J Landurbplan 2011 11 023

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

Landscape and Urban Planning 105 (2012) 34–42

Contents lists available at SciVerse ScienceDirect

Landscape and Urban Planning


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

Plant traits link people’s plant preferences to the composition of their gardens
Dave Kendal a,b,∗ , Kathryn J.H. Williams a , Nicholas S.G. Williams a,b
a
Melbourne School of Land & Environment, University of Melbourne, Victoria, Australia
b
Australian Research Centre for Urban Ecology, Royal Botanic Gardens Melbourne, Victoria, Australia

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

Article history: Gardens are ubiquitous in western cities, comprising up to a third of the total urban area and often con-
Received 25 March 2011 taining a majority of the vegetation present. Gardens are the cumulative result of many individual plant
Received in revised form choices, yet we know relatively little about the role of preference in these choices. We investigated peo-
21 November 2011
ples’ preference for different garden plants and reasons for plant choices using a postal questionnaire
Accepted 30 November 2011
(n = 224) containing 30 colour photos of garden plants and questions about gardening behaviour. Pref-
Available online 30 December 2011
erences were compared with the plants growing in the gardens of 48 randomly selected respondents.
Objectively measured plant traits were used to relate preferences to the plants growing in people’s gar-
Keywords:
Landscape preference
dens. Significant relationships were found between survey responses and both the traits and taxonomy
Native plants of plants growing in respondent gardens. The results also show that people’s preferences are very diverse,
Horticulture and that these preferences were related both to aesthetic traits such as flower size, leaf width and foliage
Biodiversity colour, and non-visual traits such as nativeness and drought tolerance. Together these findings provide
Plant traits evidence that garden floras have responded to their social environment, and suggests that the very high
levels of diversity observed in gardens can in part be attributed to the heterogeneity observed in this
social environment.
Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.

1. Introduction across disciplines identifying a range of factors influencing the


occurrence of plants in residential gardens, including attributes of
As the world is becoming more urbanised (United Nations, people such as household income (Martin, Warren, & Kinzig, 2004)
2010), residential gardens are becoming an increasingly important and cultural background (Fraser & Kenney, 2000), attributes of
contributor to people’s health and wellbeing (Dunnett & Qasim, plants such as flowering (Marco, Barthelemy, Dutoit, & Bertaudière-
2000) and the ecological functioning of cities by maintaining bio- Montes, 2010) and fruiting (Acar, Acar, & Eroğlu, 2007) and physical
diversity (Daniels & Kirkpatrick, 2006b; Smith, Warren, Thompson, attributes such as rainfall (Daniels & Kirkpatrick, 2006a). Preference
& Gaston, 2006) and through the provision of ecosystem services (Behe & Nelson, 1995; Berghage & Wolnick, 2000; Townsley-
(Tratalos, Fuller, Warren, Davies, & Gaston, 2007). Gardens are the Brascamp & Marr, 1994) and the related concept of aesthetics (Head
cumulative result of many individual decisions about plant choice & Muir, 2006; Marco et al., 2010) influence people’s plant choices.
over time that combine to determine the social and biophysical However, there has been little research into the ways that peo-
benefits provided. These benefits can vary greatly between gar- ple’s preferences for plants actually shape the composition of the
dens and depend on the characteristics (traits) of the individual plants occurring in their gardens. This study strengthens our under-
plants in them. For example, gardens with many native plants standing of this relationship by examining people’s preferences for
may provide habitat that suits native birds more than exotic birds plants and plants in gardens using measures that are more objective
(e.g. Daniels & Kirkpatrick, 2006b) or increase ‘belonging’ for some and more comparable than approaches typically used in landscape
people (Head & Muir, 2006) while those with large canopies may preference studies.
provide more cooling (McPherson et al., 1997) and increase health The few studies that do explore whether preferences influ-
benefits (Mitchell & Popham, 2008). There is a body of research ence the plants grown in gardens have been limited in several
ways. Some are dependent on coarse, subjective measures of
self-reported garden composition or practice that are difficult to
generalise to other contexts. Kurz and Baudains (in press) found
∗ Corresponding author at: Australian Research Centre for Urban Ecology, Royal
that people who self-reported choosing native plants had much
Botanic Gardens Melbourne, c/o School of Botany, University of Melbourne, Victoria
higher preference for high-habitat gardens, and Larsen and Harlan
3010 Australia. Tel.: +61 3 8344 0267; fax: +61 3 347 5460.
E-mail addresses: dkendal@unimelb.edu.au (D. Kendal), kjhw@unimelb.edu.au (2006) found that people have self-reported garden styles that
(K.J.H. Williams), nsw@unimelb.edu.au (N.S.G. Williams). mostly matched their expressed preferences. There have been

0169-2046/$ – see front matter. Crown Copyright © 2011 Published by Elsevier B.V. All rights reserved.
doi:10.1016/j.landurbplan.2011.11.023
D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42 35

several studies at the garden landscape level that link preference to old city by Australian standards. The city has grown slowly but con-
the garden style presented in photographic stimulus, but provide sistently, with older houses tending to be towards the urban centre
little insight into how people choose particular plants. For example, and newer housing towards the fringe. The socioeconomic diversity
van den Berg and van Winsum-Westra (2010) show that people’s in Ballarat, with census districts ranging from the 1st to the 89th
preferences vary by the formality of the garden design, and Yabiku, percentile of the Australia wide Index of Relative Socio-economic
Casagrande, and Farley-Metzger (2008) found that people’s pref- Advantage and Disadvantage, allows detailed exploration of socioe-
erences for mesic and xeric garden styles differed in the desert conomic factors. However, we do not attempt to explore the relative
city of Phoenix, Arizona. Other studies have related psychologi- importance of cultural background as there is relatively little cul-
cal traits other than preference to garden characteristics. Zagorski, tural diversity in Ballarat as measured by all four core variables
Kirkpatrick, and Stratford (2004) found that garden species compo- in the Standards for Statistics on Cultural and Language Diversity
sition was related to attitude towards the effort of gardening and (Australian Bureau of Statistics, 1999). For example, the propor-
to the environment, while van den Berg and van Winsum-Westra tion of the population speaking a language other than English
(2010) found that the personality trait Personal Need for Structure (LOTE) was 8% in Ballarat compared to a Australian average of 22%,
was related to self-reported garden style. and the proportion of Indigenous Australians was 1% compared
Our study builds on this body of work by relating people’s to an Australian average of 2.3% (Australian Bureau of Statistics,
preferences to actual observations of plants in their gardens, and 2006).
by characterising plants using objective measures of plant traits. Like much of Australia, the vast majority of the housing in Bal-
The use of plant traits provides significant benefits for the under- larat is detached with both front and rear gardens, and the majority
standing the relationship between plant preferences and plants of houses are owner occupied (Australian Bureau of Statistics,
occurring in gardens. Our focus on plant traits draws on approaches 2006). Front gardens typically consist of garden beds of shrubs and
in ecology, where plant traits are commonly used to analyse the herbaceous plants around the perimiter enclosing an area of turf
functional response of plants to their environment, and to allow and possibly trees and bisected by a path and/or vehicle drive-
global comparisons of vegetation communities that are taxonom- way, although not all elements are present in all gardens. Many
ically distinct (Cornelissen et al., 2003). The study of plant traits of the plants grown in these gardens are exotic, as is the case in
in ecology has focussed explicitly on how the functional traits of many other western cities (e.g. Daniels & Kirkpatrick, 2006a; Loram,
plants (such as leaf mass per area and canopy height) vary in rela- Thompson, Warren, & Gaston, 2008; Marco et al., 2008; Martin et al.,
tion to physical environmental drivers such as climate and nutrient 2004). Remnant native vegetation around Ballarat largely consists
levels (e.g. Wright, Westoby, & Reich, 2002). Garden plants exist of grassy and shrubby woodlands, with a relatively sparse over-
in an environment that has both physical (e.g. temperature and storey of Eucalyptus sp. and Acacia sp. and an understorey of grasses
water availability) and social drivers (e.g. people’s preferences), and and shrubs such as Joycea pallida (R.Br.) H.P.Linder, Dianella revoluta
consequently may show trait responses to both. Loosely defined R.Br, Acacia stricta (Andrews) Willd., Leptospermum continentale Joy
plant traits have been used in many landscape preference stud- Thomps., Daviesia leptophylla A.Cunn. ex Don and Tetratheca cili-
ies, typically by expert categorisation of stimulus photographs (e.g. ata Lindl. Native vegetation differs in appearance from the plants
small, medium or large foliage) (Williams, 2002) or in respondent being cultivated in gardens by generally having smaller flowers
self-assessment of garden style categories (e.g. wild, manicured and narrower foliage that tends to be dull green or grey-green in
or romantic) (van den Berg & van Winsum-Westra, 2010). How- colour.
ever, the use of categories with loose definitions makes comparison
between studies and generalisation to other contexts difficult. This 2.2. Preference survey and sampling
study uses objective measures of plant traits to explore the rela-
tionship between preferences and the plants in people’s gardens, an The questionnaire was designed to explore people’s preferences
approach which will allow our results to be objectively compared for garden plants and their reasons for choosing the plants in their
with results from other places. Plant traits that have been related to front garden. The questions used were generated from preliminary
people’s preference in garden and non-garden contexts include leaf semi-structured interviews where a purposive sample of 14 people
and flower colour and size (Kaufman & Lohr, 2004, 2008; Kendal, was interviewed about the reasons they chose the plants in their
Williams, & Armstrong, 2008; Todorova, Asakawa, & Aikoh, 2004; front garden. Respondents were asked to circle words (or write
Townsley-Brascamp & Marr, 1994), leaf width (Williams, 2002), the other words) that described the desired characteristics and func-
provision of shade or fruit (Fraser & Kenney, 2000), tree canopy tional roles of plants they had chosen (see Figs. 1 and 2 for the
shape (Lohr & Pearson-Mims, 2006; Sommer & Summit, 1995; terms used). Some basic demographic questions identified in past
Williams, 2002) and nativeness (Williams, 2002). This study will research as influencing people’s landscape preferences and garden-
explore the role of preference in people’s garden plant choices by ing behaviour were also included (e.g. Bhatti, 2006; Martin et al.,
focussing on plant traits including flower size, leaf colour, native- 2004; Williams, 2002). The preference section asked participants to
ness and leaf width. We hypothesise that garden plants will have rate how much they liked each of 30 garden plants using a 5 point
traits that show responses to both social and physical drivers, and Likert scale, presented as colour photographs in a separate booklet
that people’s preference for aesthetic plant traits such as foliage (Table 1). All photographs were taken by the primary author using
colour and flower size will be reflected in the traits of the plants in a digital SLR camera with a near standard focal length lens and
their gardens. in overcast conditions to ensure relatively uniform lighting con-
ditions. The photographs were of individual plants in public and
private gardens in Ballarat and Melbourne, Australia, and the back-
2. Methods ground of each image was blurred using Gimp v2.6 to minimise any
distractions from the plant. The plants used were chosen to repre-
2.1. Study area sent a variety of trait values, and were stratified by the presence
of flowers, nativeness, foliage texture and colour, plant type (shrub
Ballarat is a regional city in south-eastern Australia (latitude or tussock forming perennial) and evidence of clipping (pruning
37.56◦ S, longitude 143.86◦ E), and is one of Australia’s largest to a regular shape such as a cube or sphere). Two presentation
inland cities with a population of approximately 82,000. Ballarat orders were determined randomly, and each address was randomly
was founded during the gold rush of the 1850s and is a relatively assigned to one of the presentation orders. The final photographs
36 D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42

Fig. 1. Responses to the question “Which things about the plants you have chosen in
your front garden influenced your decision?” Results are shown as the proportion
of respondents in each preference group with 95% confidence intervals. For each Fig. 2. Responses to the question “Which functions of the plants you have chosen in
response, values with the same letter were not significantly different (Tukey HSD your front garden influenced your decision?” Results are shown as the proportion
P < 0.05). of respondents in each preference group with 95% confidence intervals. For each
response, values with the same letter were not significantly different (Tukey HSD
P < 0.05).

used in the questionnaire were also selected based on findings from


the preliminary interviews, where a multiple sorting technique was Questionnaires were successfully posted to 695 addresses in
used to identify people’s plant preference categories. April 2010. This included 127 addresses that were part of a floristic
For each of 30 garden plants, a number of traits were deter- survey of the plants cultivated in people’s front gardens in which
mined. Visual traits were measured directly from the photographs addresses were randomly selected within eight census districts
used in the preference study. As there was only one individual plant stratified by socioeconomic status and geographic location. The
per photograph, all trait measurements were taken from a single remaining 568 addresses were randomly selected from the local
individual per taxa. Following Cornelissen et al. (2003), an aver- phone directory. Reminder letters were sent 2 weeks after the sur-
age of 5–10 measurements was used for each trait calculation. Leaf vey was initially posted, and included a URL where the survey could
width was measured as the maximum diameter, in image pixels, of be completed online. Additional responses for the online question-
a circle that fits completely within the leaf. Flower size was mea- naire were sought through local community newsletters. The adult
sured as the average diameter of inflorescences in image pixels. most responsible for decisions about the garden was asked to com-
Flowering was measured as the proportion of canopy covered in plete the questionnaire. A total of 238 responses were received,
flowers. Additional trait measurements were taken from published 199 from the random survey (a response rate of 29%) and 39 from
literature (e.g. University of Melbourne, 2002). Binary coded vari- responses to newsletters.
ables were used for grey foliage, green foliage, native (originating
from south-eastern Australia), clipped (evidence of pruning into a 2.3. Garden survey and sampling
regular shape) and tufted (tussock forming perennials). Drought
tolerance was taken from the Burnley Plant Directory (University Respondents were asked whether they consented to have
of Melbourne, 2002) and coded as an ordinal variable from 1 (low) their front garden surveyed to allow comparison between stated
to 4 (high). preferences and the traits of the plants in their gardens. Front
D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42 37

Table 1
Images used in preference survey that load most heavily on the four preference dimensions identified. PCA component loadings for the three most
popular plants for each preference dimension are given in brackets.

gardens were used as there is often a great deal of similarity flowering which were coded as ordinal variables with values from
between the plants in people’s front gardens and back gardens 1 to 5 from published information (e.g. University of Melbourne,
(Daniels & Kirkpatrick, 2006a), and this approach was less intru- 2002). Frequency was calculated as the proportion of gardens sur-
sive than a survey of back gardens. Approximately one third (n = 48) veyed that a taxa was observed in. Trait signatures were calculated
of the gardens of respondents who had consented to a garden for each garden. For binary coded traits (native, grey foliage, green
survey were randomly selected, a sample size allowing basic sta- foliage, tufted, clipped), trait signatures were calculated as the
tistical comparisons between respondent groups based on their proportion of taxa in the garden with that trait. For continuous
preferences. variables (leaf width, flower size, flowering and drought tolerance),
For each selected garden, all cultivated plants (excluding turf) trait signatures were calculated as the mean unweighted trait value
were identified to the lowest taxonomic level possible, usually for all taxa in the garden. The proportion of front yard covered
species, and the number of plants (abundance) of each taxa by turf, garden beds and hard surfaces and the age of the house
recorded. The number of taxa and total number of plants was associated with the garden were estimated from published infor-
calculated for each garden. Plants with obvious morphological dif- mation on historical housing styles (Heritage Council of Victoria
ferences, such as variegated leaves or purple foliage, were identified et al., 2007). Digital aerial photos (10 cm resolution) taken on 17
as separate taxa. Traits were assigned using the same method as the December 2007 were obtained from the City of Ballarat and used
preference study, with the exception of leaf width, flower size and to measure the width and depth of the front garden.
38 D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42

2.4. Statistical analysis Table 2


The plant traits loading on each preference dimension, measured by the Pearson’s
correlation coefficient between traits and PCA component loadings.
All statistical analysis performed using R v2.11.1
(R Development Core Team, 2010). The distribution of the Plant traits Preference dimensions
data were tested to ensure they met the assumption of normality Fine foliage Flowering Strappy Clipped
for parametric statistical tests. The final analysis only included
Native 0.52 −0.59 0.16 −0.07
data from the 224 survey forms in which the preference section Tussock forming perennials −0.20 −0.35 0.89 −0.20
was completed. Less than 1% of preference ratings were missing Evidence of clipping −0.34 −0.08 −0.31 0.74
and these were replaced by the respondent’s mean (Tabachnick & Leaf width −0.42 0.49 −0.10 −0.10
Fidell, 2001). A plant photograph similarity matrix was generated Grey foliage 0.39 −0.21 −0.04 −0.25
Green foliage −0.40 0.24 0.15 −0.08
as the Pearson correlation of the preference responses for each Flowering −0.22 0.54 −0.30 −0.37
plant photograph. A Principal Components Analysis (PCA) with Flower size −0.46 0.60 −0.11 −0.25
Varimax rotation (using the principal function in the psych package Frequency in gardens −0.42 0.62 −0.34 −0.09
in R) was used to group plant photographs that were treated simi- Drought tolerance 0.36 −0.48 0.19 −0.12
larily by respondents. A scree plot was used to identify the number
of components to extract. The PCA loadings for each component
less likely to be green and more likely to be grey, with smaller
were correlated with plant trait measurements to inform how the
flowers, were infrequently cultivated and more drought tolerant.
component was constructed. A respondent preference similarity
The second dimension (flowering, explaining 13% of the variation)
matrix was generated as the Pearson correlation coefficient of each
was related to plants that were less likely to be native, had broader
respondent’s preference responses. A cluster analysis using Wards’
foliage, larger flowers with a greater canopy coverage of flowers,
linkage (using the hclust function in R) was used to group respon-
were more frequently planted and less drought tolerant. The third
dents with similar preference responses. Respondent preferences
component (strappy, explaining 8% of the variation) were more
for each trait were calculated as the Pearson correlation coefficient
likely to be tussock forming perennials and were less frequently
of plant trait measurements with respondent’s preferences for
planted. The fourth component (clipped, explaining 5% of the vari-
each plant. The Pearson correlation coefficient was then calculated
ation) were more likely to show evidence of clipping and less likely
between mean preferences for each respondent cluster and plant
to have large flowers.
trait measurements to explore how trait preferences were related
There were also clearly identifiable differences between
to cluster membership.
respondents based on their preferences. The cluster analysis of
The relationship between respondent preferences and the com-
respondents’ preferences identified three main clusters of respon-
position of their gardens was explored in three ways. First, the
dents. Different visual and non-visual traits were preferred by
relationship for specific taxa was explored by comparing respon-
each respondent cluster (Table 3). Respondents in the first cluster
dent preferences for taxa that were both common in gardens
(labelled colourful, n = 97) had higher preferences for frequently
(occurring in at least three gardens) with the presence or absence
occuring plants with large flowers and a high proportion of the
of the taxa from their garden using ANOVA (the aov function in R).
canopy covered in flowers, and a lower preference for plants that
Second, respondent trait preferences were correlated with respon-
were native or drought tolerant. The second cluster (modern, n = 82)
dent garden trait values. Separate correlations were performed for
had higher preferences for tussock-forming perennials and lower
all respondents and for respondents who had resided at the same
preferences for clipped plants. Respondents in the third cluster
address for more than 5 years to determine whether length of resi-
(lush, n = 45) had higher preferences for frequently occuring plants
dence affected the relationship between preferences and the plants’
with green, coarse foliage and large flowers, and lower prefer-
growing in people’s gardens. Finally, to explore whether these indi-
ences for native plants, drought tolerant plants and plants with
vidual relationships between preference and the plants in people’s
grey foliage.
gardens affected patterns at a larger scale, the pairwise similar-
There were significant (Tukey’s HSD P < 0.05) demographic dif-
ity of taxonomic composition of respondents’ gardens and their
ferences between the respondent clusters. The respondents in the
preferences was compared across all respondents and gardens. A
colourful cluster were older (mean age = 59 years compared to 50
taxonomic similarity matrix for gardens (based on the inverse of
years for other clusters), had resided at their current address for
Bray Curtis distance, generated using the vegdist function in the
longer (mean years of residence = 16 compared to 11 for other clus-
vegan package in R) and a preference similarity matrix (the corre-
ters) and were more likely to be retired (46% compared to 21%
lation of respondent preferences) were compared with the Mantel
of respondents in other clusters). The respondents in the modern
test (using the mantel function in the vegan package in R).
cluster had higher proportions of people with a graduate educa-
tion (68% compared with 37% of respondents in other clusters)

3. Results
Table 3
The plant traits preferred by each respodent cluster, measured by the Pearson’s
3.1. Overall patterns in preferences correlation coefficient between traits of and mean preferences for each plant.

Plant traits Respondent cluster


There were clear patterns in the preferences of respondents
based on both visual (e.g. flowering, leaf size, habit, and evidence of Lush Modern Colourful
clipping) and non-visual (e.g. nativeness, frequency of occurence) Native −0.52 0.25 −0.47
traits of the plants in the preference study. The Principal Com- Tussock forming perennials 0.23 0.47 −0.31
ponents Analysis of preference responses identified four plant Evidence of clipping −0.02 −0.73 −0.18
preference dimensions (see Table 1) that explained 53% of the total Leaf width 0.32 −0.20 0.27
Grey foliage −0.37 0.22 −0.07
variation. The correlation of component loadings with plant traits Green foliage 0.53 0.15 0.29
measured from the plant photographs (see Table 2) was used to Flowering 0.31 0.15 0.73
develop a descriptive name for each dimension. The first plant pref- Flower size 0.44 0.00 0.60
erence dimension (fine-foliage, explaining 27% of the variation) was Frequency in gardens 0.56 −0.10 0.64
Drought tolerance −0.43 0.14 −0.36
related to plants that were native, with smaller leaves that were
D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42 39

Table 4 respondents who had been at the same residence longer, there
The relationship between people’s preferences and the traits of the plants in their
were much stronger relationships with all flower and foliage traits,
garden, measured by the Pearson’s correlation coefficients between respondent trait
preferences and the traits observed in their gardens. although the correlation with tussock forming plants was weakly
negative.
Plant traits All respondents Respondents more than 5
These first two measures of similarity explore the relationship
years at residence
between individual respondent preferences and the plants growing
Native 0.47 0.45 in their front gardens. At a broader scale using a pairwise com-
Flowering 0.26 0.32
parison of respondent preferences and the similarity of the taxa
Frequency in gardens 0.25 0.24
Grey foliage 0.16 0.42 growing in their gardens, people with similar preferences were also
Green foliage 0.15 0.34 more likely to have similar taxa growing in their gardens. There was
Flower size 0.07 0.23 a weak positive relationship (Mantel r = 0.15, P < 0.01) between the
Leaf width 0.07 0.20
pairwise similarity of respondents’ preferences and the taxonomic
Tufted 0.06 −0.12
composition of their gardens.

and had resided a their current address for the shortest time
(mean = 10 years). 4. Discussion

3.2. Stated reasons for choosing plants The garden plant preferences of the participants in this study
were related to aesthetic traits (e.g. flower size, leaf width and
Respondents stated (Figs. 1 and 2) that aesthetic characteris- foliage colour), context specific biophysical traits (e.g. drought tol-
tics such as flower and foliage traits and factors related to the erance) and extrinsic characteristics (e.g. nativeness). Different
plants’ suitability for a location such as drought tolerance and people preferred different kinds of garden plants. In particular,
low maintenance were the most important reasons for choosing there was a great deal of difference in people’s preferences for
plants. Other commonly identified reasons included traits such as native plants, with some people strongly disliking native plants.
fragrance and form, abstract characteristics such as beauty, site- People’s preferences were also related to the plants growing in
specific functional roles such as screening or to fill a space, and their gardens. While the statistical relationships between prefer-
personal reasons such as childhood memories. Very few alternative ences and the plants in people’s gardens were generally weak, they
answers were provided by respondents, and only one ‘cut flowers’ were much stronger for residents who had been living at the same
was provided more than once (by two respondents). There were address for more than 5 years.
significant (Tukey’s HSD P < 0.05) differences in some respondent
preferences, with respondents in the Modern cluster the most likely
to choose plants because they were native, drought tolerant or bird- 4.1. Relationship between preferences and plants in gardens
attracting and the least likely to choose plants for their flower traits,
and respondents in the colourful cluster the most likely to choose Ecological studies have shown that plant traits respond to their
plants for their fragrance. physical environment (Cornelissen et al., 2003). This study found
that the traits of plants in gardens are related to people’s prefer-
3.3. Comparison between respondent preferences and their ences, and thus provides evidence that the trait profiles of garden
gardens floras are also responding to their social environment. In the same
way that heterogeneity in the physical environment leads to greater
There were few significant differences in physical character- diversity in native plant communities, our findings suggest that
istics of the front gardens of different respondent clusters. The heterogeneity in the social environment, as we have found in
gardens of respondents in the lush cluster had significantly (Tukey’s people’s preferences, should also lead to greater diversity in cul-
HSD P < 0.05) more individual plants of each taxa in (mean = 7.6 tivated plant communities. This may go some way to explaining
plants per taxa compared with 3.7 for gardens in other respondent the extremely high levels of diversity found in gardens (Thompson
clusters). There were also trends suggesting that taxon richness was et al., 2003).
lower in gardens of respondents in the lush cluster (mean number Objectively measured traits were found to be a useful way of
of taxa per garden = 11.7, compared with 19.4 for gardens in other comparing plant preferences to the plants growing in gardens. The
respondent clusters), that respondent gardens were largest in the use of traits allowed simple statistical techniques such as correla-
colourful cluster (mean area = 152 m2 compared with 101 m2 for tion to be used to compare the strength of the relationship between
gardens in other respondent clusters) and that the proportion of preference and plants for particular groups, such as those who had
garden beds was higher in the gardens of respondents in the mod- resided at the same address for longer. The greater strength of the
ern cluster (mean = 45% compared with 29% for gardens in other relationship for residents who had been at their current address
respondent clusters). There were no noticeable differences in the for more than 5 years is not surprising and can be explained both
age of the houses associated with the gardens in different respon- by people choosing and cultivating the plants they prefer over
dent clusters. time (and removing plants they dislike), and familiarity resulting
There were positive relationships between respondents’ plant in increasing preferences for plants already in their gardens. Also,
preferences and the plants growing in their front gardens. Respon- not all plants are chosen by residents themselves. Nearly half the
dents had higher preferences for a particular taxa if it was growing population of Australia has moved house within the last 5 years
in their garden; respondents with a plant in their garden had a sig- (Australian Bureau of Statistics, 2010), often to a house with an
nificantly (ANOVA P = 0.001) higher preference (mean = 3.8) than existing garden planted by others. Survey respondents had received
those who did not (mean = 3.3). There were moderate to weak cor- 28% of plants from their family, 25% from friends and 2% from
relations between preference for traits and the traits of the plants landscapers. People may also be choosing plants that they do not
growing in people’s front gardens (see Table 4). For all respon- necessarily prefer, but that do provide desired functions such as
dents, the strongest correlation was between proportion of native the provision of screening, shade or edible fruits, or context spe-
taxa in people’s gardens and their preference for native plants, and cific requirements such as filling a space or the ability to survive
there were weaker correlations with leaf and flower traits. For difficult microclimatic conditions (Figs. 1 and 2).
40 D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42

The relationship between preference and plants in gardens is drought tolerance, through familiarity, or through understanding
consistent with the findings of studies at the garden level that both the aesthetics of foliage texture and colour contrast in broader gar-
people’s garden style preferences and self-reported practices are den landscapes. This is also consistent with the findings of Marco
similarly related to a personality trait (van den Berg & van Winsum- et al. (2010) that both physical and social factors were related to
Westra, 2010) or environmental attitudes (Kurz & Baudains, in people’s plant choices. Participants also preferred frequently occur-
press) and that a majority of people preferred a landscape style ring plants, yet it is not clear if this is a response to the familiarity
that matched their own self-reported garden style (Larsen & Harlan, of these plants, or simply that these plants are cultivated more
2006). Our finding that preference for native plants is related to frequently because they have other preferred traits.
their presence in gardens is also consistent with other studies The response to native plants was particularly strong, with some
which have shown that conservation attitudes are related to the people strongly preferring native plants and others strongly dis-
presence of native plants in people’s gardens (Kurz & Baudains, in liking them. This distinction has also been found in other studies
press; Head & Muir, 2006; Zagorski et al., 2004). of Australian gardeners (Head & Muir, 2006; Kurz & Baudains, in
press), and may be related to the relatively large taxonomic and
4.2. A trait-based approach to exploring landscape preference aesthetic differences between Australian native vegetation and the
garden vegetation of Australia’s European colonial past. Other stud-
The most useful traits were those that could be easily measured ies have shown that native plants can be preferred in a garden
using a simple metric such as leaf or flower size. The correlation context with appropriate edge treatment (Nassauer, 1995), in an
between preference ratings and size measurements from stimulus appropriate design (Helfand, Park, Nassauer, & Kosek, 2006) or
photographs allows simple calculation of the strength of the effect when neighbours have native gardens (Nassauer, Wang, & Dayrell,
for different groups (e.g. see the comparison of trait preference 2009). However, this study shows that some people do prefer native
by respondent cluster in Table 3), and should allow straightfor- garden plants without a spatial context. The relatively high educa-
ward comparison across studies even when the taxonomic identity tion levels of the respondent cluster preferring native plants is also
of the stimulus varies. The importance of flower size, the propor- consistent with the findings of studies on preference for native and
tion of canopy covered by flowers, leaf width and leaf colour to naturalistic landscapes (van den Berg & van Winsum-Westra, 2010;
the prediction of preference for most (but not all) people is gen- Williams, 2002; Williams & Cary, 2002; Zheng, Zhang, & Chen,
erally consistent with other research. Specifically, research has 2011). It has been hypothesised that people with higher education
shown that green is a preferred foliage colour (Kaufman & Lohr, levels may have a greater knowledge of environmental issues, and
2004, 2008), that only some people prefer grey foliage (Kendal this may translate into environmentally sensitive behaviours such
et al., 2008), that people prefer broader leaves (Williams, 2002), and as planting trees (Luck, Smallbone, & O’Brien, 2009). Higher educa-
that foliage colour can be an important factor in consumers’ plant tion levels may lead to both higher preferences for native garden
choices (Berghage & Wolnick, 2000; Townsley-Brascamp & Marr, plants and the planting of native species in gardens through and
1994). The importance of flower and leaf characteristics is not sur- understanding of the broader environmental benefits they provide.
prising given the emphasis of these characteristics in horticultural Overall socioeconomic and demographic factors were relatively
plant breeding (Hobhouse, 2002), and it is likely that preferences poor predictors of preference and it is likely that other individual
may be an important reason plants are selected for these traits. level factors not measured in this study contribute more substan-
These findings are partially consistent with habitat theories tially to variations in preference. Other studies at the garden level
of landscape preference, which have argued that people prefer have related preference to personality traits such as Personal Need
plants with characteristics indicating high resource availability for Structure (van den Berg & van Winsum-Westra, 2010) and the
such as large flowers and green foliage (Heerwagen & Orians, type of course students’ are enrolled in (Zheng et al., 2011), or
1995), dislike characteristics that indicate poor quality habitat such related the species composition (Zagorski et al., 2004) and style
as narrow foliage (Williams & Cary, 2002) and selectively breed (Grampp, 1990) of gardens to the gardener’s attitudes towards gar-
plants to increase the psychological benefits provided by them dening effort. Cultural background has been related to variations
(Haviland-Jones & Wilson, 2005). Others have argued that land- in garden preference and practice (Fraser & Kenney, 2000; Kurz &
scape preferences may be culturally based rather than biologically Baudains, in press), although this was not likely to be a significant
based (van den Berg, Vlek, & Coeterier, 1998), and these findings factor in this study due to the relative cultural homogeneity of Bal-
are also partially consistent with the idea that Australian garden- larat. Further research could reveal whether these factors are also
ing preferences and practices are derived from western-European related to garden plant preferences.
traditions. While these theories provide some explanation for pref-
erence for large flowers, broad leaves and green foliage, in this 4.3. Implications for garden practice and policy
study these preferences were not uniform and some people clearly
did not prefer these characteristics. Instead, one group of respon- This study shows that people have diverse preferences that are
dents preferred garden plants with narrow leaves and grey foliage related to the plants in their gardens. Programs that wish to regulate
and strongly preferred some non-visual traits such as nativeness gardening behaviour, such as introducing collective management
and drought tolerance. This group was also the most likely to state of gardens to achieve biodiversity outcomes (Goddard, Dougill, &
they chose plants for these non-visual traits (nativeness, drought Benton, 2010) must consider the diverse preferences of residential
tolerance and attractiveness to birds) (Figs. 1 and 2). Landscape gardeners, particularly in relation to the use of native plants. For
preference theory suggests that while landscape preferences may some people, preferences appear to be related to extrinsic envi-
be evolutionary adaptations or culturally based, preferences can ronmental benefits rather than intrinsic characteristics, and this
also be based on individual knowledge (Bourassa, 1991). The results perhaps provides a greater opportunity to shape people’s gardening
of our study are consistent with the idea that some people have behaviour than top-down approaches. Activities based on factors
been able to overcome their innate or cultural preferences, and that have been shown to affect preference for native plants such
existing knowledge of the origin and tolerances of specific plant as familiarity (Herzog, Herbert, Kaplan, & Crooks, 2000), for exam-
species or the relationship between visual proxies (such as leaf ple through the public planting of native plants, and education to
width and foliage colour) for the these non-visual traits are being increase awareness of their ecological benefits (Gobster, 1999) may
used in the determination of preference. This is perhaps through also lead to the increasing cultivation of native plants in urban
understanding other benefits such as ecosystem functioning or landscapes.
D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42 41

There are also implications for the study of landscape prefer- Bhatti, M. (2006). ‘When I’m in the garden I can create my own paradise’: Homes
ences. The use of objectively measured traits allows comparison and gardens in later life. The Sociological Review, 54(2), 318–341.
Bhatti, M., & Church, A. (2000). ‘I never promised you a rose garden’: Gender, leisure
between studies in different places and with taxonomically dis- and home-making. Leisure Studies, 19(3), 183–197.
tinct floras, and is more suited to generalisation to other contexts. Bourassa, S. (1991). The aesthetics of landscape. Belhaven Press.
Little is known about how specific landscape elements influ- Cornelissen, J. H. C., Lavorel, S., Garnier, E., Diaz, S., Buchmann, N., Gurvich, D. E.,
et al. (2003). A handbook of protocols for standardised and easy measure-
ence human health and wellbeing (Velarde, Fry, & Tveit, 2007), ment of plant functional traits worldwide. Australian Journal of Botany, 51(4),
and the exploration of people’s response to plant traits may 335–380.
also help to reveal the specific characteristics of plants (such as Daniels, G. D., & Kirkpatrick, J. B. (2006a). Comparing the characteristics of front
and back domestic gardens in Hobart, Tasmania, Australia. Landscape and Urban
canopy size, flower size, leaf width or foliage colour) that con-
Planning, 78(4), 344–352.
tribute to people’s health and wellbeing. The findings of this Daniels, G. D., & Kirkpatrick, J. B. (2006b). Does variation in garden characteristics
study show that people respond very differently to different influence the conservation of birds in suburbia? Biological Conservation, 133(3),
326–335.
plant traits, suggesting that the plant characteristics relevant
Dunnett, N., & Qasim, M. (2000). Perceived benefits to human well-being of urban
to health and wellbeing may in fact be different for different gardens. HortTechnology, 10(1), 40–45.
people. Fraser, E., & Kenney, A. (2000). Cultural background and landscape history as factors
Other research exploring patterns of diversity and distribution affecting perceptions of the urban forest. Journal of Arboriculture, 26(2), 106–112.
Gobster, P. (1999). An ecological aesthetic for forest landscape management. Land-
of cultivated plants has largely focussed on census-district level scape Journal, 18(1), 54–64.
social factors (e.g. Kirkpatrick, Daniels, & Zagorski, 2007; Martin Goddard, M., Dougill, A., & Benton, T. (2010). Scaling up from gardens: Biodiversity
et al., 2004). Future research could focus more on individual fac- conservation in urban environments. Trends in Ecology & Evolution, 25(2), 90–98.
Grampp, C. (1990). Social meanings of residential gardens. In M. Francis, & R. Hestor
tors less tied to the geography of census districts and shown to (Eds.), The meaning of gardens (pp. 178–283). Cambridge: MIT Press.
be related to gardening behaviour such as gender (Bhatti & Church, Haviland-Jones, J., & Wilson, P. (2005). An environmental approach to positive emo-
2000), age (Bhatti, 2006), personality (van den Berg & van Winsum- tion: Flowers. Evolutionary Psychology, 3, 104–132.
Head, L., & Muir, P. (2006). Suburban life and the boundaries of nature: Resilience
Westra, 2010), cultural background (Head, Muir, & Hampel, 2004) and rupture in Australian backyard gardens. Transactions of the Institute of British
and conservation beliefs (Zagorski et al., 2004). Geographers, 31(4), 505–524.
Head, L., Muir, P., & Hampel, E. (2004). Australian backyard gardens and the journey
of migration. Geographical Review, 94(3), 326–347.
5. Conclusion Heerwagen, J., & Orians, G. (1995). Humans, habitats and aesthetics. In S. Kellert, &
E. Wilson (Eds.), The biophilia hypothesis (pp. 138–172). Washington, DC: Island
Press.
This study shows that the plants growing in people’s gardens are Helfand, G., Park, J., Nassauer, J., & Kosek, S. (2006). The economics of native
related to their preferences. We found that people’s preferences for plants in residential landscape designs. Landscape and Urban Planning, 78(3),
garden plants are related to both aesthetic traits such as flower size, 229–240.
Heritage Council of Victoria, Building Commission, Heritage Victoria. (2007). What
leaf width and foliage colour and non-visual traits such as native- house is that? A guide to Victoria’s housing styles. Melbourne: Heritage Victoria.
ness and drought tolerance. This provides some evidence that the Herzog, T., Herbert, E., Kaplan, R., & Crooks, C. L. (2000). Cultural and developmental
trait profile of cultivated garden floras is responding to heterogene- comparisons of landscape perceptions and preferences. Environment and Behav-
ior, 32(3), 323–346.
ity in the social environment as well as the physical environment, Hobhouse, P. (2002). Flower gardens. London: Frances Lincoln.
and that the high levels of diversity observed in gardens is at least Kaufman, A. J., & Lohr, V. I. (2004). Does plant color affect emotional and physiological
partly the result of people’s diverse preferences. Objectively mea- responses to landscapes? Acta Horticulturae, 639, 229–233.
Kaufman, A. J., & Lohr, V. I. (2008). Does it matter what color tree you plant? Acta
sured plant traits are a useful way of exploring preference for Horticulturae, 790, 179–184.
particular landscape elements that allow more meaningful com- Kendal, D., Williams, K., & Armstrong, L. (2008). Preference for and performance of
parisons between studies, and may help reveal the plant traits some Australian native plants grown as hedges. Urban Forestry & Urban Greening,
7(2), 93–106.
influencing people’s health and wellbeing. Further research on indi-
Kirkpatrick, J. B., Daniels, G. D., & Zagorski, T. (2007). Explaining variation in front
vidual household level factors may help to explain further variation gardens between suburbs of Hobart, Tasmania, Australia. Landscape and Urban
in patterns of urban vegetation. Planning, 79(3–4), 314–322.
Kurz, T., & Baudains, C. (in press). Biodiversity in the front yard: An investigation
of landscape preference in a domestic urban context, Environment and Behavior.
Acknowledgements doi:10.1177/0013916510385542.
Larsen, L., & Harlan, S. (2006). Desert dreamscapes – residential landscape preference
and behavior. Landscape and Urban Planning, 78(1–2), 85–100.
This project was supported by an Australian Postgraduate Lohr, V., & Pearson-Mims. (2006). Responses to scenes with spreading, rounded and
Award scholarship. Additional support was provided by the Uni- conical tree forms. Environment and Behavior, 38(5), 667–688.
Loram, A., Thompson, K., Warren, P., & Gaston, K. (2008). Urban domestic gardens
versity of Melbourne. We thank the residents of Ballarat who (XII): The richness and composition of the flora in five UK cities. Journal of
participated in this survey and permitted us to survey their gardens. Vegetation Science, 19(3), 321–330.
We are grateful for the thoughtful comments of three anonymous Luck, G., Smallbone, L., & O’Brien, R. (2009). Socio-economics and vegetation change
in urban ecosystems: Patterns in space and time. Ecosystems, 12(4), 604–620.
reviewers and the editor that have resulted in a greatly improved
Marco, A., Barthelemy, C., Dutoit, T., & Bertaudière-Montes, V. (2010). Bridging
manuscript. human and natural sciences for a better understanding of urban floral patterns:
The role of planting practices in Mediterranean gardens. Ecology and Society,
15(2)
References Marco, A., Dutoit, T., Deschamps-Cottin, M., Mauffrey, J.-F., Vennetier, M., &
Bertaudière-Montes, V. (2008). Gardens in urbanizing rural areas reveal an unex-
Acar, C., Acar, H., & Eroğlu, E. (2007). Evaluation of ornamental plant resources pected floral diversity related to housing density. Comptes Rendus Biologies,
to urban biodiversity and cultural changing: A case study of residen- 331(6), 452–465.
tial landscapes in Trabzon city (Turkey). Building and Environment, 42(1), Martin, C., Warren, P., & Kinzig, A. (2004). Neighborhood socioeconomic status is a
218–229. useful predictor of perennial landscape vegetation in residential neighborhoods
Australian Bureau of Statistics. (1999). Standard for statistics on cultural and language and embedded small parks of Phoenix, AZ. Landscape and Urban Planning, 69(4),
diversity. Canberra: ABS. 355–368.
Australian Bureau of Statistics. (2006). Census of population and housing. Canberra: McPherson, G., Nowak, D., Heisler, G., Grimmond, S., Souch, C., Grant, R., et al. (1997).
ABS. Quantifying urban forest structure, function, and value: The Chicago Urban For-
Australian Bureau of Statistics. (2010). Moving house. In: Australian social trends. est Climate Project. Urban Ecosystems, 1(1), 49–61.
Canberra: ABS. Mitchell, R., & Popham, F. (2008). Effect of exposure to natural environment on
Behe, B., & Nelson, R. (1995). Consumer preferences for Geranium flower color, leaf health inequalities: An observational population study. The Lancet, 372(9650),
variegation, and price. Highlights of Agricultural Research, 42(4), 7–8. 1655–1660.
Berghage, R. D, & Wolnick, D. J. (2000). Consumer color preference in New Guinea Nassauer, J. (1995). Messy ecosystems, orderly frames. Landscape Journal, 14(2),
Impatiens. HortTechnology, 10(1), 206–208. 161–170.
42 D. Kendal et al. / Landscape and Urban Planning 105 (2012) 34–42

Nassauer, J., Wang, Z., & Dayrell, E. (2009). What will the neighbors think? Cul- University of Melbourne. (2002). Burnley Plant Database. Melbourne: University of
tural norms and ecological design. Landscape and Urban Planning, 92(3–4), Melbourne.
282–292. van den Berg, A., & van Winsum-Westra, M. (2010). Manicured, romantic, or wild?
R Development Core Team. (2010). R: A language and environment for statistical The relation between need for structure and preferences for garden styles. Urban
computing. Vienna, Austria: R Foundation for Statistical Computing. Forestry & Urban Greening, 9(3), 179–186.
Smith, R., Warren, P., Thompson, K., & Gaston, K. (2006). Urban domestic gardens van den Berg, A. E., Vlek, C. A. J., & Coeterier, J. F. (1998). Group differences in the aes-
(VI): Environmental correlates of invertebrate species richness. Biodiversity and thetic evaluation of nature development plans: A multilevel approach. Journal
Conservation, 15(8), 2415–2438. of Environmental Psychology, 18(2), 141–157.
Sommer, R., & Summit, J. (1995). An exploratory study of preferred tree form. Envi- Velarde, M., Fry, G., & Tveit, M. (2007). Health effects of viewing landscapes – Land-
ronment and Behavior, 27(4), 540–557. scape types in environmental psychology. Urban Forestry & Urban Greening, 6(4),
Tabachnick, B., & Fidell, L. (2001). Using multivariate statistics. Allyn & Bacon. 199–212.
Thompson, K., Austin, K. C., Smith, R. M., Warren, P. H., Angold, P. G., & Gaston, Williams, K. (2002). Exploring resident preferences for street trees in Melbourne,
K. J. (2003). Urban domestic gardens (I): Putting small-scale plant diversity in Australia. Journal of Arboriculture, 28(4), 161–170.
context. Journal of Vegetation Science, 14(1), 71–78. Williams, K., & Cary, J. (2002). Landscape preferences, ecological quality, and biodi-
Todorova, A., Asakawa, S., & Aikoh, T. (2004). Preferences for and attitudes towards versity protection. Environment and Behavior, 34(2), 257–273.
street flowers and trees in Sapporo, Japan. Landscape and Urban Planning, 69(4), Wright, I., Westoby, M., & Reich, P. (2002). Convergence towards higher leaf mass
403–416. per area in dry and nutrient-poor habitats has different consequences for leaf
Townsley-Brascamp, W., & Marr, N. E. (1994). Evaluation and analysis of con- life span. Journal of Ecology, 90(3), 534–543.
sumer preferences for outdoor ornamental plants. Acta Horticulturae, 391, 199– Yabiku, S. T., Casagrande, D. G., & Farley-Metzger, E. (2008). Preferences for landscape
208. choice in a Southwestern desert city. Environment and Behavior, 40(3), 382–400.
Tratalos, J., Fuller, R., Warren, P., Davies, R., & Gaston, K. (2007). Urban form, biodi- Zagorski, T., Kirkpatrick, J. B., & Stratford, E. (2004). Gardens and the bush: Garden-
versity potential and ecosystem services. Landscape and Urban Planning, 83(4), ers’ attitudes, garden types and Invasives. Australian Geographical Studies, 42(2),
308–317. 207–220.
United Nations. (2010). World urbanization prospects: The 2009 Revision. New York: Zheng, B., Zhang, Y., & Chen, J. (2011). Preference to home landscape: Wildness or
United Nations Department of Economic and Social Affairs, Population Division. neatness? Landscape and Urban Planning, 99(1), 1–8.

You might also like