Landscape and Urban Planning 101 (2011) 244–252

Contents lists available at ScienceDirect

Landscape and Urban Planning

journal homepage: www.elsevier.com/locate/landurbplan

Temporal and spatial variation in garden and street trees in six eastern
Australian cities
J.B. Kirkpatrick ∗ , G.D. Daniels, A. Davison
School of Geography and Environmental Studies, University of Tasmania, Private Bag 78, GPO, Hobart, Tasmania 7001, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Trees are an economically, socially and culturally important component of cities, yet, in single city studies,
Received 15 November 2010 appear to be less dense in areas of low income and educational status than in areas of high income and
Received in revised form 9 February 2011 education status. We found that this pattern occurred in six Australian cities over the period 1961–2006,
Accepted 13 February 2011
with conditions in 1961 predicting those in 2006. Tree presence in gardens conformed similarly to pre-
Available online 16 March 2011
dictors between cities, but the presence of street trees and the type of both street or garden trees did
not. Our data suggest that it would be possible to plan to double the number of street trees in Australian
Keywords:
cities in present circumstances, but that significant increases in garden tree numbers would depend on
Education
Household income
increasing the income and higher education attainment of lower socioeconomic groups.
Private yard trees © 2011 Elsevier B.V. All rights reserved.
Public trees
Urban forest
Wealth

1. Introduction people actively prevent tree establishment (Kirkpatrick, 2006). If

we can understand the social correlates of variations in tree abun-
Trees in cities, and their component suburbs, have well- dance and tree type it may be possible to utilise planning and
documented virtues to suit a wide variety of social aspirations: as educational mechanisms to influence social conditions that in turn
highly cost-effective ‘machines’ for saving money through hydro- influence tree presence, if such knowledge suggests that this policy
logical, climatic, noise and pollution moderation (Dwyer et al., goal is appropriate.
1992; McPherson et al., 1997); as habitats for wildlife (Chamberlain There is substantial variation in tree abundance per unit area
et al., 2004; Daniels and Kirkpatrick, 2006a; Gaston et al., 2006; between cities in the United States (e.g. Nowak et al., 1996, 2001).
MacGregor-Fors, 2008); as a means of enhancing mental and phys- There have been no detailed investigations of socioeconomic rea-
ical health (Kuo, 2001; Urlich, 1984); as imparters of high realty sons for this substantial inter-urban variation although ecoregion
value (Jensen et al., 2004; Luttik, 2000; Tyrvainen and Miettinen, and city size appear to be predictors (Nowak et al., 1996, 2001). In
2000); as components of social capital (Elmendorf, 2008; Kuo, other parts of the world comparisons between urban areas use the
2003); as factors in perceptions of safety (Kuo et al., 1998; Kuo and proportion of green space or woody vegetation abundance rather
Sullivan, 2001); and as participants in spiritual experience (Dwyer than tree abundance. In the case of Europe, the proportion of green
et al., 1991; Konijnendijk, 2008). The degree to which different space increases with city area, while mildly declining with pop-
types of trees provide the social and environmental benefits listed ulation density (Fuller and Gaston, 2009). In the case of woody
above is less well-documented, although some data exists on res- vegetation cover in residential neighbourhoods in eight country
ident preferences for different street trees (Williams, 2002). It is, towns in Victoria, Australia, no attempt was made to assess the
therefore, important not only to understand variations in abun- effect of individual towns, but the variation between neighbour-
dance of trees within cities, but also variation in the abundance of hoods peaked at medium density values, with high educational
types of trees. levels and with high proportions of immigrants (Luck et al., 2009).
There are many treeless spaces in cities that are biophysically In the individual city, the urban forest has been found to be com-
suitable for the spontaneous establishment of trees, suggesting that prised of numerous fragments with clumped distributions (Wang
et al., 2009). Low abundances of trees within parts of cities are typi-
cally most strongly associated with low levels of household income
∗ Corresponding author. Tel.: +61 3 62262460; fax: +61 3 62262989. and individual educational attainment (Heynen and Lindsey, 2003;
E-mail addresses: [email protected] (J.B. Kirkpatrick),
Hope et al., 2003; Iverson and Cook, 2000; Jensen et al., 2004;
[email protected] (G.D. Daniels), [email protected] (A. Davison). Kirkpatrick et al., 2007; Landry and Chakraborty, 2009; Martin

0169-2046/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.landurbplan.2011.02.029
 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252 245

et al., 2004; Mennis, 2006), although other socioeconomic fac- well defined spatial level at which census statistics are presented.
tors such as ethnicity (Heynen et al., 2006) and housing tenure These suburbs were stratified into six classes by median weekly
(Perkins et al., 2004) can be influential. Luck et al. (2009) and Boone household income (≥$AU1264, $AU1264–$1025, <$AU1025) and
et al. (2010) document legacy effects for woody plant cover in Aus- percentage of people with tertiary education (≥20.42%, <20.42%)
tralian country towns and tree cover in Baltimore respectively, with using the 2006 Australian Census data, as these variables were
the Baltimore study showing that past socioeconomic conditions highly predictive of trees and garden types with trees in a pre-
are better predictors than present conditions, while the Australian vious study of Hobart (Kirkpatrick et al., 2007). Ensuring no more
study showed a strengthening of the links between socioeconomic than 5 suburbs from any one city in each of the 6 socio-demographic
conditions and urban tree cover through time. Despite the impres- classes, 169 suburbs were randomly selected from the original sam-
sive volume of recent work on trees in cities, we do not know ple. Some socio-demographic classes were uncommon in smaller
whether increases in levels of income and education are associ- cities, which prevented the selection of 5 suburbs in each of the 36
ated with increases in the abundance of urban trees, or changes basal classes.
in lifeform composition, whether the relationships between urban Because educational qualifications and household income were
trees and these predictive socioeconomic variables vary between not recorded in the 1961 Australian Census, advertised house sale
cities in the one country, and whether there are differences in these price was taken as a surrogate for economic and social status. News-
relationships between private garden and public street trees. papers published in 1961 and 2006 in each city were consulted for
Australia provides an excellent context in which to seek answers these data. Prices were recorded until a running mean stabilised.
to these questions. In the early 1960s, the distribution of income Stabilisation required between 9 and 22 houses in each suburb, in
in Australia was more equal than in any other country, communist each time period. Median house price by suburb, for 1961 and 2006,
or capitalist (Pusey, 2003; Saunders, 1993), there was little unem- were then converted to an equivalent value on the basis of variation
ployment by the standards of the twenty-first century (Nickell in the Consumer Price Index. It was possible to collect these data
et al., 2005), and a low proportion of people with tertiary educa- for 53 suburbs, covering all 6 cities and 6 education/income classes.
tional qualifications (De la Fuente and Doménech, 2001). By 2006, These 53 suburbs were used as the study sample for the mapping
income inequality had grown significantly relative to the 1960s of tree presence.
(Baum and Gleeson, 2010; Jordan and Stillwell, 2007), the unem- Within the 53 suburbs, tree density was measured for individual
ployed constituted at least one in twenty of the workforce, and the private gardens. Using a subsample of 10 of the 53 suburbs, 18 was
underemployed many more, and approximately one in five adult found to be the number of gardens needed to gain stable mean val-
Australians had a tertiary educational qualification (Australian ues. Thus, within each of the 53 suburbs, 18 cadastral parcels were
Bureau of Statistics, 2008a). randomly selected by overlaying a contemporary cadastre map of
In relation to the question of intra-national urban forest varia- the suburb with a transparent grid and using random numbers as
tion, Australian cities occur from the tropics to the cool temperate grid co-ordinates. Most suburbs displayed little variation in block
zone, vary in population from less than 100,000 to several mil- size, preventing a bias towards larger blocks that could result from
lion, and occur in six States and two Territories, all of which have selecting parcels by this method. For those few suburbs in which
different legislative, regulatory and planning frameworks as well there were discrete areas of multi-acre properties, as well as for
as different ethnic and cultural histories (Gleeson and Low, 2000; more densely built up suburbs, parcels were selected from each
Troy, 1995). area in proportion to the number of residences. To be selected, a
The majority of urban trees are either situated in private gar- parcel had to have a permanent structure in 2006. Most structures
dens, or in public lands. The typical Australian suburb has detached were detached residential buildings, some were business sites, and
houses with front and back gardens and streets with nature strips others apartment blocks.
(verges) and footpaths (sidewalks) (Fig. 1). Householders in one For each of the 18 parcels within each suburb, the following
city in Australia have been shown to reflect their personalities and data were collected, mostly through interpretation of historical
preferences in locally highly individuated gardens (Kirkpatrick et and contemporary aerial photography: parcel area in 2006 (m2 ,
al., 2009) that may, or may not, include trees (Kirkpatrick et al., measured from cadastre maps); area of house (or main building;
2007). In Australia, the management of public trees, including street including roofs adjoined to the house/building, unless they could
trees planted on nature strips, in cities is largely the responsibil- be determined to be wall-less) in 2006 (m2 ); total area covered
ity of municipal government (with most Australian cities typically by artificial surfaces (outbuildings, tennis courts, swimming pools,
encompassing several local government areas (LGAs)), supported and impervious surfacing, in addition to the house area) in 2006
by rates on property owners. LGAs have populations that vary from (m2 ); presence/absence of a house on the parcel in 1961; if a house
the tens of thousands, as in Hobart, to the millions, as in Brisbane. that was present in 1961 had been renovated between then and
In the present paper we document the patterns and socio- 2006 (renovations were determined through visual estimation of
demographic correlates of temporal and spatial variation in urban house size and shape); if any evident housing renovations had
trees in six eastern Australian cities, for the period 1961–2006, increased, decreased, or not changed the area of the home; the
and ask whether there are legacy effects and whether spatial pat- addition of a tennis court or swimming pool between 1961 and
terns in public and private urban trees are the products of different 2006 (because these two items can comprise a large proportion
sociopolitical processes. The implications of these results for urban of the artefactual surfacing of a contemporary property, and their
planning are discussed. presence/absence is easily discerned on 1961 photos, which is not
the case for lesser amounts of impervious surfacing, such as garden
2. Methods paths and driveways); total number of trees growing within the
boundaries of the property in 2006, and in 1961; and total number
2.1. Data collection of public trees growing directly adjacent to the property (i.e. street
trees growing on the median strip of, or within, roads bordering
Adelaide (South Australia), Brisbane (Queensland), Hobart (Tas- the parcel) in 2006, and in 1961. In all cases, trees were considered
mania), Melbourne (Victoria), Sydney (New South Wales) and as canopy forming plants greater than 4–5 m tall. Size relative to
Townsville, including the adjacent city of Thuringowa (Queensland) common structures, such as houses and cars, and shadow length,
were selected for study to represent wide variation in size, jurisdic- were indicators of height in areas where street level images were
tion and environment. These cities have a total of 1617 suburbs, a not available, or available aerial images were of low resolution.
246 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252

Fig. 1. A typical street in an Australian suburb.

Not all areas within the 53 suburbs were developed in 1961 to eucalypts, (n) the percentage of identified garden trees that were
the same extent as in 2006, meaning that some of the parcels were deciduous; (o) the percentage of identified garden trees that were
markedly different in visual distinctiveness in the two time peri- gymnosperms; (p) the percentage of identified garden trees that
ods. However, when analysing 1961 imagery, the location of such were evergreens, but not eucalypts or gymnosperms. The indepen-
parcels could be determined with good accuracy from parcels and dent variables were those listed in the previous paragraph plus city,
roads that were already established at that time, or other identi- the percentage of blocks (parcels) that were native forest in 1961,
fiable landmarks, such as trees, buildings, and bodies of water. To the percentage of blocks with a residence in 1961; the percentage
document these broad differences, the type of land use prevalent of blocks that were orchard in 1961, the percentage of blocks that
in the immediate area of each parcel in 1961 was classified into the were rural paddocks (fields) in 1961, the mean block size (m2 ),
categories of residential, orchard, natural bushland, paddock, and the mean house size (m2 ) and the mean artificial garden surface
other (farms, buildings other than residential). For the 1961 period, area (m2 ). Independent variables were only included in analyses
street tree counts were only conducted on parcels which already if there was reasonable theoretical expectation that they might
had road frontage, regardless of whether or not a house had yet influence the dependent variable. Thus, the variables derived from
been built. If the 2006 parcel had no more road frontage than the the 2006 census were not included in analyses designed to explain
width of the driveway, then street tree counts were not done for tree data collected for 1961. The Best Subsets option in Regression
either time period. in Minitab (Minitab Inc., 1998) was used to help select the theo-
The quality of 2006 vertical aerial photography, as well as retically feasible model with the highest explanatory power. This
the availability of eye level horizontal images, enabled far more routine was used to determine the two most explanatory regres-
detailed classification of trees than was possible for 1961. Thus, for sion models with each of 1–5 variables for each of the dependent
the 2006 time period only, in addition to counting the total numbers variables. No more than five independent variables could be used
of yard and street trees, definitively identified trees in each zone in the present study because of the size of the sample of suburbs.
were counted in the classes: eucalypts (Eucalyptus sensu lato and Models were rejected if the residuals did not approximate normal-
Angophora), pines and pine relatives (Gymnospermae), evergreen ity and if any of the components, other than the constant, were
angiosperms (Angiospermae) other than eucalypts, and deciduous insignificant at P > 0.05. The non-rejected Best Subsets model with
angiosperms. the highest R2 was selected. If dependent variables were signif-
In addition to median weekly household income and the per- icantly (P < 0.05) related to the class variable ‘city’ in a oneway
centage of people with tertiary qualifications, other independent ANOVA, the Best Subsets procedure was used on the residuals from
variables derived from the 2006 Australian Census were: median this relationship, with up to four variables selected, then a general
age of people; percentage of population born in Australia; per- linear model using a logit-link function, in which the selected con-
centage of workforce unemployed; percentage of workforce in tinuous variables were used as covariates and city as a factor, was
professional, managerial, and clerical/administrative positions; developed. One way ANOVA was used to test the significance of
and percentage of households renting their domicile. variation in dependent variables between cities, with Tukey’s test
used to determine the significance of differences between classes.
Pearson’s product moment correlation coefficient was used to test
2.2. Data analysis
the strength and significance of relationships between continuous
variables.
The spatial scale for all analyses was the suburb. The depen-
dent variables were: (a) the percentage of residences with one or
more trees in 2006, (b) the percentage of residences with one or 3. Results
more trees in 1961, (c) the percentage of nature strips adjacent
to residences with one or more trees in 2006, (d) the percentage 3.1. Patterns of spatial variation and temporal change
of nature strips adjacent to residences with one or more trees in
1961, (e) a − b, (f) c − d, (g) the number of garden trees per hectare More than half of the gardens had at least one tree in 1961, with
in 2006, (h) the number of garden trees per hectare in 1961, (i) a mean density of 54 trees per ha of garden (Table 1). There was
the mean number of street trees adjacent to residences in 2006, (j) no significant variation between cities in either of these variables
the mean number of street trees adjacent to residences in 1961, (k) (Table 1). Although the number of street trees per house did not
g − h, (l) i − j, (m) the percentage of identified garden trees that were vary significantly between cities, the presence of street trees did,
 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252 247

Table 1
Suburb means by city for garden and street trees for 1961 and 2006. GT%: the percentage of residences with one or more garden trees; ST%: the percentage of nature strips
adjacent to residences with one or more street trees; GT/ha: number of garden trees per hectare; 61all: trees per hectare in 1961 in the area occupied by residences in 2006;
61res: trees per hectare in 1961 in the area occupied by residences in 1961; STnumber: mean number of street trees adjacent to residences. Cities with the same letter in
columns have statistically identical values at P > 0.05.

City GT% ST% GT/ha STnumber

1961 2006 1961 2006 61all 61res 2006 1961 2006

*** *** ***
All 57.45 77.53 22.04 41.03 54.2 25.56 84.74 0.477 0.799***

Townsville 50.67a 82.33a 10.00b 40.00ab 30.12a 17.47a 89.60ab 0.134a 0.864abc
Brisbane 50.82a 80.64a 10.91b 29.27bc 62.95a 22.90a 103.43a 0.264a 0.525bc
Sydney 73.50a 85.00a 32.80a 48.30a 64.13a 34.55a 81.31ab 0.623a 0.862ab
Adelaide 59.10a 72.80a 32.00a 53.80a 42.90a 23.32a 62.73b 0.813a 1.189a
Melbourne 44.11a 70.33a 31.11a 53.78a 53.66a 19.08a 109.11a 0.491a 0.969ab
Hobart 65.57a 73.86a 8.57b 18.43c 63.71a 35.36a 56.19b 0.401a 0.305c

P 0.202 0.333 0.008 0.000 0.220 0.069 0.002 0.170 0.002

***
P < 0.001 (paired t-test, for GT/ha pertaining to both comparisons with 2006).

with Melbourne, Sydney and Adelaide having higher percentages dens with one or more trees in 1961 and the density of trees in
than the other cities (Table 1). gardens in 1961 (Table 3). Large blocks, small houses, low unem-
Between 1961 and 2006 there were strong overall increases in ployment rates, high percentages of blocks that were residential in
the presence and density of both garden and street trees (Table 1). A 1961, and low proportions of Australian-born residents predicted
strong overall increase of trees in gardens occurred even when the the percentage of gardens with one or more trees in 2006. The 2006
blocks that were developed between 1961 and 2006 were included garden tree density was predicted by low proportions of unem-
in the analysis (Table 1). This latter strong increase happened in all ployed and of Australian-born residents, low 1961 house prices,
cities except Hobart (Table 1). high proportions of rented dwellings and high proportions of peo-
By 2006, more than three quarters of the gardens in the sam- ple with tertiary education (Table 3). Increase in the percentage
ple had at least one tree and the mean density of trees per hectare of gardens with one or more trees between 1961 and 2006 was
of garden was 85 (Table 1). While there was no variation between predicted by low proportions of blocks that were bush in 1961, low
cities in the percentage of gardens with one or more trees, Mel- proportions of blocks that were residential in 1961 and low median
bourne and Brisbane had higher garden tree densities than Adelaide age of people in 2006, whereas increases in garden tree density in
and Hobart (Table 1). Both the percentage of houses with street the same period were predicted by high levels of tertiary education,
trees and their number per house were highest in Melbourne and low levels of unemployment, high proportions of households that
Adelaide and lowest in Hobart (Table 1). rent, a high proportion of blocks that were orchards in 1961 and
In 2006, eucalypts had their highest proportions in gardens in low median house prices in 1961 (Table 3).
Hobart and Adelaide and their least in the two northernmost cities, In 1961, both the percentage of houses with adjacent street
Brisbane and Townsville (Table 2). Melbourne, Adelaide and Hobart trees and the mean number of street trees varied strongly by city
had higher proportions of deciduous trees than the other three and, within cities, were influenced by land use variables (Table 4).
cities. Other angiosperms constituted most of the trees in Brisbane In 2006, there was a strong effect of city for both adjacent street
and Townsville, whereas gymnosperms were prominent in Sydney, trees and the mean number of street trees, with low unemploy-
but not elsewhere (Table 2). ment rate and a high proportion of blocks that were residential
in 1961 also influencing the presence of adjacent street trees, and
high median household income strongly influencing mean number
3.2. Correlates of variation of street trees (Table 4). There was a low level of explanation for
changes in the presence of street trees between 1961 and 2006,
The use of house price as a surrogate for household income with only the proportion of blocks that were orchards in 1961 con-
in 1961 is supported by an R2 of 60.9% for the relationship tributing. The more explanatory model for change in street tree
between median house price and median household income in number had this same variable and median household income as
2006 (F = 79.54, P < 0.001). The best model for 2006 median house positive components and median house price in 1961 and the pro-
price included a negative effect from the percentage of blocks that portion of blocks that were bush in 1961 as negative components
were paddocks in 1961, a positive effect from the median age of (Table 4).
persons, a negative effect from the percentage of people who were There was a strong influence of city on all models for the dif-
Australian born, a positive effect from median household income ferent types of trees in gardens. Eucalypt presence was strongly
and a positive effect from the percentage of gardens with one or related to garden size, deciduous trees to high levels of managerial,
more trees in 2006 (F = 44.16, P < 0.001). The other components professional and administrative employment, high levels of renting
added 21.5% to the R2 for median household income. and low levels of tertiary education, and both other angiosperms
There were significant relationships between the change in and gymnosperms to the nature of block land use in 1961 (Table 5).
median house price ($AU 1989) between 1961 and 2006 and: the
percentage of blocks with garden trees in 1961 (r = 0.304, P = 0.027),
the density of garden trees in 1961 (r = 0.276, P = 0.046) and the fre- 4. Discussion
quency of street trees in 1961 (r = 0.413, P = 0.002), the percentage
of blocks with garden trees in 2006 (r = 0.298, P = 0.030) and the fre- 4.1. Explaining temporal and spatial variation in garden tree
quency of street trees in 2006 (r = 0.405, P = 0.003). There were no abundance
significant correlations with change in any of the tree variables.
The 1961 median house price, the percentage of blocks that were Even within the relatively egalitarian structure of Australian
residential in 1961 and the percentage of blocks that were bush society in 1961, tree occurrence and density in gardens was higher
(native forest) in 1961 best predicted both the percentage of gar- where house prices were higher. Established suburbs tended to
248 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252

Table 2
The percentage of identified trees in gardens by taxonomic group by city in 2006. Figures with the same adjacent letter in columns are identical at P > 0.05.

Eucalypts Deciduous Other angio. Gymnosperm

Hobart 40.84a 18.54abc 23.40b 17.22b

Melbourne 24.82ab 30.47a 25.82b 18.89b
Adelaide 31.20a 22.73ab 26.41b 19.67b
Sydney 25.88ab 7.06b 29.97b 37.08a
Brisbane 9.41b 3.11c 80.65a 6.84b
Townsville 3.73b 0.69c 91.92a 3.67b

F 3.99 6.62 27.9 6.19

P 0.004 <0.001 <0.001 <0.001
R2 (%) 29.8 41.3 74.8 39.7

Table 3
Multiple regression models for garden trees by suburb, 1961 and 2006 and garden tree change (2006–1961).

1961 2006 2006–1961

GT% GT/ha GT% GT/ha GT% GT/ha

Constant −43.2** −3.74*** 110.3*** 133.59** 145.82 1.28

Median house price 1961 (AU$) 0.006*** 0.0002* – −0.0065* – −0.008*
Blocks native forest 1961 (%) 1.7*** 0.08*** – – −1.23** –
Blocks residential 1961 (%) 0.74*** 0.026*** 0.208* – −0.65*** –
Blocks orchard 1961 (%) – – – – – 2.07*
Mean block size (m2 ) – – 0.04*** – – –
Mean house size (m2 ) – – −0.14** – – –
Mean artificial garden surface (m2 ) – – – – – –
Households renting (%) – – – 1.286** – 1.53**
Median age of person (%) – – – – −1.78* –
Australian born (%) – – −0.52* −0.859* – –
Unemployed (%) – – −3.55** −9.86** – −5.85*
Tertiary educated (%) – – – 1.75** – 1.54*

F 17.67 19.63 7.58 12.71 6.7 8.64

P <0.001 <0.001 <0.001 <0.001 0.001 <0.001
R2 (%) 52.0 62.1 44.6 57.5 29.1 47.9

Asterisks indicate the level of significance for individual components of models: *** P< 0.001, ** P< 0.01, * P< 0.05.

have more trees than newer suburbs on the city margins that did positive influence of rented dwellings and the negative influence
not replace native forest. From the 1950s, a growing minority of of proportion of Australian-born in these models were surprising,
middle-class Australians began to reflect changing environmental but explicable. Rented dwellings were concentrated in older sub-
values by building and purchasing suburban houses located within urbs in the sample as indicated by a significant positive correlation
a bush setting, in preference to housing created by the destruc- between the proportion of blocks that had residences in 1961 and
tion of pre-existing bushland (Timms, 1999, 2006). The positive the proportion of dwellings that were rented in 2006 (r = 0.386,
influence of suburb age on woody plant cover has been previously P = 0.004). In 1961 the innermost ring of suburbs around the central
observed elsewhere (Heynen and Lindsey, 2003; Luck et al., 2009), business districts (inner cities) of Australia tended to be occupied
although it is not universal (Troy et al., 2007). by the poor (Stimson, 1982). The gentrification of these suburbs
The strong association between high levels of unemployment was substantially complete by the turn of the millenium (Baum
and low frequencies and densities of trees in gardens in 2006 is con- and Gleeson, 2010), with house prices too high for purchase by uni-
sistent with the income hypothesis, and the education hypothesis versity students, young managers and professionals who preferred
is also consistent with the model for 2006 garden tree density. The to live near the city centre. So although the proportion of rental

Table 4
General linear models for the percentages of identified trees in gardens by taxonomic groups.

Eucalypts Deciduous Other angio Gymnosperm

Factor
City *** *** *** ***
Covariates
Constant 1.34 −45.52* 155.1*** −29.46
Mean block size (m2 ) 0.053*** – – –
Mean artificial surface area (m2 ) −0.065* – – –
Managerial, professional and
administrative employment (%) – 1.5* – –
Households renting (%) – 0.41* – –
Tertiary educated (%) – −1.49* – –
Blocks native forest 1961 (%) – – −1.25** –
Blocks residential 1961 (%) – – −1.1** 0.48**
Blocks orchard 1961 (%) – – −1.76** –
Blocks paddocks 1961 (%) – – −1.01* 0.67**

R2 (%) 52.69 51.52 80.65 50.41

Asterisks indicate the level of significance for individual components of models: *** P< 0.001, ** P< 0.01, * P< 0.05.
 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252 249

Table 5
Multiple regression or general linear models for street trees, 1961 and 2006 and street tree change (2006–1961).

1961 2006 2006–1961

ST% STno ST% STno ST% STno

Factor
City ** ** *** *** – –
Covariates
Constant −16.22 −2.24* 33.61*** −0.081 18.29*** 0.365
Median house price 1961 (AU$) – – – – – −0.00009*
Blocks native forest 1961 (%) 0.7* 0.049*** – – – −0.025**
Blocks residential 1961 (%) 0.41** 0.027** 0.229** – – –
Blocks paddock 1961 (%) – 0.025* – – – –
Blocks orchard 1961 (%) – – – – 1.008* 0.028*
Median household income (AU$) – – – 0.00075*** – 0.0004*
Unemployed (%) – – −2.48* – – –

F – – – – 5.74 6.29
P – – – – 0.02 <0.001
R2 (%) 41.3 37.21 58.92 52.18 10.1 34.4

Asterisks indicate the level of significance for individual components of models: *** P< 0.001, ** P< 0.01, * P< 0.05.

0.50 Rented dwellings among migrants, with the poorest migrants having limited access
Education to the wealthy areas with high tree cover and potentially perceiving
Unemployed trees as a luxury. This relationship may vary spatially within and
0.25 between migrant communities given the varied tree-related atti-
Second Component

House price 2006

Managers tudes expressed by ethnic groups elsewhere (Fraser and Kenney,
Income 2000; Jay and Schraml, 2009; Lohr et al., 2004) and the varied asso-
0.00 House price 1961 ciations between migrant and ethnic communities and urban forest
cover (Heynen et al., 2006; Talarchek, 1990; Troy et al., 2007).
The model for the change in percentage frequency of garden
-0.25 trees by suburb simply reflected the fact that, in a situation in which
the garden tree population is increasing almost everywhere, those
suburbs with already high frequencies of garden trees are limited
-0.50 Australian-born
Median age (persons) in their potential for further increase (Troy et al., 2007; Wu et al.,
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 2008). Thus, the increase in frequency was concentrated in those
suburbs that were neither well-established nor largely forested in
First Component
1961. The model for change in tree density between 1961 and 2006
Fig. 2. Loading plot on first two principal components for socioeconomic variables was highly similar to that for tree density in 2006 because tree
showing significant Pearson product moment correlations (P < 0.05). Negative cor- density varied much less in 1961 than in 2006 (Table 1). Thus, the
relations are shown as dashed lines.
increasing inequality in income and education between 1961 and
2006 appears to have been expressed in increasing variation in gar-
den tree density, while the increasing levels of both income and
dwellings was negatively related to median household income, it education between 1961 and 2006 have been accompanied by a
was not related to 2006 house price (Fig. 2). This contrasts with the general increase in both the frequency and density of garden trees.
situation in other cities where tree cover is positively correlated
with levels of home ownership among residents (Heynen et al.,
2006; Perkins et al., 2004), indicating that contemporary associ- 4.2. Explaining variation in tree types
ations between renters and measures of the urban forest may be
influenced by city-specific socioeconomic conditions. Urban (Lohr et al., 2004) and rural (Erickson and de Young, 1993)
In 2006 the proportion of Australian-born was inversely related residents place a high value on the climate moderating services
to both the unemployment rate and median house prices (Fig. 2) that urban trees provide. In the present study the desire for cli-
indicating that migrants dominated both the highest and lowest mate moderation was expressed in the effects of inter-city climatic
socioeconomic strata of society. It may be that the Australian- differences on the proportions of the four types of trees. Decid-
born have a more negative attitude to urban trees on average than uous trees have the most value in gardens where there is a cold
the more prosperous of those who were born elsewhere, perhaps season, as they can be planted to allow the sun through in winter
because of the family links of many adult urban Australian-born and to create shade in summer. In subtropical and tropical climes,
people to the rural landscape, where trees might still be largely the main utility of trees is shade, generally best provided by the
regarded as an obstacle to production and a danger to life and dense canopies of evergreen angiosperms, rather than the perpet-
property. Indeed, within the Netherlands, people with a farming ually sparse crowns of eucalypts, the seasonally absent crowns of
background prefer views of managed landscapes more than views deciduous trees or the columnar structures of the gymnosperms.
of wild natural landscapes (Van den Berg and Koole, 2006). In con- The most popular of the gymnosperms in Queensland is Araucaria,
trast, American adults who were raised (at least partially) on farms a genus that casts little shade. The prominence of gymnosperms in
consider trees to be a more important factor in determining the Sydney might reflect their outstanding architectural qualities in a
quality of city living than people that have spent their entire lives city with a climate sufficiently warm in winter and sufficiently cool
in cities (Lohr et al., 2004). Thus, rather than indicating arboriphobia in summer to not be severely demanding of climate moderation
among Australian-born, the inverse relationship between tree pres- services.
ence and proportion of Australian-born residents reported here, The composition of urban plant species communities can be
and also observed by Luck et al. (2009), may indicate arboriphilia influenced by past land uses (Hope et al., 2003), the socioeconomics
250 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252

of former residents (Boone et al., 2010) and changed planning Bureau of Statistics, 2008b, 2008c). The fact that the number of
practices (Galvin, 1999; Thaiutsa et al., 2008). Within the present street trees in 2006 was greater where median household income
study, the model for other angiosperms suggests that they are was higher could result from a reluctance of the poorer LGAs to
proportionately most common in the suburbs that were neither spend on a perceived luxury, but, given the socio-economic het-
well-established in 1961, nor on the rural/urban fringe, whereas erogeneity of most LGAs may also reflect pressure from ratepayers
the model for gymnosperms suggests that they may be partly a to plant street trees in richer suburbs and pressure not to plant
legacy of use as windbreak and shelter trees in paddocks and partly trees, or to remove them, in poorer suburbs. In a 2010 case in the
a legacy of their popularity in gardens in Victorian and Edwardian suburb of Midway Point, Hobart, the Sorell Council had to abandon
times. partly implemented plans to place trees along a road in the face of
The prominence of eucalypts in Hobart may be a function of the a hostile reception from locals, who not only used the mass media
large number of its suburbs that are on the forest–urban interface to communicate their view, but also took direct action by destroy-
(Zagorski et al., 2004). Similar native forest legacies are evident ing plantings. At the same time, dozens of mature deciduous trees
within cities of the United States (Nowak et al., 2001). The promi- were planted in already well-treed streets in the wealthy University
nence of eucalypts in Adelaide, a city with scarce and rationed water suburb of Sandy Bay, with no apparent public opposition. Anecdotal
resources (Daniels and Tait, 2005), probably reflects selective plant- observations within Hong Kong suggest that between 10 and 15%
ing on the basis of drought resistance. Garden area was the other of newly planted public trees suffer mortality through vandalism
strong influence on the presence of eucalypts, which are more likely and that the extent of vandalism varies spatially (Jim, 1987).
to be retained when not close to a house because of their reputation Our data suggest that the ratio of ratepayers favouring street
for high propensity for limb breakage and high flammability. trees to those opposing street trees has changed in favour of the
The proportion of deciduous trees may be higher in inner city former between 1961 and 2006. The model for change in frequency
suburbs than elsewhere, given that high levels of renting and man- was poor, but the model for change in number intimated that the
agerial, professional and administrative employment are positive well-established suburbs and suburbs on bush margins of 1961 lost
components of their model. The opposite direction in the model of street trees by 2006, while suburbs that were replacing orchards in
the highly positively correlated employment and education vari- 1961 gained trees, as did the suburbs with the high median house-
ables (Fig. 2) suggests an effect of education independent of income, hold incomes in 2006. The reason that the percentage of blocks that
also found by Heynen and Lindsey (2003) and Kirkpatrick et al. were orchards in 1961 was included as positive in both the change
(2007). models was that, unlike the situation with paddocks, orchards were
generally situated in places with no roadside vegetation remnants.
4.3. Explaining variation in street trees
4.4. Comparison of correlates of street and garden trees
Whereas the frequency and mean number of garden trees was
independent of the factor city for both 1961 and 2006, city was The association of urban trees with household prosperity, as
an important influence on street trees. Variation in climate might indicated by house price and income, was strong in both 1961
have influenced local government perceptions of the utility of street and 2006 indicating consistent spatial processes. A significant posi-
trees, particularly environmental services or disservices related to tive correlation between change in house prices between 1961 and
shading (Mesa et al., 2010) and possibly potential maintenance 2006 and trees in both streets and gardens in 1961 could be taken
costs in storm-prone areas (Jim and Liu, 1997). However, it seems to indicate that trees attracted the wealthy. However, the much
more likely that there were cultural differences between times and stronger correlation between house prices in 1961 and change in
cities that were expressed through the action or inaction of local house prices between 1961 and 2006 (r = 0.668, P < 0.001) suggests
government. The data for Townsville and Brisbane, where street that the wealthy were predominantly attracted by the wealthy,
trees were rare in 1961, but highly abundant in 2006 do tend to rather than by the trees. The magnitude of the house price increase
suggest the latter explanation. Indeed, the Brisbane City Council was unrelated to the magnitude of tree increase for all tree vari-
recently concluded a decade long program involving the planting ables, suggesting that other influences, such as change in education
of 1 million urban trees. Furthermore, numbers of street trees were levels, were important in influencing tree numbers, an inference
consistently highest in Adelaide, a city which would seem to require supported by the model for change in garden tree density.
street trees for climatic moderation less than some others. The influences on frequency and density of garden trees were
Within the city of Bangalore, narrow streets have been found to not city specific. Thus, the social and economic factors that affect the
contain fewer trees than wide streets (Nagendra and Gopal, 2010). propensity of individuals to plant or remove trees in their private
Thus, the consistently low frequencies and densities of street trees gardens are not greatly influenced by cultural and climatic varia-
in Hobart could be attributed to the narrow streets in its early colo- tion between cities. The process is one in which increasing wealth
nial inner suburbs. However, the paucity of street trees in Hobart and education motivate greater tree planting than removal by indi-
was equally evident in post-colonial suburbs with wider streets. It viduals, whether in tropical Townsville or cool temperate Hobart.
may be that the extensive forested mountains visible from every In contrast, the types of tree that were planted in or removed from
part of Hobart have reduced the motivation to plant trees closer at gardens by individuals were very strongly influenced by city, prob-
hand, rather than street trees being abjured because of cool winters, ably largely as a function of climate-related viability and utility
which deciduous trees well suit. rather than as a result of interurban cultural differences. The trees in
In 1961 street tree frequency and number was higher in the streets seemed most subject to local government decision-making,
well-established suburbs and in areas in the process of urban devel- with differences between cities and times being difficult to explain
opment. In these latter areas, dirt roads with remnants of the native in any other way. These findings may indicate that intercity vari-
vegetation on their edges had large numbers of trees, later to be ations in the extent of urban forest elsewhere (Fuller and Gaston,
removed in favour of kerbs and footpaths. In 2006 there was a 2009; Nowak et al., 1996, 2001) are primarily due to variations in
high frequency of street trees in suburbs that were well-established the number of public trees, rather than of trees in private gardens,
in 1961, but a low frequency of street trees in suburbs with high or that our sample of Australian cities have comparatively homoge-
unemployment rates. By 2006, the poor had been largely banished nous ecoregions. For instance, eastern Australia has no desert cities,
to a mix of older middle-ring suburbs and the more physically which have particularly low levels of tree cover in the United States
undesirable parts of the outer margins of suburbia (e.g. Australian (Nowak et al., 1996, 2001).
 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252 251

4.5. Planning implications ral increases in levels of income and education are associated with
increases in urban tree density, that the causes of frequency and
The capacity for planners and land managers to increase the density of garden trees do not vary between our six cities, although
number of trees in their cities is greatest on streets, with 60% of city is an important element in models for street trees and for the
our sampled houses having no tree in front of their block, while four tree types. Most importantly, our data strongly suggest that
only 20% lack a tree in their garden. If it is assumed that the 20% it would be possible to double the number of street trees without
of householders who do not have a tree in their garden would social resistance. Although previous work within one of the cities in
react negatively to attempts to plant trees on their streets, there our sample (Kirkpatrick et al., 2009) suggests that there are limited
is still a capacity to double the number of houses with street trees. local planning options for changing the numbers of garden trees, a
A contingent valuation study conducted in 15 cities of Aotearoe more equal and educated society seems to be one that would have
New Zealand found that many residents actively support efforts a denser urban forest than that of today.
to increase the numbers of urban trees (Vesley, 2007). Further The correlative approach we used had the virtue of allowing
increases are possible where frontages are long or where trees have quantification of relationships between trees and tree types and
to be small because of overhead wires. In side streets, trees planted potential socioeconomic predictors, and, because of the random
in the middle of the street are good traffic-calming devices, again nature of the sampling process, should adequately represent these
increasing the potential for additions to the urban forest. The eco- relationships for the six Australian cities. However, the results and
nomic returns from street trees have been assessed at three to four conclusions may not be transportable to other cities and correla-
times the cost (McPherson et al., 1997; Soares et al., 2011), making tion does not conclusively establish causation. We are therefore
it foolish for local governments not to plant as many as is physically currently using content analysis, questionnaire data and qualitative
possible (McPherson et al., 2011) and socially bearable (Lohr et al., interviews to increase our depth of understanding of the quantita-
2004). There may even by spillover benefits for those individuals tive relationships revealed by the present study.
without a street tree directly adjacent to their property (Donovan
and Butry, 2010). Acknowledgment
In some areas, garden trees do not provide the same eco-
nomic benefits as streets trees (Donovan and Butry, 2010). This research was supported by the Australian Research Council
However, in other areas, relative to cost, garden trees provide (DP0878249 ‘Patterns, causes and consequences of tree retention,
greater economic benefits than trees in any other urban situa- establishment and removal in Australian cities’).
tion (McPherson et al., 1997). Increasing the number of private
trees may also be more beneficial than increasing the number References
of public trees due to the additional health benefits associated
with their nurturing (Kaplan, 1973). The potential for increase in Australian Bureau of Statistics, 2008a. Year Book Australia 2008, Catalogue No.
the number of private yards trees may be limited to some extent 1301.0. ABS, Canberra.
Australian Bureau of Statistics, 2008b. Sydney: A Social Atlas Catalogue No. 2030.1.
by housing density (Luck et al., 2009), but is also strongly influ- ABS, Canberra.
enced by the social and demographic characteristics of residents Australian Bureau of Statistics, 2008c. Melbourne: A Social Atlas Catalogue No.
(Green et al., 2011; Troy et al., 2007). 2030.2. ABS, Canberra.
Baum, S., Gleeson, B., 2010. Space and place: social exclusion in Australia’s suburban
The capacity for increasing the number of trees in private heartlands. Urban Policy Res. 28, 138–159.
gardens through command and control mechanisms, or even by Boone, C.G., Cadenasso, M.L., Grove, J.M., Schwarz, K., Buckley, G.L., 2010. Land-
incentives, would seem to be low in Australian conditions, where scape, vegetation characteristics, and group identity in an urban and suburban
watershed: why the 60 s matter. Urban Ecosyst. 13, 255–271.
private gardens are not generally regarded by their owners as social Chamberlain, D.E., Cannon, A.R., Toms, M.P., 2004. Associations of garden birds with
goods (Daniels and Kirkpatrick, 2006b; Kirkpatrick et al., 2009). An gradients in garden habitat and local habitat. Ecography 27, 589–600.
increase in private tree cover would probably result from a reduc- Daniels, C., Tait, C., 2005. Adelaide: Nature of a City. BioCity, Adelaide.
Daniels, G.D., Kirkpatrick, J.B., 2006a. Does variation in garden characteristics influ-
tion in income inequality, desirable for many other reasons, given
ence the conservation of birds in suburbia? Biol. Conserv. 133, 326–335.
that the number and frequency of trees did not increase over a Daniels, G.D., Kirkpatrick, J.B., 2006b. Comparing the characteristics of front and
median 2006 weekly household income of $1200.00. If incomes back domestic gardens in Hobart, Tasmania, Australia. Landscape Urban Plann.
78, 344–352.
continue their tendency to become more unequal, the benefits of
De la Fuente, A., Doménech, R., 2001. Education Attainment in the OECD, 1960–1995.
trees could be spread more evenly by implementing planning poli- OECD, Paris (CEPR Discussion Paper No. 3416).
cies that increase social mix, although this is likely to be a recipe for Donovan, G.H., Butry, D.T., 2010. Trees in the city: valuing street trees in Portland,
increased neighbourhood conflict over trees. Improving access to Oregon. Landscape Urban Plann. 94, 77–83.
Dwyer, J.F., Schroeder, H.W., Gobster, P.H., 1991. The significance of urban trees and
tertiary education might also ultimately be effective in increasing forests: toward a deeper understanding of values. J. Aboricult. 17, 276–284.
urban tree abundance on private lands, although research is needed Dwyer, J.F., McPherson, E.G., Schroeder, H.W., Rowntree, R.A., 1992. Assessing the
to determine whether there is any effect of the content or discipline benefits and costs of the urban forest. J. Aboricult. 18, 227–234.
Elmendorf, W., 2008. The importance of trees and nature in community: a review
of tertiary education on the propensity to plant trees. of the relative literature. J. Arboricult. 34, 152–156.
One area in which government could strongly influence tree Erickson, D., de Young, R., 1993. Management of farm woodlots and windbreaks:
numbers and types on private land in Australian cities is in the some psychological and landscape patterns. J. Environ. Syst. 22, 233–247.
Fraser, E.D.G., Kenney, W.A., 2000. Cultural background and landscape history as
limitation of tree clearing when single dwellings are replaced by factors affecting perceptions of urban forest. J. Arboricult. 26, 106–112.
multiple dwellings, a common phenomenon in Australian cities. At Fuller, R.A., Gaston, K.J., 2009. The scaling of green space coverage in European cities.
present, it is common for existing all garden trees to be removed Biol. Lett. 5, 352–355.
Gaston, K., Warren, P., Thompson, K., Smith, R.M., 2006. Urban domestic gardens
during such development. Regulations that require retention of at
(IV): the extent of the resource and its associated features. Biodivers. Conserv.
least some trees, perhaps of particular types, would have an imme- 14, 3327–3349.
diate effect. Galvin, M.F., 1999. A methodology for assessing and managing biodiversity in street
tree populations: a case study. J. Arboricult. 25, 124–128.
Gleeson, B., Low, N., 2000. Australian Urban Planning: New Challenges, New Agen-
5. Conclusions das. Allen & Unwin, Sydney.
Green, C.S., Millward, A.A., Ceh, B., 2011. Who is likely to plant a tree? The use of pub-
This has been the first study to develop models to explain vari- lic socio-demographic data to characterize client participants in a private urban
forestation program. Urban For. Urban Green., doi:10.1016/j.ufug.2010.11.004.
ation in street and garden tree frequency and density between Heynen, N.C., Lindsey, G., 2003. Correlates of urban forest canopy cover – implica-
cities in both space and time. We have established that tempo- tions for local public works. Public Works Manage. Policy 8, 33–47.
252 J.B. Kirkpatrick et al. / Landscape and Urban Planning 101 (2011) 244–252

Heynen, N.C., Perkins, Roy, P., 2006. The political ecology of uneven urban green Mesa, A., Arboit, M., de Rosa, C., 2010. Solar obstruction assessment model for
space. Urban Aff. Rev. 42, 3–25. densely forested urban environments. Archit. Sci. Rev. 53, 224–237.
Hope, D., Gries, C., Zhu, W., Fagan, W.F., Redman, C.L., Grimm, N.B., Nelson, A.L., Minitab Inc., 1998. User’s Guide 2: Data Analysis and Quality Tools Release 12.
Martin, C., Kinzig, A., 2003. Socioeconomics drive urban plant diversity. Proc. Minitab Inc., USA.
Natl. Acad. Sci. U.S.A. 100, 8788–8792. Nagendra, H., Gopal, D., 2010. Street trees in Bangalore: density, diversity, compo-
Iverson, L.R., Cook, E.A., 2000. Urban forest cover of the Chicago region and its relation sition and distribution. Urban For. Urban Green. 9, 129–137.
to household density and income. Urban Ecosyst. 4, 105–124. Nickell, S., Nunziata, L., Ochel, W., 2005. Unemployment in the OECD since the 1960s:
Jay, M., Schraml, U., 2009. Understanding the role of urban forests for migrants – what do we know? Econ. Rec. 115, 1–27.
uses, perception and integrative potential. Urban For. Urban Green. 8, 283–294. Nowak, D.J., Rowntree, R.A., McPherson, E.G., Sisinni, S.M., Kerkmann, E.R., Stevens,
Jensen, R., Gatrell, J., Boulton, J., Harper, B., 2004. Using remote sensing and geo- J.C., 1996. Measuring and analyzing urban tree cover. Landscape Urban Plann.
graphic information systems to study urban quality of life and urban forest 36, 49–57.
amenities. Ecol. Soc. 9, 5. Nowak, D.J., Noble, M.H., Sisinni, S.M., Dwyer, J.F., 2001. People & trees: assessing
Jim, C.Y., 1987. The status and prospects of urban trees in Hong Kong. Landscape the US urban forest resource. J. For. 99, 37–42.
Urban Plann. 14, 1–20. Perkins, H.A., Heynen, N., Wilson, J., 2004. Inequitable access to urban reafforesta-
Jim, C.Y., Liu, H.H.T., 1997. Storm damage on urban trees in Guangzhou, China. Land- tion: the impact of urban political economy on housing tenure and urban forests.
scape Urban Plann. 38, 45–59. Cities 21, 291–299.
Jordan, K., Stillwell, F., 2007. Financial futures: a touch of inequity? Dialogue 26, Pusey, M., 2003. The Experience of Middle Australia: The Dark Side of Economic
1–18. Reform. Cambridge University Press, Cambridge/New York.
Kaplan, R., 1973. Some psychological benefits of gardening. Environ. Behav. 5, Saunders, P., 1993. Longer run changes in the distribution of income in Australia.
145–162. Econ. Rec. 69, 353–366.
Kirkpatrick, J.B., 2006. The Ecologies of Paradise. Pandani Press, Hobart. Soares, A.L., Rego, F.C., McPherson, E.G., Simpson, J.R., Peper, P.J., Xiao, Q., 2011.
Kirkpatrick, J.B., Daniels, G.D., Davison, A., 2009. An antipodean test of spatial con- Benfits and costs of street trees in Lisbon, Portugal. Urban For. Urban Green.,
tagion in front garden character. Landscape Urban Plann. 93, 103–110. doi:10.1016/j.ufug.2010.12.001.
Kirkpatrick, J.B., Daniels, G.D., Zagorski, T., 2007. Explaining variation in front gardens Stimson, R., 1982. The Australian City: A Welfare Geography. Longman Cheshire,
between suburbs of Hobart, Tasmania, Australia. Landscape Urban Plann. 79, Melbourne.
314–322. Talarchek, G.M., 1990. The urban forest of New Orleans – an exploratory analysis of
Konijnendijk, C.C., 2008. The Forest and the City: The Cultural Landscape of Urban relationships. Urban Geogr. 11, 65–86.
Woodland. Springer, Dordrecht. Thaiutsa, B., Puangchit, L., Kjelgren, R., Arunpraparut, W., 2008. Urban green space,
Kuo, F.E., 2001. Coping with poverty: impacts of environment and attention in the street tree and heritage large tree assessment in Bangkok, Thailand. Urban For.
inner city. Environ. Behav. 33, 5–34. Urban Green. 7, 219–229.
Kuo, F.E., 2003. The role of arboriculture in a healthy social ecology. J. Arboricult. 29, Timms, P. (Ed.), 1999. The Nature of Gardens. Allen & Unwin, Sydney.
148–155. Timms, P., 2006. Australia’s Quarter Acre: The Story of the Ordinary Suburban Gar-
Kuo, F.E., Bacaioca, M., Sullivan, W.C., 1998. Transforming inner city landscapes: den. Miegunyah Press, Melbourne.
trees, sense of safety, and preferences. Environ. Behav. 30, 28–59. Troy, P. (Ed.), 1995. Australian Cities: Issues, Strategies and Policies for Urban
Kuo, F.E., Sullivan, W.C., 2001. Aggression and violence in the inner city: impacts of Australia in the 1990s. Cambridge University Press, Melbourne.
the environment via mental fatigue. Environ. Behav. 33, 543–571. Troy, A.R., Grove, J.M., O’Neill-Dunne, J.P.M., Pickett, S.T.A., Cadenasso, M.L., 2007.
Landry, S.M., Chakraborty, J., 2009. Street trees and equity: evaluating the spatial Predicting opportunities for greening and patterns of vegetation on private
distribution of an urban amenity. Environ. Plann. A 41, 2651–2670. urban lands. Environ. Manage. 40, 394–412.
Lohr, V.I., Pearson-Mims, C.H., Tarnai, A., Dillman, D.A., 2004. How urban resi- Tyrvainen, L., Miettinen, A., 2000. Property prices and urban forest amenities. J.
dents rate and rank the benefits and problems associated with trees in cities. Environ. Econ. Manage. 39, 205–233.
J. Arboricult. 30, 28–35. Urlich, R.S., 1984. View through a window may influence recovery from surgery.
Luck, G.W., Smallbone, L.T., O’Brien, R., 2009. Socio-economics and vegetation Science 224, 420–421.
change in urban ecosystems: patterns in space and time. Ecosystems 12, Van den Berg, A.E., Koole, S.L., 2006. New wilderness in the Netherlands: an investi-
604–620. gation of visual preferences for nature and development landscapes. Landscape
Luttik, J., 2000. The value of trees, water and open space as reflected by house prices Urban Plann. 78, 362–372.
in the Netherlands. Landscape Urban Plann. 48, 161–167. Vesley, E.-T., 2007. Green for green: the perceived value of a quantitative
MacGregor-Fors, I., 2008. Relation between habitat attributes and bird richness in a change in the urban tree estate of New Zealand. Ecol. Econ. 63, 605–
western Mexico suburb. Landscape Urban Plann. 84, 92–98. 615.
Martin, C.A., Warren, P.S., Kinzig, A.P., 2004. Neighbourhood socioeconomic status Wang, Y., Wu, Z., Wang, X., 2009. Urban forest landscape patterns in Ma’anshan City,
is a useful predictor of perennial landscape vegetation in residential neighbour- China. Int. J. Sust. Dev. World 16, 346–355.
hoods and embedded small parks of Phoenix, AZ. Landscape Urban Plann. 69, Williams, K., 2002. Exploring resident preferences for street trees in Melbourne,
355–368. Australia. J. Aboricult. 28, 161–170.
McPherson, E.G., Nowak, D., Heisler, G., Grimmond, S., Souch, C., Grant, R., Rowntree, Wu, C., Xiao, Q., McPherson, E.G., 2008. A method for locating potential tree-planting
R., 1997. Quantifying urban forest structure, function, and value: the Chicago sites in urban areas: a case study of Los Angeles, USA. Urban For. Urban Green.
Urban Forest Climate Project. Urban Ecosyst. 1, 49–61. 7, 65–76.
McPherson, E.G., Simpson, J.R., Xiao, Q., Wu, C., 2011. Million trees Los Angeles Zagorski, T., Kirkpatrick, J.B., Stratford, E., 2004. Gardens and the bush:
canopy cover and benefit assessment. Landscape Urban Plann. 99, 40–50. gardeners attitudes, garden types and invasives. Aust. Geogr. Stud. 42,
Mennis, J., 2006. Socioeconomic-vegetation relationships in urban, residential land: 207–220.
the case of Denver, Colorado. Photogramm. Eng. Remote Sens. 72, 911–921.