16%

Percent of City’s total street trees per zone 78%

l tia

sid

en

sid Re

Re

er

th

78%

4% 7% 8% 0% 2%

Zoning Residential Parks & Public Land Commercial& Industrial Agricultural Other Not Zoned

Created 3/10/2014

zoning contributes the lowest percentage

and urban forest structure, particularly

l

tia

en

sid

er

of tree canopy relative to its land area.

th

s & In Pu dus b t C lic rial om La Ag me nd ric rci ul al tu ra l

relationships between neighborhood age

O

and tree canopy; conversely, industrial Re

street trees. She discovered distinct differences in the type and abundance

in Eugene is consistent with that of other

of various tree species in neighborhoods

studies: Private property, residential, and

developed during different time periods.

sloped areas often have greater tree canopy

Before 1910, most street trees were native

than other zoning or land use categories

species – predominately big leaf maple

(Wilson et al., 2003; Mincey et al, 2013).

(Acer macrophyllum); while the newest

rk

The relative distribution of tree canopy Pa

% of total street trees per zone

10% 6% 5% 2% 4% 7% 8% 0% 2%

O

en

tia

l

City area per zone % of total street trees per zone

16%

In P du Pa ark st rk s & ria s & InP l u d Pu ubsl bCl o tirci L C icm ala om Lm n A a d Ag mgrei nedrc ric rccuil ial ul atlur tuO a rath l l e r

61%

Eugene’s urban trees have long been

neighborhoods, developed between 1946-

viewed as a community asset. In the early

1978, were dominated by flowering plums

1900s, promotional magazines published

(Prunus cerasifera) (Butler, 1987).

by local merchants were sent to the mid-

Eugene has long been at the forefront of

west, describing Eugene’s wide streets as,

urban forestry planning and management.

“lined with stately oaks, pine, maple, and

In 1947, a governmental research bureau

ash trees.” (Anybody’s vol. 1 p. 13, as cited

at the University of Oregon published a

in Butler, 1987, p. 60)

guide titled, “Street Trees for Cities,” which

Sally Butler’s 1987 thesis, “A cultural

stated,

history of the American urban forest – using Eugene, Oregon as a case study,” used historic photographs, lithographs, and

field

observations

to

analyze

33


Tree Canopy Map

R RIVE

HW

Figure 11: Map of tree canopy by land ownership in Eugene, OR. Most of the canopy is over private property and in the south hills.

9N Y9

CAL YOUNG

I-5

D EL TA H WY

COBURG

BELTLINE HWY

Grogan, 2014

HARLOW ROOSEVELT

I-105

HIGH WILLAMETTE

JEFFERSON

18TH 24TH

30TH

2,040 acres

N

5,254 acres

0

“Although planting of street trees is common practice in many Oregon cities, full value has not generally been achieved from such planting because of the lack of a comprehensive community plan.� (As cited in City of Eugene, 1992, p. 3) The City of Eugene was the first city in Oregon, and perhaps in he country,

34

KLIN

ZO

Tree canopy over private property

FRAN

AMA

Tree canopy over public property

7TH 13TH

CHAMBERS

11TH

BERTELSEN

6TH

1

2

4

Miles

to create an integrated municipal urban forest management plan which addressed trees on both public and private property. Organized efforts to create the plan began in 1985, and it was published in 1992 (City of Eugene, 1992). In addition to responding to an increased desire for a coordinated approach to address both public and private trees, the plan also outlined proposed policies and actions to achieve its goals. One of


Street Tree Density Map Figure 12: Map of street tree density in Eugene. The highest density of street trees is surrounding the downtown area. Grogan, 2014

the proposed actions was, “Inventory all

called for a tree inventory and master plan.

trees and available planting spaces in street

Subsequently, the City conducted a partial

rights-of-way to determine composition

inventory (likely its first) and assumed

and planting needs,” (City of Eugene,

responsibility for street tree planting,

1992).

maintenance, and removal.

According to the 1992 management plan, the City produced a report titled “A Street tree Program for Eugene, Oregon,” in 1963, which recommended the City take responsibility for all planting and maintenance in street rights-of-way and

35


Eugene’s Street Tree Inventory

of total trees. For the inventory data to stay useful and accurate, the City needs to

The City of Eugene has conducted and maintained a GIS-based street tree inventory since 2005. Data collection has historically been conducted by both fulltime and seasonal City staff, and contains information on approximately 60,000 trees. The City is beginning to implement neighborhood inventory updates, based on the City of Portland’s neighborhood tree inventory program, and with support from

neighborhood

and

community

groups. These inventory updates will rely primarily on trained volunteers to collect the inventory data. (City of Eugene Parks and Open Space, 2013). The city’s first neighborhood inventory update was a collaboration between City staff, the Jefferson Westside Neighborhood Association, and the Eugene chapter of Friends of Trees, a communitybased tree advocacy organization. The Jefferson Westside Street Tree inventory report presents a detailed analysis of the neighborhood’s street trees. Table 4 provides a summary of the report’s findings. Figure 11, a screenshot from the interactive map that was created using the updated inventory data, shows how one can view information for each street tree in the neighborhood.

36

conduct regular inventory updates so that each tree’s data is never more than several years old. The data recorded by city staff for each tree includes species, size, condition, defects, and planter width, described below. Data is typically recorded either on paper or via personal digital assistant (PDA), then imported into the City’s GIS database.

Species Tree species is identified by data collector,

using

either

botanical

or

common name. If unsure, the data collector can choose to identify the tree at the genus level or to record the species as ‘unknown’ and move on to collecting the rest of the tree’s attributes. Often times, data collectors will enlist their colleagues for help in identification of a particular tree – by describing the tree or bringing a sample back to the office. Occasionally another staff member with more experience in identification will inspect the tree via Google Street View imagery, or return to the actual location of the tree to determine its species.

Size Three aspects of tree size are recorded.

All tree inventory data is connected

Diameter at breast height (DBH) is

to the city-wide GIS database through

measured at 4.5’ above ground level using

TreeWorks10, an extension for ArcGIS.

a DBH tape. The height for measuring

Though individual tree records are updated

diameter is estimated by data collector. Tree

to reflect tree maintenance work, these

height and canopy spread are also visually

updates only apply to a very small portion

estimated by data collector. Relative heights


Jefferson Westside Street Trees by the Numbers Number 3,403 $16.6 million $285,232

%

Metric Total trees in JWN rights-of-way Total replacement value Total annual environmental benefits

32

Notes

Page # 10 10 7, Appendix B &F

Calculated using iTree Streets Calculated using iTree Streets

171 10" 366 3233

11% 95%

Plant families represented Plant species and cultivars represented Average diameter at breast height (DBH) Trees greater than 24" DBH Broadleaf deciduous trees

123

4%

Tree/sidewalk conflicts

Tree causing damage to sidewalk infrastructure

14

3200

94%

Trees rated "fair" or "good" condition

Rated by volunteers on 4 point scale "Good, Fair, Poor, Dead"

12

3403 of 4012

85%

Stocking level

13

729

21%

Oversized trees

% of suitable planting sites that currently have trees Trees likely to outgrow their planter width

637

19%

Undersized trees

Trees smaller than planter width could have supported

13

Table 4: Summary findings of the Jefferson Westside Street Tree Inventory. From City of Eugene Parks and Open Space, 2014

8 8 9

13

Table 1: Key metrics of JWN Street Tree Inventory

Figure 13: Screenshot from interactive online map of street trees in the Jefferson Westside Neighborhood.

Top 10 Most Abundant Trees Rank

1 2 3 4 5 6 7 8 9 10

Common Name

Norway Maple Big-leaf Maple Red Maple American Sweet Gum Hawthorn Cherry Red Oak Raywood Ash Thundercloud Plum Green Ash Totals

Botanical Name

Acer platanoides Acer macrophyllum Acer rubrum Liquidambar styraciflua Crataegus spp. Prunus spp. Quercus rubra Fraxinus oxycarpa ‘Raywood’ Prunus cerasifera ‘Thundercloud’ Fraxinus pennsylvanica

Number of Trees

% of Total

355 273 201 158 137 128 104 89 77 75

10.4% 8.0% 5.9% 4.6% 4.0% 3.8% 3.1% 2.6% 2.3% 2.2%

1,597

50%

From City of Eugene Parks and Open Space, 2014

Table 2: Most abundant tree types in JWN

5

37


Figure 14: Structural defects recorded in Eugene’s street tree inventory.

Entire Decline Dieback-major Dieback-minor Lean

Crown Dead top Hanger >1” Deadwood > 4” Crown scar Crown cavity Crown decay Trunk Co-dominant stems Crack Included bark Split Trunk cavity Trunk decay Trunk scar Trunk seam

Basal Basal decay Basal scar Basal cavity Exposed roots Girdled

such as typical height of floors of a building or average height of utility lines are used to inform estimates.

of the tree. There are two options for tree

defects

are

presented as a checklist for each tree. The defects are grouped into three categories: structural defects, biotic defects, and cultural defects. The data collector checks off those defects which apply to the tree based on a visual inspection. No data entered is equivalent to “no defect” present. Figure 14 illustrates structural defect categories recorded in Eugene’s inventory.

38

Tree condition is judged in the field based on the overall health and vitality

Defects Pre-determined

Condition

recording tree condition, an ordinal scale of dead, poor, fair, good, and excellent; and a numeric scale from zero to one hundred. Treeworks software automatically ascribes a numeric value to each ordinal category, and vice versa. For example a tree recorded as “Good” would also be assigned a condition score of 70/100.


Planter width The width of the planting strip the tree is in is visually estimated or measured with a tape measure by the data collector. The width is the length of growing area between the edge of the street and the beginning of the sidewalk, measured perpendicular to the centerline of the street (illustrated in Figure 4, p. 22). If the tree is planted near a curb-tied sidewalk or the width of unimpervious area around the tree is greater than 20 feet, it is recorded as either “>20” or “open”. Tree pits, which are tree planting areas surrounded by concrete on all sides, and typically less than 4 feet wide, are recorded separately. In Eugene, most street trees are planted in strips between six to eight feet wide, and are mostly in residential areas.

39


40


RESEARCH DESIGN

3


Research Design Introduction

characteristics of the urban environment

This project explores how Eugene’s street tree inventory data could be analyzed to inform an understanding of tree health causes over time and to inform management decisions. For purposes of this paper, the term tree inventory refers exclusively to a ground-based inventory (for a comparison of urban forestry assessment methods Table 1 p.21). Eugene’s inventory contains information on over 55,000 street trees and is estimated to represent about 75% of the City’s total

population – and what types of trends are occurring at the city-scale. City GIS data on land use, age of development, and soils can be associated with individual tree records and provide means to perform statistical analyses. Existing tree inventory fields such as diameter at breast height (DBH), width of planting strip, species, and tree condition can then be analyzed in relation to urban environmental attributes. See Figure 4, p.22 for typical site

street tree population. Trees that are not

descriptors of concern to urban foresters.

included are mostly naturalized trees which

Simplified Research Process

happen to grow in the public right-of-way. Though these trees are technically street trees, they were not planted intentionally, and are not actively managed. They tend to occur in forested areas where trees naturally regenerate – unlike the flat lands upon which most of the city is built (See Figure 2, p.17).

The City’s first two neighborhood

inventory

reports,

for

Westside

Neighborhood

the

Jefferson

and

West

University Neighborhood, describe current conditions and suggest opportunities for future action. Though these reports are significant milestones in the city’s urban forestry program, looking at the street tree population as neighborhood units should not be the only scale or type of analysis, as discussed in Chapter 2. Eugene’s tree inventory data presents an opportunity to test if, and how,

42

are having an influence on the street tree

This project proposes a method of combining Eugene’s street tree inventory data with GIS data, and analyzing the relationships

between

them,

as

an

additional means to quantify and describe the street tree population. It provides results which can inform urban forest management in Eugene, and also a method for analyzing street tree inventory data in other cities. The following is a conceptual description of the research process. A more detailed description which discusses statistical methods employed in this project is provided in the next section. The method is a simple, threestep process summarized in Figure 15 and described in detail in Table 5. In step 1, variables that were statistically independent were separated from and those they tend to predict, the dependent “response” outcomes. These variables were


Simplified Research Process

3

Species (iTree size code)

steps

1. Identify variables

Planter width

DO THESE FACTORS Independent Variables

2. Statistical analysis of tree size & condition

Figure 15: Simplified diagram of research process presented in this study.

Soils (soils shrink/swell) Age of development (year annexed) Land use Zoning

3. Descriptive analysis of tree defects

PREDICT THESE? Dependent Variables

Size (DBH) Condition Defects

Detailed Research Process Step 1 - Identification & selection of variables Considerations Identify tree and site descriptors

Urban forestry best practices and management plans

Determine predictor and response variables

Local expertise and urban forestry literature

Identify available/appropriate data source

Easily accessible (GIS). Quality and consistency (inventory data).

Prepare data for analysis

Spatial join GIS data, simplify categories. Verify and/or modify data type (categorical vs. continuous) for analysis type.

Step 2- Statistical analysis

Results

Outcome

Descriptive statistics

Frequency tables and charts, means plots

Reveal character of the data

One-way ANOVA

Means, standard deviation, f value, significance level (p value)

Tests for significant variation of mean response variables within a predictor variable set.

Means, standard deviations for each class of a predictor variable

Identifies which categories within a predictor set are significantly different from each other and which are not

R2 value, estimated and standardized coefficients, and significance level.

Identifies the percentage of variation in response variable that can be explained by each predictor variable.

Summary tables and charts

Easily produce graphic and tabular summary of frequency or percentage of a response variable occuring within categories of predictor variables.

If significance, post-hoc tests

Regression analysis

Table 5: Detailed description of research process presented in this study.

Step 3 - Descriptive analysis*

Microsoft pivot tables and charts

*Descriptive analysis can be conducted independently of other steps.

43


Table 6: The identification and selection of variables in Step 1 of the research process.

Identification and selection of variables for this study Predictor (independent) variables

Data selected for analysis

Data type

Species

iTree size code

Categorical, translated to continuous

Planter width

Planter width

Categorical, translated to continuous

Soils

Soil shrink/swell capacity

Categorical

Age of development

Annexation date (year)

Continuous

Land use

County land use classification (simplified)

Categorical

Zoning

City zoning classification (simplified)

Categorical

Size

Diameter at breast height (DBH)

continuous

Condition

Condition rating

continuous

Defects

Structural defects

categorical

Response (dependent) variables

selected based on local expertise and urban

variables. Sometimes variables can take on

forestry literature. Size codes from iTree, a

both roles across different analyses.

USDA urban forest assessment tool, were

Then, predictor variables of significant

assigned to each tree to help account for

interest can be used in a multiple regression

species variation in health. Soil shrink/

analysis to test their ability to predict

swell capacity was used to roughly capture

variability in a given response variable.

variation in soil qualities. Several land use

Regression analysis requires numeric data,

and zoning categories were combined to

so categorical data will need to be modified

simplify analysis.

before being employed in regression

The statistical analyses employed in

analyses, or analyzed using other methods.

step 2 essentially seek to answer whether

Categorical response variables, which

the selected independent variables shown

may not be easily converted to numeric

in Table 6 reliably predict the response

data, such as tree defects, are analyzed

variables, and if so – how much?

using descriptive methods. These results

The third step, descriptive analysis,

illustrate the relative frequency that a

focused on the distribution and abundance

categorical response variable occurs within

of tree defects by comparing the relative

each category of a predictor variable (e.g.,

frequency

percentage of total defects that are crown

of

certain

defects

within

different independent variable categories.

Detailed Research Process As variables are selected in step 1, they must be identified as either predictor (independent) or response (dependent)

44

defects within each zoning classification). Many street tree inventories include some site information, but site descriptors such as land use and age of development are rarely included, and will likely require access to a municipal GIS database. GIS-


Simplification of land use categories Original land use categories Single Family Duplex Agriculture Timber Educational Religious Government General Services Group Quarters Multi-family residential Industrial

Table 7: Original land use categories and simplified categories used in this analysis.

Simplified land use categories Single Family Residential & Duplex Agriculture & Timber Education & Religious Government & General Services Group and Multi-family residential Industrial

Communication Alleys, Walkways, Bikepaths Railroad

Infrastructure

Utilities Mobile Home Park Mobile Homes on Lots Recreation Parks Retail Wholesale

Mobile Home Parks & Recreation Retail

Vacant

Vacant

Water

Water

based tree inventory records are then spatially joined with the desired GIS layers from the database. Some modification of variables, as described later, may be necessary to improve the interpretability of results or facilitate statistical analysis. Before conducting statistical analysis, it is useful to get a quick visual survey of the data set by looking at frequency tables and charts for each variable. This process helps develop an understanding of the general distribution of the data, and to spot any interesting or suspicious trends.

Methods employed in this analysis Part 1: identification selection of variables Some

zoning

and

land

and use

classifications were combined to reduce the number of categories. See Table 7 for original and simplified land use classifications employed in this analysis. Though this study did not analyze tree species, it attempts to account for the influence of tree species on response variables by ascribing size codes to each species. These size codes, specific to the Pacific Northwest Region, are used by iTree Streets for quantification of street

45


tree benefits, and include tree type

were not analyzed by ANOVA or multiple

(broadleaf or coniferous), habit (deciduous

regression, instead they were analyzed

or evergreen), and size (small, medium, or

using descriptive methods in part 3 of the

large). Rather than speculating, tree species

research process.

which were not listed in iTree’s list (12,116 or 23.7% of the inventoried trees) were omitted from the analysis. Due to the very small number of broadleaf evergreen trees in the population, these too were omitted (168 or 0.3% of the inventoried trees). The

planter

width

variable

was

converted from a categorical variable to an interval. All width categories were averaged (E.g., 4-5’ = 4.5 ; 5-6’ = 5.5) with the exception of ‘less than 4 feet’ which was estimated at 3 feet wide, and greater than 20’, which was substituted with the estimated diameter of a tree’s root zone. For these widths, trees less than 8” in DBH were calculated using a formula to estimate root spread based on DBH that was reported by Day & Wiseman (2009): The radius of a tree’s root spread averages about 39 times its DBH. For trees in these widths greater than 8” DBH, the diameter of root spread was estimated at 20 meters (65.6 feet), an estimate by this author, based on a conservative interpretation of the maximum extent of tree roots reported by Stone & Kalisz (1991). It should be noted that only 13% of the trees inventoried occur in planter widths greater than 20 feet. Tree defect categories were simplified to reflect the relative location (basal, trunk, crown) and extent (dead tree) of the injury (See Figure 14, p. 38). Due to the complexity of statistical analysis of categorical response variables, tree defects

46

Part 2: statistical analysis Frequency charts for all predictor and response variables being analyzed were generated to visually describe the data and to check for any anomalies or suspected errors, such as dramatic outliers. These charts did not reveal anything suspicious, and all identified variables were included in the statistical test. For frequency charts, see Appendix A. In

order

to

determine

whether

predictor variables have a statistically significant (p<.05) effect on response variables, a one-way ANOVA test was conducted for both DBH and Condition to determine whether mean values varied significantly across some data categories within classes of each predictor variable type illustrated in Table 6. Since the defect category was removed from this portion of the analysis, the only response variables tested were DBH and Condition. Means plots and frequency tables and graphics illustrated the distribution of mean response variable values per predictor class. When the one-way ANOVA produced statistically significant (p<.05) results, statistically significant differences between categories of the predictor variable were identified using post-hoc tests (Tukey’s HSD, Lane, 2010). Means plot graphs facilitated

identification

of

statistical

significance (p<.05) by inspection of 95%


confidence interval overlaps. Results of

of correlation - the highest correlation

the post-hoc test were judged for their

between two variables was 0.36, between

statistical and practical significance.

the deciduous/evergreen size code and

Finally,

predictor

variables

with

the small, medium, large size code. For

a statistically significant (p<.05) and

selected results of correlation analysis, see

practically significant effect on DBH were

Appendix B.

tested across many forms of a multiple regression model, and the best combination was selected and reported. The most significant classes within zoning

classification

and

land

use

categories were selected for inclusion in the reported regression model as “dummy� or indicator variables so they could be tested as binary (0 or 1 value) predictors (E.g., testing residential zoning or nonresidential zoning by including it as the only zoning variable). This allowed the regression analysis to test the ability of an individual category of a predictor variable to explain variance in the response variable. For example, how much of variation in DBH can be explained by residential zoning alone? The multiple regression model was adjusted by increasing or decreasing the

number

of

predictive

variables

Part 3: descriptive Analysis As mentioned previously, ANOVA or regression analyses were not employed to predict defects due to the complexity in analyzing effects of categorical and continuous variables on a categorical response variable. Instead, the distribution and abundance of structural defects across the same predictor variables were analyzed using descriptive tables and crosstabulation. Tables for the most explanatory independent variables were reported and contained predictor classes on one axis and the type of structural defect on the other axis. Since the spatial extent and number of trees in each predictor category varied widely, a defect type’s percent of total defects within each category was chosen. The results were compared using 100% bar graphs (see Findings).

until a balance was judged to select the highest predictive ability (R2 value) and interpretability of results, without appreciably reducing the F value or statistical significance of the overall model, while avoiding multicollinearity errors. Since multicollinearity violates the assumptions of regression analysis, a separate analysis of independent variables used in the regression model was employed to confirm they did not have a high degree

47


48


FINDINGS

4


Findings Results of One-Way ANOVA For diameter at breast height (DBH), all predictor variable sets produced statistically significant results (<.01) using the one-way ANOVA test, meaning that each predictor variable set contained a minimum of one significant categorical difference between at least one pair of predictor categories (Table 8). Posthoc tests (Tukey’s HSD) revealed which categories within each predictor variable set were significantly different from each at predicting DBH (Lane, 2010). In the case of ‘Soil shrink/swell’, the post-hoc test revealed that the significant difference (high shrink/swell differed from low and medium shrink swell) tended to forecast only about 1” in diameter. This predicted difference has little practical significance so this variable was removed from subsequent analyses. Significant predicted differences in condition class were similarly impractical (differing by only 1 point on a 100 point scale) and also omitted from further analysis.

50


Results Predicting DBH for five Univariate ANOVAs

Table 8: Results of One-way ANOVA for predictor variable sets.

Predictor Variable Age of Development

Type III Sum of Squares: 210,880.31

F 3,568.72

r2 = 0.067

p <.01

Planter width*

Type III Sum of Squares: 164,290.68

F 2,903.38

r2 = 0.059

p <.01

F 122.18

p <.01

*Categorical planter width data was converted to numeric interval (see methods)

Zoning Classification

Type III Sum of Squares: 46,800.38 Data categories:

N

Agriculture

Mean DBH*

SD

221

4.81

4.53

1,859

5.71

5.94

40,853

8.53

7.99

Natural Resource & Public

1,944

9.30

8.04

Other

1,607

10.57

9.39

Commercial

4,365

10.78

8.32

F 53.72

p <.01

Industrial Residential

*Data categories listed by ascending mean values

Land Use Classification

Type III Sum of Squares 41,231.28 Data categories:

N

Agriculture and Timber

Mean DBH*

SD

44

3.66

3.21

395

5.16

6.50

2,076

6.72

7.12

Water

79

6.81

9.19

Mobile

858

7.17

4.96

Industrial

445

8.10

8.28

35,365

8.50

8.00

Educational and Religious

1,452

9.01

7.84

Parks & Rec

1,504

9.20

8.62

Retail

2,200

9.99

7.80

Government and General Service

2,755

10.33

8.21

Group Housing and Multi-Family Residential

3,425

10.50

9.02

F 27.42

p <.01

Infrastructure Vacant

Single Family Residence and Duplex

*Data categories listed by ascending mean values

Soils

Type III Sum of Squares: 5,318.97 Data categories:

N

Mean DBH*

SD

High shrink/swell

5,732

7.83

6.65

Low shrink/swell

12,340

8.74

7.90

Moderate shrink/swell

32,683

8.84

8.32

*Data categories listed by ascending mean values

51


12

Mean diameter (DBH) inches

Figure 16: Mean DBH values, 95% confidence intervals, and sample size for each zoning category.

10.78

10.57 10

9.30 8.53 8

6

5.71 4.81

4

2

0

Agriculture

Industrial

N = 221

1,859

Residential Natural Resource 40,853 & Public 1,944

Error Bars: 95% CI

Other

Commercial

1,607

4,365

Zone_name

Homogeneous Subsets of DBH by Zoning Classification Table 9: Results of Post-hoc tests for = 44 zoning.NMeans in the same column are not significantly (p<.05) different from each other.

Tukey’s HSD Subset

Zoning Classification* Agriculture Industrial Residential

1

N

2

221

4.81

1859

5.71

3

40853

8.53

Natural Resource & Public

1944

9.30

Other

1607

10.57

Commercial

4365

10.78

p-value

0.18

0.88

*Data categories listed by ascending mean values

The group sizes are unequal. The harmonic mean of the group sizes is 320.547.

52

9.30

0.22


Zoning classification In Figure 16, mean DBH values, the number of trees in each DBH class, and

duplex residential – in other words it is a seemingly unintelligible mix of land use classes.

error bars representing a 95% confidence

Residential zoning’s mean DBH, 8.53

interval around the mean are shown

inches, is significantly different from

for each zoning category. For a diagram

every other zoning class except “Natural

explaining how to interpret this means

Resource & Public”. The very large sample

plot, see Appendix C. The difference

size for “Residential”, 40,853 trees, is likely

between the highest and lowest mean

due to this zoning class making up 61% of

values is nearly six inches, a fairly dramatic

the City’s total area, and to the relatively

difference considering these are all street

higher density of street trees in residential

trees (growing within the public right-of-

zones compared to other zoning classes

way adjacent to city streets). The lowest

(See Figures 10, 11, & 12 pp. 33-35).

mean DBH occurs in both agricultural

The Natural Resource and Public

and industrial zoning classes. The means

Land zoning class contains parks, schools,

from these two zoning categories are not

natural areas, and other publicly-owned

significantly different from one another, but

lands. It makes sense that these street

are significantly different from every other

trees would have a relatively high DBH

zoning class. An analysis of Agricultural

since these lands also have the highest

zoning by land use indicates that it is

percentage of large planter widths for trees

93% residential, which is consistent with

(21.8% are large, next highest is Residential

residential development along the edge of

with 14.2%).

the urban growth boundary. Table 9 shows

One potential explanation for the high

“homogenous subsets” of zoning classes,

mean DBH in both “Commercial” and

where subsets of means that do not differ

“Other” categories is they share the two

from each other at p < .05 (according to the

oldest ages of development, so they can be

Tukey multiple comparison procedure)

expected to have older trees (Tables 11 &

are grouped into the same column, while

12).

those that do differ significantly go into separate columns. Commercial zoning has the highest mean diameter at 10.78 inches, and is significantly different from all other classes except “Other”. A comparison of zoning classifications with land use classifications reveals that the “Other” zoning category is 24% government and general service, 12.4% retail, and 35% single family and

53


Land use Compared with Zoning, the Land Use

with the highest mean DBH also have an

categories showed much more overlap

earlier mean year of annexation : Group

between the 95% confidence interval for

and multifamily residential has a mean

mean DBH across categories (See Figure 17

DBH of 10.5 inches and a mean annexation

& Table 10). This indicates that only a few

year of 1925; Government and general

land use categories exhibit clearly different

service has a mean DBH of 10.3 inches

mean DBH values. The “Agriculture &

and a mean annexation year of 1910; Retail

Timber” and “Infrastructure” categories

has a mean DBH of 10 inches and a mean

had the lowest mean DBH, and this pair

annexation year of 1920. See Table 12.

of classes was significantly different from all others except for “Vacant,” “Water,” and “Mobile Home.” Table 10 shows “homogenous subsets” of zoning classes, where subsets of means that do not differ from each other at p<.05 (according to the Tukey multiple comparison procedure) are grouped into the same column, while those that do significantly differ are shown in separate columns. Another land use category which

Age of development As expected based on observed relationships between mean DBH and age of development in zoning and land use categories, a scatterplot indicates a statistically significant negative trend – more recently annexed portions of the City tend to have a lower mean DBH (See Figure 18).

Other predictive variables

stands out is “Mobile Home,” with

The two other predictive variables

a significantly low mean DBH. It is

identified in (Figure 15, p. 43) were

noteworthy that this land use, with lower

species size codes and planter width. One-

household incomes and improvement

way ANOVA and comparison of means

values than other residential categories,

confirmed that general tree species size

has a lower mean DBH than the other two

categories (derived from iTree’s species

residential classes in the analysis – “Single

list for the Pacific Northwest) produced

& Duplex Residential” and “Multi Family

significant differences corresponding to

Residential”. Compared to these other

relative differences in mean DBH as one

residential land uses, mobile homes are

might expect. One-way ANOVA and also

typically on very small lots with little room

produced a significant positive correlation

for trees and historically did not have street

between DBH and the planter width

tree planting requirements.

variable, with larger planter widths having

Similarly to the zoning classification

a significantly higher mean DBH. These

data, the relationship between mean DBH

variables were included in the multiple

and land use are likely heavily influenced by

regression analysis described next.

age of development. Mean annexation year

54

for land use shows that those categories


Mean diameter (DBH) inches

12

10.33

9.99

10

8.10

8

6.78

6.72

9.20

9.01

8.50

10.50

Figure 17: Mean DBH values, 95% confidence intervals, and sample size for each land use category.

7.17

6

5.16 4

3.66

2

0

Infrastructure Agriculture 395 Vacant & Timber 2,076

Water

Industrial

79

445

Mobile Home 858

N = 44

Retail

Ed. & Religious

Single & Duplex Residential

1,452

35,365

Parks & Recreation

Group/Multi Govt. & Residential 3,425 Gnrlsvc

2,200

1,504

2,755

Error Bars: 95% CI

LandUse_SIMP

Homogeneous of Subsets of DBH by Land Use Tukey’s HSD Subset

Land Use Classification* Agriculture and Timber Infrastructure Vacant

N

1 44

3.66

395

5.16

2

3

4

5

6

7

5.16

2,076

6.72

6.72

79

6.81

6.81

6.81

Mobile Home

858

7.17

7.17

7.17

7.17

Industrial

445

8.10

8.10

8.10

8.10

35,365

8.50

8.50

8.50

8.50

8.50

9.01

9.01

9.01

9.01

9.20

9.20

9.20

Water

Single Family Residence and Duplex Educational and Religious

1,452

Parks & Rec

1,504

Retail

2,200

9.99

9.99

Government and General Service

2,755

10.33

10.33

Group Housing and Multi-Family Residential

3,425

p-value

Table 10: Results of Post-hoc tests for land use. Means in the same column are not significantly (p<.05) different from each other.

10.50 .59

.05

.30

.07

.13

.06

.15

*Data categories listed by ascending mean values

The group sizes are unequal. The harmonic mean of the group sizes is 273.872.

55


R2 Linear = 0.067

Figure 18: Scatterplot showing relationship between DBH and development age.

120

diameter (DBH)

100

80

60

40

20

0

1850

1900

1950

2000

2050

Annexation Year

Mean DBH and Age of development Mean DBH and Age of by Zoning Category Development by Land Use Zoning

Mean DBH*

Mean age of development

Agriculture

4.81

1982

Industrial

5.71

1964

Residential

8.53

1956

Natural Resource & Public

9.3

1936

Other

10.57

1906

Commercial

10.75

1909

*Data categories listed by ascending mean values

Tables 11 & 12: Mean DBH and mean development age for zoning and land use categories. Smaller mean DBH values tend to have a younger mean age of development, and vice versa.

Mean DBH

Mean age of development

Agriculture & Timber

3.66

1975

Infrastructure

5.16

1945

Vacant

6.72

1968

Water

6.81

1971

Mobile

7.17

1970

Industrial

8.1

1956

Single Family Residential & Duplex

8.5

1956

Education & Religious

9.01

1931

Parks & Recreation

9.2

1945

Retail

9.99

1922

Government & General Services

10.33

1910.25

Group & Multi Family Residential

10.5

1925

*Data categories listed by ascending mean values

56


Results of Regression Analysis

more than twice as much variation in

Species size code, which was separated

DBH than whether it is classified as small,

into two independent variables, was the most powerful individual predictor of DBH, explaining about 10 percent of the variation in DBH (R2 = 0.104, See Table 13) and was statistically significant. Evergreen trees have a predicted DBH more than 6.5 inches greater than deciduous trees. For each increase in the ordinal size code, small, medium, or large, predicted DBH increases by about 1.4 inches. The standardized coefficient, Beta, indicates that whether a tree is deciduous or evergreen explains

medium, or large. The negative coefficient for age of development indicates that for each increase in year, i.e., a “younger� development age, predicted DBH decreases by a little less than 1/16 of an inch. This variable also explains almost 7 percent of the variation in DBH with statistical significance (r2 = 0.067, See Table 14). For each one foot increase in planter width, predicted DBH increases by a little more than 1/8 inch. Planter width explains

Multiple Regression Analysis for DBH Predictor Variable

Estimated coefficient

Standard error

Beta

t

Species Size Code Decidous = 0, Evergreen = 1

6.664

0.264

0.129

51.494*

Species Size Codea Small, medium, large= 1,2,3

1.345

0.116

0.059

22.688*

Constant

5.782

a

Regression statistics:

R2 = 0.104

adjusted R2 =0 .104

Table 13: Results of multiple regression for DBH variance explained by size code

F= 2,244.783

*Statistically significant at p<.01

Simple Regression Analysis for DBH Predictor Variable Age of Development Constant Regression statistics:

Estimated coefficient

Standard error

Beta

-.056

-0.258

t

0.001 -59.739*

117.137 r2 = 0.067

adjusted r2 = 0.067

Table 14: Results of multiple regression for DBH variance explained by age of development

F= 3,568.720

*Statistically significant at p<.01

Simple Regression Analysis for DBH Predictor Variable

Estimated coefficient

Planter width

0.142

Constant

6.784

Regression statistics:

r2 = 0.059

Standard error

Beta 0.242

adjusted r2 = 0.059

0.003

t 53.883*

Table 15: Results of multiple regression for DBH variance explained by planter width

F= 2,903.376

*Statistically significant at p<.01

57


nearly 6 percent of the variation in DBH

attention to species selection and planter

and is statistically significant (r2 = .059, See

width.

Table 15).

Table 16: Results of multiple regression for DBH variance explained by age of development, size code, and planter width

Subcategories of zoning and land

The most powerful predictors for

use predictors each explained less than

DBH, in descending order, were Species

1 percent of the variance in DBH - but

Size Code, Age of Development, and

were statistically significant. Estimated

Planter width. Together, these variables

coefficients for both of these predictors

explain about 20 percent of the variance

reiterate findings from the means plots

in DBH with statistical significance (R2

discussed previously, and are statistically

= 0.207, See Table 16). This suggests

significant. However, they explain very

that managers who want trees to reach

little variance in DBH (Zoning R2 = 0.012,

their full potential should pay particular

Land use R2 = 0.009) Tables 17 & 18.

Multiple Regression Analysis for DBH Predictor Variable Age of Development

Estimated coefficient

Standard error

Beta

t

-0.067

-0.312

Species Size Code Decidous = 0, Evergreen = 1

5.677

0.215

0.139

40.834*

Planter width sizeb

0.098

0.171

0.003

34.348*

Species Size Code Small, medium, large= 1,2,3

1.116

0.100

0.056

19.749*

a, b

a

Constant Regression statistics

0.001 -64.860*

135.082 R2 =0 .207

adjusted R2 = 0.207

F= 2,255.543

*Statistically significant at p<.01

Table 17: Results of multiple regression for DBH variance explained by three individual zoning categories

Multiple Regression Analysis for DBH Predictor Variable

Standard error

Beta

t

Is zoning Residential?

-1.053

-0.052

.136

Is zoning Industrial?

-3.870

-0.090

.227 -17.071*

Is zoning Commercial?

1.202

0.042

Constant

9.582

Regression statistics *Statistically significant at p<.01

58

Estimated coefficient

R2 = .012

adjusted R2 = .012

-7.736*

.178

F= 206.758

6.762*


Multiple Regression Analysis for DBH Predictor Variable

Estimated coefficient

Standard error

Beta

t

Is land use Single Family Residenial and Duplex?

-1.256

-0.071

.084 -14.977*

Is land use Vacant?

-3.040

-0.075

.190 -15.981*

Is land use Infrastructure?

-4.597

-0.050

.410 -11.217*

Constant Regression statistics

Table 18: Results of multiple regression for DBH variance explained by three individual land use categories

9.757 R2 = .009

adjusted R2 = .009

F= 146.622

*Statistically significant at p<.01

59


Results of descriptive analysis

patterns in the city may hinder application

Descriptive analysis of structural

of this information to maintenance and

defects was employed to examine the relative frequency of various defect types within categories of predictor variable types. Since the number of trees in each predictor category varies dramatically, defect types were summarized by their percentage of the total defects for that class, using comparative tables and 100% bar graphs. Analysis

of

defect

types

across

zoning categories revealed some dramatic differences in their frequency, especially in industrial zoning, where basal defects make up over 20% of total defects (including ‘None’). The next highest category for basal defects is in commercial zoning, with about 6%. Across all the zoning categories except ‘Other’, the highest percentage of structural defects is ‘None’ (for ‘Other’ trunk defects are most abundant). Generally the second most frequent structural defect type across categories is ‘Trunk’ defects (Figure 19). Structural defects in land use categories similarly

indicated

fairly

dramatic

differences in the relative frequency of defect types in each category. Like the industrial zoning category, industrial land use stood out as having the highest incidence of basal defects, 25%, with the next highest rate of basal defects occurring in the ‘Vacant’ category, 11%. Another similarity between zoning and land use defect frequency is that ‘None’ and ‘Trunk’ made up the vast majority of defects across all categories (Figure 20). The complexity and distribution of land use

60

management activities. The frequency of structural defect types across planter width revealed that basal defects, while a relatively small percentage of total defects, decreased as planter width increased (See Figure 21). Defects which affect the entire tree, which include tree decline, dieback, and lean, generally increase as planter width increases, with the exception of planter widths less than 4 feet. Other defects appear at similar levels across the planter width categories. These results are intuitive, assuming that trees in larger planter widths are more likely to mature, while smaller planter widths make trees more vulnerable to damage. Analysis of structural defects by the species proxy, species size code, revealed that coniferous evergreen trees generally had fewer structural defects than broadleaf deciduous trees. Structural defects were higher in small broadleaf deciduous trees, and were mostly trunk defects (Figure 22). Though it is likely that small deciduous species have the most defects because they are often chosen for difficult sites (small planter width, overhead utilities, etc.), the relatively few defects in evergreen trees raises some questions. Is the lack of defects related to their excurrent growth pattern, and/ or data collectors having more difficulty spotting defects through evergreen foliage? Further analysis is needed to determine if coniferous evergreens actually do have fewer structural defects – and if this is


Structural defect type by zoning 2% Crown Entire

1% 10% 5%

1%

6%

1% 10%

11%

Figure 19: Structural 4% defect types by zoning 8% Crown categories

1% 12%

14%

Entire

21%

Trunk

41%

21%

None

12% 6%

Trunk

38%

Basal

5%

39%

Basal

49%

None 3% 6%

83%

88%

6% 3%

52%

Agriculture

% 12%

Industrial

dead

0%

0%

Crown

0%

1%

Entire

2%

Trunk

10% 4% 8% 5%

6% 1% 21%

Basal Crown none Entire

83%

5% 12% 6%

Trunk 49%

49%

Residential

Parks & Nat’l Resource

0%

0%

9%

38% 4% 3%

18%

49% 31%

1%

9%

1% 41% 6%

None

%

9%

Agriculture & Timber

Water

Agriculture & Timber

Figure 20: Structural defect types by land use categories

0%

69%

1% 39% 13% 6%

1% 49% 5% 13% 3% 35%

58%

55%

Water

dead Crown Entire

1%

1% 12%

Trunk 14%

30%

Basal Crown none Entire

Trunk 37%

39%

39%

11%

Basal

40%

None 25%

49%

35%

Other

Other

39%

3% 77%

Commercial

24% 39%

5% 88%

35%

0%

42%

4%

Basal

39%

1% 0% 1% 1% Structural defect type14% by land 12% use 10% 11%

21% 52%

42%

77%

8%

6%

42%

42%

Infrastructure Vacant Multi Residential Mobile home Retail Single & Duplex Residential

40%

Industrial

8%

6%

39%

39%

Government Education & Religious

true, it would strengthen the argument 0%

0% along 0%city for planting more ofdead these trees

1%

streets.

0%

0%

0%

0%

0%

0%

0%

0%

Crown

0%

0%

1%

0%

1%

1%

1%

0%

1%

1%

1%

Entire

4%

5%

9%

4%

9%

9%

13%

13%

5%

12%

14%

49%

Trunk

8%

12%

18%

31%

24%

39%

37%

39%

30%

39%

40%

3%

Basal

0%

6%

4%

5%

11%

3%

8%

6%

25%

8%

6%

35%

none

88%

77%

69%

58%

55%

49%

42%

42%

40%

39%

39%

12%

d

0%

C

E

T

B

n

61


Structural defect type by planter width

by planter width Figure 21: Structural 1%

defect types by land use classification. 13%

1% Crown Entire

8%

39%

36%

39%

6%

5%

47%

49%

47%

45%

4-5 feet

6 to 9 feet

10 to 19 feet

>20 feet

13%

45% 41%

Basal None

2% 60% 45%

41%

median

<4 feet

40%

2%

6%

Large

1% 11%

Trunk 40%

1%

1% 10%

11%

2%

0924911

0.002538071

dead

0%

0%

0%

0%

0%

0%

5926281

0.010659898

Entire

0%

10%

8%

11%

13%

11%

Crown 1% Trunk Basal14%

0%

1%

1%

1%

1%

0%

40%

36% 16% 5%

39% 7% 6%

40% 2%

41% Crown 6%

49%

47%

45%

41% Entire

use

Figure 0.126565144 22: Structural 1% defect 1%types by iTree 0.398138748 12% 14% size codes. 0.015397631

074678 1% 7189641 5%

1963552

3248835 30%

0.446700508

Structural defect type by iTree size codes

Crown Entire Trunk

39%

40%

Basal

none 14%

0% 22% 60% 18%

1%

None 6%

2%

39%

Industrial

0%

0%

%

1%

1%

1%

3%

5%

12%

14%

9%

30%

39%

40%

%

25%

8%

6%

2%

40%

39%

39%

47%

Trunk 34%

7%

40%

Basal None 6%

5%

51%

47%

2%

47%

46% 27%

Government Education & Religious

0%

62

7%

39%

%

10%

69% 59%

40%

40%

39% 15%2%

61%

25% 8%

30%

1%

Structura

dead Crown

Large Coniferous Evergreen

Small Coniferous Evergreen

Medium Coniferous Evergreen

0.001729107

0

0

0.000378812 0.000798722 0.001342883

0

0.004735147 0.013578275 0.001939719

0.0129683 0.003816794

Medium Broadleaf Deciduous

Large Broadleaf Deciduous

Small

Small Broadleaf Deciduous

Entire 0.142651297 0.217557252 0.163972286 0.072416188 0.145899894 0.09773202 Trunk 0.136887608 0.183206107 0.307159353 0.390113012 0.336395101 0.610564011 Basal

0.011239193 0.003816794 0.020785219 0.066860282 0.048189563 0.021933751

none

0.694524496 0.591603053 0.508083141 0.465496559 0.455138445 0.266487616

dead

0.00068

Crown

0.00

Entire

0.078

Trunk

0.39170

Basal

0.05647

none

0.46738


DISCUSSION

5


Discussion Despite

the

growing

body

of

“Land use should be determined by the inventory team based on impressions out in the field (i.e., not from land use maps). This field describes how the land is being used, which is not necessarily the same as the ownership of the land”

municipal tree inventory data, it has not been widely examined for the effects of site descriptors such as land use, zoning, or age of development, even though analysis of urban tree canopy data has incorporated these variables for decades (Rowntree, 1984; Wilson et al., 2003; Mincey, 2013). The objective of this project was to demonstrate methods and results of analysis of Eugene’s street tree data that incorporated these under utilized variables, and to suggest their management implications. As urban forest managers increasingly recognize and employ land use and other site descriptors in creation of urban forest management plans, use of tree inventory data should be analyzed within these categories, instead of, or in addition to, the simple jurisdictional (city) and sociopolitical (neighborhood) boundaries which have become the norm (Steenberg et al., 2013). It should be noted that i-Tree Eco requires collection of land use data within 13 pre-determined categories for its sample inventories. iTree Eco has determined these categories based on their recognized influence on urban forest attributes. The importance of on-the-ground observations of land use is described in the iTree Eco manual,

(i-Tree Eco Manual, N.D., p. 33). The software can generate reports that provide estimated tree benefits,

DBH

distribution, and condition ratings, and other variables for each land use category. Age of development may be difficult to judge with accuracy in the field, and this category is not captured in i-Tree Eco. Eugene’s street tree inventory data and site information from the City’s GIS database suggest street tree data may be better analyzed and understood using site predictors such as those used in this study, particularly age of development. When

seeking

relationships

between

characteristics of the urban environment and tree attributes, it will be important to account for the influence of soil volume and size characteristics of different species. This study has demonstrated the use of a fairly simple approach to account for this variability by simplifying tree species into general size codes, using iTree’s classification system as a resource, and using planter width as a proxy for rooting area. Interpretation of this study must be tempered by the omission of street tree density from the analysis. This metric

64


was omitted because it could not be easily

this is that we are seeing the maturation

associated with each individual tree record,

and decline of large trees planted during

unlike the variables that were analyzed.

the City Beautiful era (described in the

Similarly, species abundance and diversity

history portion of this paper) - trees that

were not investigated. This study should

benefited from better growing conditions

be repeated with these variables included

and regular maintenance than trees that

in order to further analyze the influence of

are have been planted since. Another

site descriptors, since tree density is one of

potential explanation is a “survival of the

the most important measures of a healthy

fittest” scenario, where a few trees that do

urban forest. Though this study focused on

survive grow to maturity while most die

sociopolitical site descriptors, the method

before reaching a size threshold where they

could easily be translated to biophysical

are more likely to persist.

attributes, such as riparian zones, slope,

Given the short average lifespan of

and aspect. Given the highly variable soil

urban trees (Galvin, 1999), a high mean

conditions in urban areas (Urban, 2008),

DBH may indicate that there are not

however, it is unlikely that accurate soil

enough younger trees planted to account

data could be collected on the scale that

for natural decline of the mature trees

would inform this type of analysis. The

that are bringing up the average DBH.

lack of practical results for tree condition

Add to this the common development

could possibly be addressed by analyzing

pressures on older areas of town, which

them using an ordinal classification system

tend to be near the city’s center, and the

rather than a one-hundred point scale.

result is a neighborhood with mature

Interpreting results for mean DBH This study suggests that, after species characteristics, age of urban development (here equivalent to the year a given tree’s location was annexed into the city) is the best predictor of the variation in DBH across Eugene’s street trees. The oldest areas of Eugene have the highest mean DBH. It is important to note that this does not suggest that these areas have the greatest number of trees or the highest percentage of tree canopy. It is likely that these areas simply have fewer trees and they tend to be older - actually a concern for managers rather than a relief. One explanation for

trees coming to the end of their lifespan. Increased development often results in any new trees being planted in less favorable growing conditions. For example, street tree planting is often a permit requirement for redevelopment, which results in trees planted in areas with less soil volume and more impervious surfaces than before redevelopment. Sociocultural

trends

influence

the species selected by developers and landscape architects (or required by the urban forester) and may also contribute to the negative relationship between age of development and DBH. In Eugene, Sally Butler (1987) noted three distinct periods

65


of species preferences in the city’s street trees. The oldest neighborhoods (pre 1910) had the greatest number of native tree species (including Acer macrophyllum, Fraxinus latifolia, and Populus spp.), which tended to be large-growing trees. The medium and youngest neighborhoods each had small to medium trees as the most abundant species (Crataegus spp., Prunus cerasifera ‘Thundercloud’). These observations suggest the growth of the nursery industry has increased the number of species available to consumers, and promoted the aesthetic qualities of certain species in particular. The

statistically

significant

and

dramatic differences in mean DBH across zoning and land use categories supports the common perception that these attributes influence characteristics of the

urban

forest (Wilson et al., 2003; Mincey, 2013; City of Portland Parks and Recreation, 2004; City of Gresham, 2011; Duh and Flanagan, 2009). For example, industrial zoning had one of the lowest mean DBH values, likely a reflection of the decreased lifespan of trees in these areas - in other words, they rarely grow to maturity. One of the most surprising mean DBH values is for the commercial zoning category, and the somewhat analogous retail land use category. In Eugene, these trees have some of the highest mean DBH values, though this phenomena may be more related to age of development, as discussed previously.

Interpreting results for structural defects Though

the

descriptive

analysis

of structural defects does not test for significant differences or have the ability to explain variation across predictor variables or their subcategories, results have

some

obvious

management

implications. These results can directly inform management and maintenance activities by bringing attention to the zoning and land use categories with the most abundant structural defects. In this case, basal and trunk defects were the most common, and were highest in industrial zoning and land use categories. These defects also have a disproportionate influence on tree health, since they can disrupt a tree’s ability to transport nutrients between its leaves and its roots. Likely causes include vandalism and negligence; for example, vehicle impacts, lawn mowers, and using trees as bike racks. Industrial zoning and land use both have the highest incidence of combined basal and trunk defects. These results, combined with the low mean DBH for industrial zoning, suggest an opportunity for intervention by urban forest managers. More detailed observation of these areas is necessary to determine the primary causes of the damage in order to develop strategies to address them. If these defects are caused by injury, trunk protection devices such as bollards or other barriers may reduce these incidents. If the defects are caused by abiotic

66


conditions, such as sun scald, heat, or drought, species and cultural requirements could be specified which could mitigate these phenomena. Since data is collected on-the-ground, it is possible that crown defects are underrepresented due to sampling bias - they are not as easy to see because they are further from the data collector. This hypothesis could be tested by conducting a targeted sample inventory with a revised protocol for assessment of crown defects. Trees with an coniferous evergreen size code fared better in terms of health, described here as a lower incidents of structural defects, and also were of higher mean DBH. The fact that the City of Eugene has banned these trees from planting as street trees for the past twenty years could be influencing their mean DBH, since it means there are few young evergreens to bring down the average. The lack of structural defects supports the notion that evergreen conifers have thicker bark and tend to recover from damage more quickly than broadleaf deciduous trees. Given the increased stormwater benefits of evergreen foliage, these results provide additional reasons to increase planting of evergreen trees in appropriate sites within urban areas. Appropriate sites for these trees would likely include areas where the tree would not impede solar access to homes and where planter width was adequate to support a large tree.

Interpreting the attributes of the urban environment Since age of a development is a powerful predictor for variation in DBH, and has been proven to significantly influence urban tree canopy (Heynen & Lindsey, 2003), more research is needed to determine exactly why these areas tend to support larger trees. How much of the variation is related to time, and how much is related to factors which can be influenced by urban forest managers? In Eugene, commercial zoning had the oldest mean year of development. As city zoning is highly dependent on historic sociopolitical factors at the municipal scale, it is unlikely that this is common across other cities. The mean age of development for land use categories, however, may prove more universal - with government land use having the oldest mean year of annexation, and agricultural land uses having youngest mean age. It is easy to imagine the government parcels being acquired early in a city’s development, while agricultural and timber land uses are continually annexed as cities expand. Many

urban

forest

professionals

have speculated on the reasons why older neighborhoods have larger trees - such as different species selection, better soil conditions, more consistent tree planting and maintenance regimes. Given appropriate data, many of these conditions could be analyzed using this method, and reintroduced if deemed worthy of investment.

67


The zoning and land use classifications

used to tailor notions of what types of goals

were poor predictors of DBH but did reveal

are achievable in a given area, by providing

differences in their relative frequency of

real data that supports the notion of urban

structural defect types. This suggests there

forest management units based on land

are some aspects of the urban forest which

use and conditions. Perhaps areas without

hold promise for further research into their

adequate planter widths and with high

relationship with land uses. As Wilson

mortality should not be planted with trees,

(2003) pointed out, “zoning ordinances,

in order to plant more trees in places with

while regulatory in nature, are only a guide

a greater likelihood of tree success.

for development of a landscape� (p. 304).

Though it is only the first step, this

For this reason, land use information is

research provides an example of how tree

more likely to reflect the actual qualities

inventory data can be analyzed to develop

of a particular area, and therefore will

a better understanding of what types of

probably be a more accurate variable for

factors have the most influence on tree

urban forest analysis.

size and health. My research process could

Implications for managers The significant differences urban forest size and health across zoning, land

availability of any city.

Implications for scientists

use, planter width, species, and age of

Further research is needed to untangle

development could help managers develop

the age of development variable and

new ways to plan maintenance activities

separate those factors which urban forest

and estimate costs and benefits, based on

managers could control, like planter width

actual differences in the urban forest in

and species selection, from those which it

response to these factors. For example, the

cannot, such as time.

high mean DBH of commercial zoning in

In order to analyze a street tree

this study could be targeted for increased

population for species diversity and tree

tree planting and maintenance activities in

density, we need to develop appropriate

order to respond to the inevitable decline

metrics. Diversity indices used in ecology,

of these mature trees.

such as the Simpson and Shannon-Weaver

The dramatic difference in basal defects

indices, provide a logical starting point for

in industrial zoning and land use described

ranking species diversity. The widespread

here provide an example of how this type

use of cultivars in urban forestry may

of analysis could inform and support the

require some modification of these indices,

revision of city code or ordinances, in

in order to account for diversity at the

order to require specific maintenance or

cultivar level.

tree protection measures that avoid or mitigate the cause of this damage. It is my hope that these results will be

68

be tailored to the specific needs and data

For street trees, tree density is limited by available planting sites. This is often somewhat analogous to the amount of


unpaved right-of-way. Tree inventory

interested in further research of municipal

reports often calculate stocking level, which

tree inventory data. Different cities and

is the percentage of suitable planting sites

study areas will likely be interested in

that currently have trees. This calculation

investigating different types of urban

could provide the basis for a tree density

environmental attributes.

metric.

The largest limitations of this study

In creating diversity and density

are the lack of measures of species

metrics, it will be important to consider

diversity and tree density. These variables

their ability to be calculated for relatively

have a dramatic influence on the health

small units of the urban forest, in order to

and structure of the urban forest and its

produce a fine “grain� of data for statistical

benefits. While it may be challenging to

analyses. These metrics could be calculated

calculate these measures for an individual

for census blocks, or for an arbitrary

tree, aggregating diversity or density values

uniform grid overlaid on the study area.

to a census block scale would allow for a

Finally, statistical analysis of tree defects, possibly using indicator variables

balance of relatively fine grain analysis using easily accessible GIS data.

in a multiple regression analysis should

The replicability of the study should

be conducted to quantify differences in

be fairly straightforward for those with

the occurrence of defects across different

the available skills and data. Descriptive

predictor variables.

analysis of tree inventory data can easily be conducted using pivot tables, a feature

Conclusion This study has tested the ability for a more comprehensive approach toward analysis of municipal tree inventory data to inform urban forest management and maintenance. While results indicate this process can produce some information which

can

be

directly

applied

to

management and maintenance activities, further research is needed to expand these results and develop a suite of strategies that specifically address unique opportunities and

challenges

of

different

urban

environmental attributes. The advanced analysis of urban tree canopy in relationship to these attributes should serve as a guide for the types of analysis which could inform those

of Microsoft Excel. Statistical analysis will likely require a specialized statistical software package and knowledge of statistical methods. This project set out to determine how the value of tree inventory data could be improve management and maintenance activities. It is clear that many of the urban environmental

attributes

investigated

have significant relationship with urban forest structure and health. Though neighborhood tree inventories should be encouraged for their many social and economic benefits, urban forest managers should not limit their analyses to these boundaries, since they ignore the factors which have the most impact on these trees.

69


REFERENCES Anderson, Patrick, Mike Sherwood, and Brian Frierson. The Town of Davidson Street Tree Inventory and Management Plan. Street Tree Inventory and Management Plan. Davidson, North Carolina: Bartlett Tree Experts, 2007. http:// www.ci.davidson.nc.us/DocumentCenter/Home/ View/585.

City of Eugene Parks and Open Space. Jefferson Westside Street Tree Inventory. Street tree inventory report. Eugene, OR, 2013.

American Public Works Association. “Urban Forestry Best Management Practices for Public Works Managers: Urban Forest Management Plan.” Accessed May 12, 2014. http://www2. apwa.net/documents/About/CoopAgreements/ UrbanForestry/UrbanForestry-4.pdf.

City of Los Angeles. (2006). Million Trees LA. http://www.milliontreesla.org/mtabout1.htm

Bartenstein, Fred. “Breaking New Ground in Urban Forestry  ; 1.” Pinchot Institute for Conservation Studies, U.S. Department of Agriculture Forest Service, 1981. Bassuk, Nina L. “Using Excel 2007 to Analyze the Street Tree Inventory.” Community Forestry, Cornell University, June 13, 2013. http://www.hort. cornell.edu/commfor/inventory/usingExcel2007. html Bond, Jerry, and Beth Buchanan. Tree Inventories. International Society of Arboriculture, 2006. Bond, J., and Bryan Kotwica. Tree Inventories. Champaign, IL: International Society of Arboriculture, 2013. Print. Butler, Sally A. “A Cultural History of the American Urban Forest Using Eugene, Oregon as a Case Study.” M.A., Geography, University of Oregon, 1987. Christy, John A., and Edward R. Alverson. “Historical Vegetation of the Willamette Valley, Oregon, circa 1850.” Northwest Science 85, no. 2 (July 2011): 93–107. doi:10.3955/046.085.0202. City of Austin, Urban Forestry Board. Austin’s Urban Forest Plan; A Master Plan for Public Property. Urban forest management plan. Austin, Texas, 2013. http://www.austintexas.gov/ department/austin%E2%80%99s-urban-forest-plan. City of Eugene Urban Forest Management Plan. Urban forest management plan. Eugene, OR: City of Eugene Public Works Department, Maintenance Division, December 1992. https://scholarsbank. uoregon.edu/xmlui/bitstream/handle/1794/9600/ Eugene_Urban_Forest_Management_Plan_1992. pdf?sequence=1.

70

City of Gresham. Urban Forestry Management Plan. Urban forest management plan. Gresham, Oregon, July 2011. http://greshamoregon.gov/ urbanforestryplan/.

City of Portland, Parks and Recreation. Portland Urban Forestry Management Plan. Urban forest management plan, 2004. http://ufmptoolkit. com/pdf/Portland-Urban-Forestry-Plan-2004.pdf. City of Seattle Urban Forest Coalition. City of Seattle Urban Forest Management Plan. Seattle, WA, April 2007. http://www.seattle.gov/trees/ management.htm. Conway, Tenley M., and Kirstin S. Bourne. “A Comparison of Neighborhood Characteristics Related to Canopy Cover, Stem Density and Species Richness in an Urban Forest.” Landscape and Urban Planning 113 (May 2013): 10–18. doi:10.1016/j. landurbplan.2013.01.005. Cumming, Anne Buckelew, Daniel B. Twardus, and David J. Nowak. “Urban Forest Health Monitoring: Large Scale Assessments in the United States.” Arboriculture and Urban Forestry 34, no. 6 (2008): 341–46. D’Arcy, C. J. “Dutch Elm Disease.” The Plant Health Instructor, 2000. doi:10.1094/ PHI-I-2000-0721-02. Davey Resource Group. “City of Portland, Oregon. Initial Assessment of the Costs of Managing Street Trees as a Public Asset.” City of Portland Bureau of Planning and Sustainability, 2009. Duh, Steve, and Terry Flanagan. City of Corvallis Urban Forestry Management Plan. Urban forest management plan. Corvallis, Oregon: Corvallis Parks & Recreation, October 2009. http:// www.corvallisoregon.gov/index.aspx?page=264. Forest History Society, N.D. Essay: History of Urban Forests. Trees in your own backyard curriculum. Web. http://www.foresthistory.org/ education/curriculum/activity/activ7/activ7.html Frank, Stephen. “An Analysis of the Street Tree Population of Greater Melbourne at the Beginning of the 21st Century.” Arboriculture & Urban Forestry 32, no. 4 (2006): 155–63.


Galvin, Michael F. “A Methodology for Assessing and Managing Biodiversity in Street Tree Populations: A Case Study.” Journal of Arboriculture 25, no. 3 (1999): 124–28.

lease-on-life-Hardier-2886410.php#src=fb.

Grogan, Joel R. “Tree Canopy in the Emerald City.” Tree Canopy Analysis. unpublished, March 2014.

Mincey, Sarah K. “Zoning, Land Use, and Urban Tree Canopy Cover: The Importance of Scale.” Urban Forestry & Urban Greening 12, no. 2 (2013): 191–99.

Heynen, Nikolas C., and Greg Lindsey. “Correlates of Urban Forest Canopy Cover: Implications for Local Public Works.” Public Works Management & Policy 8, no. 1 (July 1, 2003): 33–47. doi:10.1177/1087724X03008001004. i-Tree Eco Manual. I-Tree Software Suite v5.0. (N.D.). Retrieved May, 2014, from http://www. itreetools.org/resources/manuals/Eco_Manual_ v5.pdf Johnston, Mark. “A Brief History of Urban Forestry in the United States.” Arboricultural Journal 20, no. 3 (1996): 257–78. Kenney, W. Andy, Philip JE van Wassenaer, and Alexander L. Satel. “Criteria and Indicators for Strategic Urban Forest Planning and Management.” Arboriculture and Urban Forestry 37, no. 3 (2011): 108–17. Kielbaso, J. James. “Trends and Issues in City Forests.” Journal of Arboriculture 16, no. 3 (1990): 69–76. Lane, D. (2010). Tukey’s honestly significant difference (HSD). In N. Salkind (Ed.), Encyclopedia of research design. (pp. 1566-1571). Thousand Oaks, CA: SAGE Publications, Inc. doi: http://dx.doi. org/10.4135/9781412961288.n478 Maggio, Robert. “A Geographically Referenced Tree Inventory System for Microcomputers.” Journal of Arboriculture 12, no. 10 (1986): 246–50. Latimer, Fremont. “Urban Forest Planning: A Revised Process Using Technology and Concept Development to Develop Structure and Function.” Master’s of Landscape Architecture, University of Florida, 2010. Lawrence, Henry W. City Trees: A Historical Geography from the Renaissance through the Nineteenth Century. Charlottesville: University of Virginia, 2006. Print. Malone, Rita. “The First Step: Taking a Tree Inventory.” American Forests, 1983. McCombs, Phil. “American Elm Gets New Lease on Life / Hardier Strains Are Resistant to Disease.” SFGate, August 18, 2001. http://www.sfgate.com/ homeandgarden/article/American-elm-gets-new-

Miller, Robert W. Urban Forestry: Planning and Managing Urban Greenspaces. 2nd ed. Englewood Cliffs, New Jersey: Prentice Hall, 1997.

Morani, Arianna, David J. Nowak, Satoshi Hirabayashi, and Carlo Calfapietra. “How to Select the Best Tree Planting Locations to Enhance Air Pollution Removal in the MillionTreesNYC Initiative.” Environmental Pollution 159, no. 5 (May 2011): 1040–47. doi:10.1016/j.envpol.2010.11.022. Nowak, David. “Assessing Urban Forest Structure.” Arboriculture & Urban Forestry 34, no. 6 (2008): 391–92. Nowak, David J. “The Effects of Urban Trees on Air Quality.” USDA Forest Service, Syracuse, NY. Online. ttp://www. Fs. Fed. Us/ne/syracuse/gif/trees. Pdf, 2002. Nowak, David J.; Stein, Susan M.; Randler, Paula B.; Greenfi eld, Eric J.; Comas, Sara J.; Carr, Mary A.; Alig, Ralph J. 2010. Sustaining America’s urban trees and forests: a Forests on the Edge report. Gen. Tech. Rep. NRS-62. Newtown Square, PA: U.S. Department of Agriculture, Forest Service, Northern Research Station. 27 p. Pauleit, Stephen, and Friedrich Duhme. “GIS Assessment of Munich’s Urban Forest Structure for Urban Planning.” Journal of Arboriculture 26, no. 3 (2000): 133–41. Phillips, Don. Assessment of Ecosystem Services by Urban Trees: Public Lands Within the Urban Growth Boundary of Corvallis, Oregon. 2012. http://www.itreetools.org/resources/reports/ Corvallis_Urban_Tree_Assessment_Tech_Report. pdf. Poracsky, Joseph, and Michael Lackner. “Urban Forest Canopy Cover in Portland, Oregon, 19722002.” Final Project Report. Prepared for Portland General Electric and City of Portland Urban Forestry Commission, 2004, 42. Ries, Paul D., A. Scott Reed, and Sarah J. Kresse. “The Impact of Statewide Urban Forestry Programs: A Survey of Cities in Oregon, US.” Arboriculture and Urban Forestry 33, no. 3 (2007): 168. Reynolds, Phillis C., and Elizabeth F. Dimon. Trees of Greater Portland. First edition. Portland, Or: Timber Press, Incorporated, 1993.

71


Rowntree, Rowan A. “Forest Canopy Cover and Land Use in Four Eastern United States Cities.” Urban Ecology 8, no. 1 (1984): 55–67. Safford, H.; Larry, E.; McPherson, E.G.; Nowak, D.J.; Westphal, L.M. (August 2013). Urban Forests and Climate Change. U.S. Department of Agriculture, Forest Service, Climate Change Resource Center. Santamour, Frank S, Jr. 1990. Trees for urban planting: diversity, uniformity, and common sense. Online. http://www.ces.ncsu.edu/fletcher/programs/ nursery/metria/metria7/m79.pdf accessed 5/2014 Steenberg, James W.N., Peter N. Duinker, and John D. Charles. “The Neighbourhood Approach to Urban Forest Management: The Case of Halifax, Canada.” Landscape and Urban Planning 117 (September 2013): 135–44. doi:10.1016/j. landurbplan.2013.04.003. Toronto and Region Conservation Authority. City of Mississauga Urban Forest Study. Urban forest technical report, 2011. http://www.peelregion.ca/ planning/climatechange/reports/pdf/mississ-urbforest-study-july15-2011.pdf. U.S. Census Bureau, Population Division Release Date: May 2013. Spirn, Anne W. The Granite Garden: Urban Nature And Human Design. New York: Basic Books, 1985. Urban, James. Up By Roots. ISA, 2008. Wilson, Jeffrey S., Michaun Clay, Emily Martin, Denise Stuckey, and Kim Vedder-Risch. “Evaluating Environmental Influences of Zoning in Urban Ecosystems with Remote Sensing.” Remote Sensing of Environment 86, no. 3 (August 2003): 303–21. doi:10.1016/S0034-4257(03)00084-1. Vogt, Jessica M., Sarah K. Mincey, Burnell C. Fischer, and Matt Patterson. Planted Tree Reinventory Protocol. Bloomington Urban Forestry Research Group at CIPEC, 2013. http://www.indiana. edu/~cipec/research/ProtocolBooklet_2-13-13.pdf.

72


APPENDIX


Appendix A: Frequency Charts(DBH) for selected variables Diameter Mean = 8.7 Std. Dev. = 8.047 N = 50,914

Number of trees for each DBH value in Eugene’s street tree inventory

Frequency

6,000

4,000

2,000

0

0

10

20

30

40

50

60

70

80

90

100

110

120

DBH inches

Number of trees for each planter width value in Eugene’s street tree inventory

Mean = 11.357 Std. Dev. = 13.2508 N = 46,482

16,000.0

Frequency

12,000.0

8,000.0

4,000.0

0.0

.0

20.0

40.0

growspace_Interval

74

60.0


Percentage of Street Trees per Zoning Classification 100

Percent of trees in each zoning category in Eugene’s street tree inventory

Percent of total

80

60

40

20

0

Agriculture

Commercial

Industrial

Natural Resource & Public

Other

Residential

Zoning classification

40,000

Number of trees in each land use category in Eugene’s street tree inventory

Count

30,000

20,000

10,000

0

Ag_Timb Govt_Gnrlsvc Industrial Mobile Retail Vacant Ed_Relg Group_MFR Infrastructure Parks_Rec SFR_Dup Water

LandUse_SIMP

75


Appendix B: How to read a means plot graph

Does zoning influence DBH?

Mean diameter (DBH) inches

12

10.78

10.57 10

9.30 8.53 8

6

5.71 4.81

4

no overlap= statistically different

Mean

Error bars (95% confidence)

2

0

Number of trees

Agriculture

Industrial

N = 221

1,859

Error Bars: 95% CI

Residential Natural Resource 40,853 & Public 1,944

Zone_name

N = 44

76

Other

Commercial

1,607

4,365

overlap = not statistically different


Appendix C: Correlation analyses

Correlations

iTree_ size_ Decid vs Evg

iTree_size_ b0c1 1

iTree_ size_123 .342**

Pearson Correlation Sig. (2-tailed) 0.00 N 38632 38632 iTree_ Pearson Cor- .342** 1 size_S, relation M, L Sig. (2-tailed) 0.00 N 38632 38632 **. Correlation is significant at the 0.01 level (2-tailed). Correlations diameter iTree_ size_123 diameter Correlation 1.00 .132** (DBH) Coefficient Sig. (2-tailed) 0.00 N 50914 38632 iTree_ Correlation .132** 1.00 size_S, Coefficient M, L Sig. (2-tailed) 0.00 N 38632 38632 **. Correlation is significant at the 0.01 level (2-tailed). Spearman’s rho

Correlation of Evergreen versus deciduous size code and nominal size code (small, medium, large.)

Correlation of DBH and nominal size code (small, medium, large).

Correlations iTree_ diameter size_ b0c1 Spear- iTree_ Correlation 1.00 .270** man’s size_ Coefficient rho Decid vs Sig. 0.00 Evg (2-tailed) N 38632 38632 diameter Correlation .270** 1.00 (DBH) Coefficient Sig. 0.00 (2-tailed) N 38632 50914 **. Correlation is significant at the 0.01 level (2-tailed).

Correlation of DBH and evergreen vs deciduous size code.

77


Appendix C: Correlation analyses Correlations

Correlation of planter width and DBH

planter diameter width Spearman’s planter Correlation 1.00 .216** rho width Coefficient Sig. 0.00 (2-tailed) N 46482 46482 diameter Correlation .216** 1.00 (DBH) Coefficient Sig. 0.00 (2-tailed) N 46482 50914 **. Correlation is significant at the 0.01 level (2-tailed).

Correlations

Correlation of Age of Development (year annexed) and DBH

diameter yearanx Correlation 1.00 -.308** Coefficient Sig. (2-tailed) 0.00 N 50914 49896 year Correlation -.308** 1.00 annexed Coefficient Sig. (2-tailed) 0.00 N 49896 49896 **. Correlation is significant at the 0.01 level (2-tailed). Spearman’s rho

diameter (DBH)

Correlations

Correlation of Age of Development (year annexed) and Planter width

78

yearanx planter width Correla- 1.00 -.158** tion Coefficient Sig. 0.00 (2-tailed) N 49896 45734 planter Correla- -.158** 1.00 width tion Coefficient Sig. 0.00 (2-tailed) N 45734 46482 **. Correlation is significant at the 0.01 level (2-tailed). Spearman’s year rho annexed


Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.