Abstract
Residents of Melbourne, Australia (n = 192) were surveyed regarding their preferences for 36 trees presented in black-and-white photographs. The findings indicate preference for medium-sized trees with a globular or oval form. Foliage texture also arose as an important factor, with higher preference for trees with relatively broad or coarse foliage. While there was no overall preference for either native or introduced trees, categories derived statistically from preference ratings suggest this criterion may have been a consideration for many residents. Some findings could not be predicted from the published literature, which has largely been conducted in the Northern Hemisphere.
Street trees provide a wide range of benefits for urban environments. Benefits in air quality, wildlife habitat, property values, shade and aesthetic appeal of streetscapes are widely acknowledged (Dwyer et al. 1992). There is also evidence that urban plantings increase the perceived quality of life in residential areas (Sheets and Manzer 1991) and may enhance feelings of safety (Kuo et al. 1998). Not all street trees are appreciated, however, and selection of suitable street trees remains a significant challenge for city arborists and urban land managers (Sommer et al. 1990). A large body of literature examines the characteristics of preferred trees. With the exception of a small number of cross-national studies (Sommer and Summit 1996; 1997) much of this research has been conducted in North America. The current paper extends this body of research to an Australian context, examining preferences for street trees in residential Melbourne.
Studies conducted in North America have shown high levels of appreciation for trees in urban settings, but some recent studies highlight the potential for cultural variation. Hitchmough and Bonugli (1997) conducted surveys in a Scottish town and found relatively low support for urban plantings. Similarly, a comparison of Canadian respondents from four ethnic backgrounds (Fraser and Kenney 2000) reveals considerable variation in attitudes toward urban trees, which the researchers attribute to the landscape histories of respondents’ cultures of origin. Australian attitudes toward urban trees are largely unknown, and generalization of Northern Hemisphere findings may be problematic because of differences in vegetation, landscape, and cultural values.
The issue of resident preferences for street trees is as critical in Australia as elsewhere. While management issues such as invasive roots and debris clearly are important to resident evaluations (Sommer et al. 1989), visual characteristics have a strong influence on the relative appeal of street trees. Reasonably consistent findings have arisen regarding the relationship between tree size, tree form, and resident preferences. In general, large trees tend to be preferred over small trees (Kalmbach and Kielbaso 1979; Heerwagen and Orians 1992; Sommer et al. 1993; Sommer and Summit 1995). A smaller number of studies suggest an alternative perspective. In contrast with their earlier work, a study by Summit and Sommer (1999) reports an overall preference for smaller trees. Orland et al. (1992) examined the relationship between attractiveness, house price, and tree size. They found that small to medium trees were preferred over larger trees but suggest this finding may be accounted for by an association between small trees and new homes.
A number of studies have been conducted regarding preferred tree forms. These studies consistently point to a preference for spreading and globular forms over more columnar and conical forms (Legg and Hicks 1979 cited in Summit and Sommer 1999; Sommer and Summit 1995, 1996b). On a similar issue, Heerwagen and Orians (1992) found a nonsignificant preference for trees that are broader than they are tall.
A number of theories attempt to explain human preferences for trees, but Sommer and Summit (1995) claim empirical studies of tree preferences are most consistent with evolutionary accounts of environmental aesthetics. Appleton (1990) hypothesizes that humans attend most to characteristics of trees that enhance human survival. Trunk height and width, and canopy height and shape, influence the potential for humans to see hazards or opportunities without being seen by predators (characteristics formally known as prospect and refuge). In related work, Orians (1998) predicts preference for tree forms characteristic of high-quality east African savanna; that is, acacia (Acacia spp.) trees with low bifurcation of trunk; spreading canopies; and small, compound leaves. Several studies have reported support for Orians’ habitat theory (Heerwagen and Orians 1992; Sommer and Summit 1995, 1996b; Sommer 1997).
Little is understood about human response to tree foliage. Heerwagen and Orians (1992) tested their hypothesis of preference for trees with small, compound leaves in a study of the characteristics and occurrence of various cultivars of Japanese maples (Acer palmatum). They found a strong tendency toward selection of varieties with deeply lobed leaves, which are effectively compound. Some exploratory research by Williams and Cary (2002; see also Cary and Williams 2000) also suggests tree foliage is important. We studied perception of native vegetation in rural landscapes and found that environmental preferences were related to the dominant tree species. Woodlands comprised of non-eucalypt species, such as trees from the Allocasuarina family (Buloke and Sheoke), were significantly less preferred than eucalypt woodlands. Interviews revealed that nearly one-third of participants attributed their dislike of these species to the distinct foliage (cladodes) and dark bark of these trees. The canopy of healthy Allocasuarina trees was often described as messy or scruffy, while the bark was described as fire damaged and the foliage as unhealthy. The leafless foliage of Allocasuarina trees can be seen as an adaptation to harsh conditions including rocky, sandy, and clay soils (Costermans 1981); lower preference for these trees therefore appears consistent with functional and habitat theories of landscape preference.
There is a clear need for better understanding of the tree characteristics that influence resident preferences. While evolutionary theory may provide an effective account for overall patterns of preference, researchers must be mindful of the diverse social norms and expectations that also influence resident response to street trees. While theory and empirical studies highlight the universal patterns of aesthetic response, street tree managers have reported dealing with the extremes of diverse community expectations. Cultural or ethnic differences appear to be most significantly related to these expectations (Fraser and Kenney 2000). Few gender-, age-, or training-related differences have been reported. Kalmbach and Kielbaso (1979) reported no significant differences associated with these factors. Hitchmough and Bonugli (1997) found that men generally were more supportive of street trees in a Scottish town than were women. Retired people expressed less support for street trees than employed people, but this difference was not significant. Age-related differences also were reported by Sommer and colleagues (1989); older residents had a lower opinion of European elms (Ulmus) and London plane trees (Platanus × acerifolia) than did younger residents. Sommer et al. (1992; 1993) report greater similarity than difference in the responses of people with different professional and educational backgrounds. Further research is required to better assist street tree managers in planning trees for specific neighborhoods.
Studying relative preference for a range of trees presents a logistic challenge for researchers that has been solved in a variety of ways. Some researchers have substituted actual trees with sketches of trees (e.g., Sommer and Summit 1996b; Sommer 1997), or slides or photographs (e.g., Kalmbach and Kielbaso 1979). Still others have preferred to survey residents regarding the actual trees outside their home (Sommer et al. 1989). Sommer and colleagues (1993) assert the need to use multiple methods to properly understand community response to street trees. In the current study, perceptions are examined via black-and-white photographs of trees. The use of photographic simulations has long been considered a valuable tool in environmental assessment research (Craik and Feimer 1992), and the validity of this technique has been established through several studies (Shuttleworth 1980; Stamps 1990). The use of photographic surrogates enabled participation of a diverse cross-section of Melbourne residents and enhanced the capacity to explore relative preference for a large number of trees.
This study explores preferences for street trees in Melbourne, a large metropolitan city in southeastern Australia. Frank et al. (2001) surveyed 18 of Melbourne’s 31 local councils to provide a snapshot of the city’s street tree population. Street plantings are predominantly small trees and shrubs, with a smaller proportion of medium and large trees. Just over 60% of trees are introduced from outside of Australia. The majority of trees are evergreen, although only 8% of trees are conifers. The native Eucalyptus, Acacia, and Melaleuca genera are the most common tree groups, followed by introduced Prunus and Quercus genera. Within this context, the current study seeks to identify the visual characteristics of street trees that contribute to resident preferences and to examine the relationship between street tree preferences and a small number of demographic characteristics of respondents.
MATERIALS AND METHODS
The significance of visual characteristics for preference was explored in two ways. First, the study utilized an exploratory approach adapted from the category identification method (CIM) developed by Kaplan and Kaplan (1989). This approach required collection of preference ratings from a relatively large number of respondents regarding a relatively large number (>30) of elements (in this instance, trees). Statistical techniques were then applied to identify underlying dimensions of preference and to form categories of elements that people tend to assess in a similar way (i.e., a given individual will tend to rate all scenes within a category in a similar way, although different individuals may express different preferences for the categories). This approach clarifies the relative importance of characteristics such as tree form, foliage, and size in shaping overall preferences. The second approach uses more standard statistical tests to compare mean preferences for trees regarded as having different characteristics relating to size, form, and foliage.
A photo-questionnaire was developed. Tree photographs were taken in Melbourne streets, primarily in southern and eastern suburbs. A large number of photographs was taken, targeting trees that varied in form, foliage, and origin. Photographs were taken during late summer on days of little or high cloud, by a professional photographer using a standard 50-mm-lens camera. Photographs were generally shot at an angle from the opposite side of the street to include some indication of the streetscape.
Thirty-six photographs were selected for inclusion in the photo-questionnaire. Photographs were selected to represent a range of values with regard to origin (native or introduced), form (spreading/globular or columnar/ conical or oval/upright), and foliage (broad or fine). Two orders of presentation were determined randomly. Preference judgments are influenced by context of presentation, including nearby photographs, which tend to provide a point of comparison (Brown and Daniel 1987). Using two orders of presentation helps overcome possible problems of order effect that may confound results.
Photographs were reproduced in black-and-white format and presented four to an A4 page (portrait format) in booklet form. A separate sheet was designed for participant preference ratings of street trees and responses to demographic questions.
Participants received an introductory letter by mail, explaining the project and seeking their participation. A week later, photo-questionnaires were mailed, along with a more detailed explanation of the project, a response form, and reply-paid envelope. Participants completed the questionnaire at a time and place of their own choice. They rated each of the 36 trees on a five-point scale (5 = like very much; 1 = do not like at all). Participants were asked to assess the trees shown in the photographs rather than the qualities of the photographs themselves. They then provided simple demographic information about themselves and were given an opportunity to make a general comment on street trees in their locality. Reminder notices were sent to recipients who had not returned questionnaires three weeks after the initial mailing went out.
Potential participants were identified through random selection from the Melbourne telephone directory. A total of 192 people returned correctly completed forms (93 males, 99 females), providing a response rate of 35%.
RESULTS
Visual Characteristics and Street Tree Preferences
Principal Components Analysis (PCA) with a Varimax rotation was used to identify groups of tree photographs that participants tended to assess similarly. Six components were extracted (determined by scree test), accounting for 57% of variance in preference scores. Loadings on each of these factors were used to assign tree photographs to one of six categories (one tree loaded on several categories and was omitted from this analysis). The characteristics of each group were interpreted with the assistance of four expert judges (arborists and horticulturists). These assessors examined each of the 36 photographs and classified each tree on a number of criteria: tree size, balance, form, foliage texture and size, density of canopy, health, suitability, and how commonly this species was observed as a street tree (familiarity). There was a high level of inter-judge agreement for all criteria except health, suitability, and familiarity. The trees were classified on the remaining criteria based on the majority view of expert assessers.
The characteristics of trees loading on each of the six components are shown in Table 1 and summarized in Table 2*. While the range of descriptors is incomplete, the tables suggest the six tree categories differ on a range of visual and nonvisual characteristics. Trees loading on Category 1 was primarily medium to large, Australian native trees with relatively large, coarse foliage and sparse canopies (Figure 1). Trees loading on Category 2 were almost without exception small- to medium-sized native trees with fine foliage and relatively dense, spreading, or globular canopies (Figure 2). Trees loading on Category 3 were all introduced conifers that were relatively large trees with fine foliage (Figure 3). Category 4 was associated with introduced, deciduous trees, moderate to large, with relatively large and coarse foliage (Figure 4). Categories 5 and 6 were less readily characterized. Category 5 included a small number of primarily conical or columnar trees of relatively small size. Inspection of the photographs suggested these trees also had relatively poor form (Figure 5). Trees loading on the final category were all small trees with a strongly irregular spreading form and relatively coarse foliage. Most had sparse canopies (Figure 6).
Category 1.
Category 2.
Category 3.
Category 4.
Category 5.
Category 6.
Characteristics of trees with six categories derived from survey respondent preference scores.
Summary description and mean preference for six categories of trees.
In summary, consistent with other research, this analysis suggests that tree sizes, origin, form, foliage, balance, and canopy density all contribute to lay preferences for street trees. It is interesting to note the relative importance of foliage size and texture in characterizing the six tree categories because little attention has been paid to this factor in previous research.
Preference and Tree Characteristics
Table 2 shows the mean preferences for the six tree categories. Preferences were relatively low overall, with only categories 2 (small native trees) and 4 (deciduous introduced trees) reaching “moderate” levels of preference (>2.8 on a 5-point scale). Interestingly, these two categories are typified by Acacia and Ulmus species, the two species that cross-national studies conducted by Sommer and Summit (1996b), Sommer (1997), and Summit and Sommer (1999) have shown to be most preferred. The most preferred trees were those in category 4, but this finding is not clearly related to any single characteristic of these trees.
Further analyses were conducted to examine the relationship between tree characteristics and preference. Average preferences were calculated for trees of different classes with regard to size, form, foliage texture, regularity of tree form, and tree origin.
Tree size clearly is an important predictor of preference for street trees. The most preferred trees were those of medium size, with lower preferences for smaller and larger trees (Table 3). This finding contrasts with previous research in this area, which generally has supported higher preference for large trees (Kalmbach and Kielbaso 1979; Schroeder and Cannon 1983; Sommer and Summit 1995).
Mean preference for trees compared by size [Wilk’s Λ (4,188) = .43, p = .00].
Table 4 shows relative preference for trees with spreading, conical/columnar, upright or globular form. The relatively low preference for conical/columnar forms is consistent with previous research (Sommer and Summit 1995; 1996b). In contrast, the significantly higher preference for globular forms over spreading forms is not consistent with research conducted elsewhere.
Mean preference for trees compared by form [Wilk’s Λ (3,189) = .38, p = .00].
Table 5 indicates that higher preferences were found for trees with more coarse foliage (generally broadleaved trees) over trees with fine foliage (generally conifers and some Australian native trees, many with needlelike foliage). Balance of tree form also was important; Table 6 shows significantly higher preference for trees with a regular canopy form.
Mean preference for trees compared by foliage texture [Wilk’s Λ (1,191) = .62, p = .00].
Mean preference for trees compared by balance or regularity of tree form [Wilk’s Λ (1,191) = .54, p = .00].
In southeastern Australia, there is strong interest in the relative acceptance of native and introduced trees in urban streets. Comparisons revealed no significant difference in average preference for these two broad groups (Wilk’s Λ (1,191) = 1.00, p > 0.05), suggesting that visual characteristics such as size, form, and foliage are more important predictors of preference.
Tree Preferences and Demographic Characteristics
Preference for street trees was expected to differ across social groups. This hypothesis was tested in a limited way by examining the relationship between mean preference for each tree category and a number of demographic characteristics. Multiple analysis of variance (MANOVA) found no significant association between gender and street tree preferences (Wilk’s Λ (6,185) = 0.96, p > 0.05). The age of respondents was significantly associated with street tree preferences (Table 7). Older respondents (60 years and older) tended to have lower preference for the two categories most associated with large trees (the large, generally native trees of category 1 and large conifers typical of category 3). Education also was related to street tree preference. Respondents with higher education levels expressed significantly higher preference for the large, primarily native trees of Category 1 (Table 8).
Mean preference for trees, compared by respondent age [Wilk’s Λ (12,368) = .81, p = .00].
Mean preference for trees, compared by respondent education [Wilk’s Λ (12,362=.79, p=.00].
DISCUSSION
This study supports previous research by highlighting a range of visual and nonvisual characteristics associated with lay evaluation of street trees. These characteristics include tree size and form. Exploratory analyses also indicate that tree origin, size and texture of foliage, and density of canopy also underpin residents’ evaluations. The relative importance of foliage for determining assessments suggests more attention should be paid to this characteristic in accounting for tree preferences.
The finding of higher preference for medium-sized trees is not consistent with the majority of previous research in this field. The reason for this difference is not clear. It may reflect cultural expectations unique to the Melbourne population involved in this study, possibly related to predominance of small trees and shrubs in street plantings, or local landscape characteristics such as house sizes, placement of utilities such as power lines, and width of streets and roadside paths. Street trees are assessed in the context of typical arrays, and mediumsized trees may be preferable under local conditions. Alternatively, the finding may represent a temporal shift in attitudes as residents of developed countries in general become more aware of the management problems associated with large trees in urban landscapes (Schroeder and Cannon 1983; Barro et al. 1997). The unexpected finding of preference for trees with a globular form over trees with spreading canopies may relate to similar limitations within the urban landscape.
The study indicates preference for trees with broad or coarse foliage over those with more finely textured foliage. This finding is consistent with other research undertaken in Australia (Williams and Cary 2002) and warrants further investigation. Together these studies raise the possibility that some needle-leaved trees present a visual cue that some people may associate with relatively poor habitat—in this case, environments of extreme cold or aridity—or such trees may be considered to provide less shade than broad-leaved trees. Both explanations are consistent with habitat or functional theories of environmental preferences. This speculation requires much further investigation. Native trees with very fine foliage (visually comparable to Pinus species) are relatively unusual in an Australian urban context, and conifers are not commonly chosen as street trees (Frank et al. 2001); they also may be considered ecologically undesirable. Alternate patterns of preference may arise with Northern Hemisphere populations where fine- or needle-leaved trees may have more positive associations.
The clearest pattern regarding demographic characteristics and street tree preferences is the lower preference of older people for those categories with the largest trees. This finding is consistent with the patterns described in previous research (Sommer et al. 1989; Hitchmough and Bonugli 1997) but suggests that this negative attitude may be related to specific attributes of trees rather than trees in general. The relationship between education levels and higher preference for large native trees possibly relates to broader environmental attitudes of residents. Higher education levels may be linked to more positive views of the importance of native vegetation in the landscape (Williams and Cary 2000). Further qualitative research is required to explore the meaning that residents attribute to particular tree characteristics and to thereby gain a better appreciation of the beliefs and attitudes that shape social response to urban vegetation.
CONCLUSION
An understanding of social preferences for the urban landscape is critical to planning and management of street trees. This study highlights a range of tree characteristics that should be considered in tree selection. In particular, greater empirical and theoretical attention should be paid to the significance of tree foliage. The findings were not particularly well predicted based on review of literature published in international journals. This fact highlights the need for local and regional research to support effective street tree selection.
Acknowledgments
The author would like to acknowledge the generous assistance of colleagues at Burnley College, especially Greg Moore, Jill Kellow, Peter May, Phil Kenyon, Lee Stone, and Michael Green. Nadine Michel contributed to the collection of photographs, and many local government officers assisted in identification of suitable tree locations.
Footnotes
↵*All tables and figures for this article are found on pages 167–170.
- © 2002, International Society of Arboriculture. All rights reserved.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.