Overview of study design
We selected six urban sites with greenery across an urban gradient in Gothenburg, Sweden (57°42′N, 11°58′E) (Fig. 1). The sites ranged from a residential area with lawns and mainly ornamental vegetation, to suburban woodland with predominantly natural vegetation. Thus the sites included variation in type and amount of urban greenery (see details below). We measured biodiversity to assess the biological variation in each site. People living near to each site were sent a survey on, e.g., their perception of that particular site. The respondents also rated their own environment-related attitude. We analysed various perceptions (see below) in relation to (i) the measured biodiversity on each site and (ii) environment-related attitude.
A total of 2866 households living near the six study sites, located within the city of Gothenburg, with a population of ca. 533,000 (1 Jan 2014; see Statistics Sweden 2016), were sent a survey. People aged over 18 were included in the sample. They were randomly identified from a complete register of population (“Folkbokföringsregister”). After three contacts 1347 replies were obtained; 56.8 % of them from women and 43.2 % men, distributed across six age groups of ≤25 (9.2 %), 26–35 (24 %), 36–45 (12.5 %), 46–55 (14.4 %), 56–65 (21.4 %) and 66+ (18.3 %). Data on aesthetic perception of urban green space, urban green-space-related sound perception and importance of trees and plants for the perception of bird species in urban green space will be reported in this study.
Our six sites differ in type of greenery and building structure close to green space, ranging from suburban woodland to residential area. They are distributed across the city and the areas were selected to represent different types of green space and various levels of biodiversity along an urban gradient (Fig. 1, Table 1).
Titteridamm is a forest site on the outskirts of the city. There are residential areas with row houses and small buildings adjacent to the woodland in the south. Traffic routes surround the area in the other directions. The vegetation is a mixed forest with birch (Betula pendula, B. pubescens), pine (Pinus silvestris) and spruce (Picea abies) as the predominant tree species. The shrub layer consists of e.g. alder buckthorn (Frangula alnis) and rowan (Sorbus aucuparia). The species composition is typical of a natural forest on nutrient-poor ground in the Gothenburg area. Apart from a small area (ca 0.5 ha), close to the houses, with a pond and a paved walkway, the woodland is left unmanaged and without any big trails.
Guldheden is urban woodland surrounded by three-storey buildings, taller towerblocks, a University hospital and local traffic routes. The site is dominated by deciduous trees such as oak (Quercus robur), aspen (Populus tremula) and birch, and by shrubs such as rowan and hazel (Corylus avellana). This is typical woodland on nutrient-rich ground in the Gothenburg area. A few paved walkways through the area permit easy access for the public. But maintenance of the area is minimal, i.e. removal of shrubs close to the walkways.
Sörhallsparken is a newly established park (ca 10 years) with open lawns, a few small ornamental trees (Prunus sp) and groups of bushes in a formal-garden style. However, in the middle of the area, there is a rocky knoll with natural woodland. Predominant trees on the rock are oak, birch and Swedish Whitebeam (Sorbus intermedia). There are newly built residential areas with multi-storey buildings and row houses around most of the area but the southern side of the park is borded by the Göta River.
Änggårdskolonin is an old allotment area that was founded in 1913. The area is situated between a campus area (medical faculty and biology) and a residential area with three-storey buildings. There are 50 small private gardens with huts, a common green space and walkways to be used by the public. Domesticated trees and plants are predominant in the area, e.g. apple (Malus x domestica), Prunus sp., black and red currant (Ribes spp), some crops and multiple ornamental plant species.
Kungsparken is an old park (ca 150 years) in the very centre of the city. The residential areas surrounding the park are multi-storey buildings from the late 19th century. In the north, there is a canal built in the 17th century that delimits the park. There are several busy traffic routes and paved walkways crossing the park. Veteran trees, e.g. oak, lime (Tilia cordata), beech (Fagus silvatica) and several introduced species such as horse chestnut (Aeculus hippocastanum), are predominant in the park. The ground is covered by large lawns with a few ornamental flower beds.
Wieselgrensplatsen is a residential area with three-storey buildings from the 1940s. There are some local roads across the area. Well managed lawns dominate the courtyards. There are also a few trees, e.g. crack willow (Salix fragilis) and maple (Acer sp.), and groups of ornamental plants.
Some parts of the visually dominant biodiversity were measured. Following previous studies (Fuller et al. 2007; Shwartz et al. 2014; Carrus et al. 2015) we focused on the vegetation as a major part of biodiversity being perceived by people. In addition, songbirds were monitored because they can have a great influence on perception of urban sites (Hedblom et al. 2014). We also included bumblebees because they are highly conspicous invertebrates that are noticed by the public (Bjerke and Østdahl 2004).
The vegetation in each site was investigated by counting trees, bushes and herbs in circular plots with a radius of 20 m, 10 m, and 0.28 m, respectively (method used by Swedish national monitoring, see Ståhl et al. 2011). The species richness of trees, bushes and herbs was expressed as number of species per area (trees and bushes per ha, herbs per m2). The number of species was counted irrespective of their origin, i.e. both native and non-native plants were counted and included in the estimate of species richness. A stratified sampling design was used for the vegetation in our study sites. This was done to ensure that the habitat variation in each site was captured in the samples. The number of plots per site varied between 2 and 4 (trees and bushes) and 6 and 12 (herbs) depending on size of the site and the habitat heterogeneity. As can be seen in Table 1, the standard error was typically around 10 % of the mean, indicating an acceptable precision of the measurement on each site (Southwood and Henderson 2000). But in Sörhallsparken and Wieselgrensplatsen, standard error was higher for trees and bushes due to high variation between sampling plots (lawns without trees vs groves with trees). This would not have been changed to any great extent by increasing the number of sampling plots because of the large differences in vegetation composition between sampling strata.
Our survey included two animal groups; species numbers and diversity (Simpson’s index 1/D, see Magurran 2004) of songbirds and bumblebees in each site were estimated by standardized point counts. On each site, birds were counted at two points that were visited three times in April to June 2013. Bumblebees were monitored three times at three points in each site during July and August 2013.
The six sites were redefined into three categories of biodiversity: high, medium, and low. This categorization was done to increase the number of replies per category and was based on the totalled ranking of mean values for species density of trees, bushes, and herbs, and diversity of songbirds and bumblebees (Table 1).
Measures in survey
Participants were asked to estimate their aesthetic perceptions of the nearby urban greenery on a 7-point scale, ranging from 1 (completely disagree) to 7 (completely agree). The measure comprised four statements: “Naturalistic”; “Rich in species”; “Lush”; and “Varied”.
Greenery-related sound perception
Participants were asked to estimate different types of urban greenery-related sounds on a 7-point scale, ranging from 1 (completely disagree) to 7 (completely agree). The measure comprised three statements: “Sounds of nature give me a stronger perception of the site”; “It is important for me to listen to the bird song in the area”; and “Sounds of nature are important for my perception of bird species in the area”.
Importance of trees and plants for the perception of bird species
Participants were asked to respond to two statements on a 7-point scale, ranging from 1 (completely disagree) to 7 (completely agree): “Trees are important to my perception of bird species in the area”; and “Plants are important to my perception of bird species in the area”.
Urban-oriented person attitude
This was measured on a 7-point scale: 1 (completely disagree) to 7 (completely agree) related to the question: “Are you an urban-oriented person, finding pleasure in street life, shopping, and amusements of the city?”
Nature-oriented person attitude
This was measured on a 7-point scale: 1 (completely disagree) to 7 (completely agree) related to the question: “Are you a nature-oriented person, finding pleasure in the sea, woods, and nature?”
Statistical design and analyses
A non-equivalent comparison-group quasi-experimental design (McGuigan 1983) was used. Thus, compared with a “true experiment” (Liebert and Liebert 1995), the inferences drawn about the causal relationships between independent and dependent variables might be considered weaker.
Three independent variables were included: 3 Biodiversity (high, medium, low) × 2 Urban-oriented person (high, low); × 2 Nature-oriented person (high, low). This means that the green space was redefined from six sites to three biodiversity categories of high, medium and low. Scales measuring the two attitudes of urban and nature oriented person were dichotomized into high vs. low conditions, respectively. In order to investigate contrasting positions (high vs. low) of an attitude, it is recommended to split a continuously scaled variable at the median (DeCoster et al. 2009), i.e. H-person vs. L-person. Thus, respondents lower than median were considered to be “low” and those higher than median were considered to be “high” on the respective independent variable of urban- (low vs. high) and nature-oriented (low vs. high) attitude. This was done in order to investigate urban-nature-oriented attitude polarized by “high” vs. “low” individuals respectively (e.g., Knez 2005, 2013; Knez and Thorsson 2006). This does not, however, imply that urban- and nature-oriented attitudes are orthogonal, meaning that a respondent cannot score “high” or “low” on both attitudes. People can indeed appreciate or not both the city and the nature. As shown in Table 2, most of the subjects (543) were not particularly fond of either the city or the nature. However, 165 respondents were shown to estimate themselves as high on both attitudes. It was also shown that 243 and 322 respondents estimated themselves as being high-nature-low-urban and high-urban-low-nature oriented persons respectively.
Participants’ aesthetic perception of urban green space, urban green-space-related sound perception and the importance of trees and plants for the perception of bird species in urban greenery were used as dependent variables.
Multivariate analysis of variance (MANOVA) was used because the three dependent variables involved more than one scale/statement. The independent variables of Biodiversity, Urban-oriented person and Nature-oriented person were treated as between-subject factors. The software IBM SPSS Statistics 22 was used for statistical computations.