Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Informal Urban Green-Space: Comparison of Quantity and Characteristics in Brisbane, Australia and Sapporo, Japan

  • Christoph D. D. Rupprecht ,

    christoph.rupprecht@griffithuni.edu.au

    Affiliations Environmental Futures Research Institute, Griffith University, Nathan, Queensland, Australia, Griffith School of Environment, Griffith University, Griffith University, Queensland, Australia

  • Jason A. Byrne

    Affiliations Environmental Futures Research Institute, Griffith University, Nathan, Queensland, Australia, Griffith School of Environment, Griffith University, Griffith University, Queensland, Australia

Informal Urban Green-Space: Comparison of Quantity and Characteristics in Brisbane, Australia and Sapporo, Japan

  • Christoph D. D. Rupprecht, 
  • Jason A. Byrne
PLOS
x

Abstract

Informal urban green-space (IGS) such as vacant lots, brownfields and street or railway verges is receiving growing attention from urban scholars. Research has shown IGS can provide recreational space for residents and habitat for flora and fauna, yet we know little about the quantity, spatial distribution, vegetation structure or accessibility of IGS. We also lack a commonly accepted definition of IGS and a method that can be used for its rapid quantitative assessment. This paper advances a definition and typology of IGS that has potential for global application. Based on this definition, IGS land use percentage in central Brisbane, Australia and Sapporo, Japan was systematically surveyed in a 10×10 km grid containing 121 sampling sites of 2,500 m2 per city, drawing on data recorded in the field and aerial photography. Spatial distribution, vegetation structure and accessibility of IGS were also analyzed. We found approximately 6.3% of the surveyed urban area in Brisbane and 4.8% in Sapporo consisted of IGS, a non-significant difference. The street verge IGS type (80.4% of all IGS) dominated in Brisbane, while lots (42.2%) and gaps (19.2%) were the two largest IGS types in Sapporo. IGS was widely distributed throughout both survey areas. Vegetation structure showed higher tree cover in Brisbane, but higher herb cover in Sapporo. In both cities over 80% of IGS was accessible or partly accessible. The amount of IGS we found suggests it could play a more important role than previously assumed for residents' recreation and nature experience as well as for fauna and flora, because it substantially increased the amount of potentially available greenspace in addition to parks and conservation greenspace. We argue that IGS has potential for recreation and conservation, but poses some challenges to urban planning. To address these challenges, we propose some directions for future research.

Introduction

Dunn et al. argue that global conservation efforts depend on the interest people have in nature conservation, an interest formed largely through experiencing nature within the cities that people inhabit[1]. Informal urban greenspace (IGS) such as vacant lots, brownfields and street or railway verges comprise one part of this urban nature. Research has found that IGS can play a role in exposing city dwellers to nature – as recreational space for residents and an alternative to traditional greenspace (e.g. parks and playing fields)[2][4], and as habitat for flora and fauna[5][7]. But we presently lack knowledge about the estimated total quantity of IGS in our cities – a key issue, because the quantity of space likely has a strong influence on its potential for recreation and conservation. Questions yet to be answered include: what proportion the different types of IGS contribute to the total amount of IGS in a city, and how does IGS quantity differ between cities? We know little about the spatial distribution (within a city or in different geographical settings) of IGS, its vegetation structure, or its potential accessibility, which are again important factors determining its potential for recreation and conservation. Compounding this problem, scholars presently lack a shared or agreed definition of these taken for granted socio-ecological spaces (though we propose such a definition below). Such a definition is necessary to ensure that researchers are talking about the same concept and vital to creating an integrated research agenda. Finally, we lack a reliable, comprehensive rapid assessment method that can be applied in different geographical contexts and is useful for estimating IGS quantity as a first step in urban planning initiatives to ‘green’ cities. We take up these tasks in this paper.

This paper reports the results of a study that asked the following four research questions: (1) how does the land use proportion of total IGS and individual IGS subtypes differ between urban core areas in two cities? (2) how do the characteristics (distribution, vegetation structure, accessibility) of IGS differ between urban core areas in two cities? (3) does distance from the city center influence IGS quantity, and (4) how accurate is the IGS land use proportion survey method employed for estimating potential IGS quantity? This study contributes new knowledge in two ways. Our study has for the first time examined how much land likely consists of a wide variety of IGS types in an urban core. Second, it represents the first comprehensive examination comparing IGS quantity and type within the urban core area of two cities, potentially allowing scholars to examine IGS composition and quantity in other geographical settings.

Methods and Data Collection

Informal urban greenspace (IGS) definition and typology

Cities consist of a patchwork of different spaces, from densely built areas to green space such as urban forests or parklands. But besides these exist also more ambiguous, ‘liminal’ vegetated spaces, that Jorgensen and Tylecote refer to as ‘ambivalent landscapes’ [8]. This heterogeneous group of vacant lots, railway verges, utility corridors and waterway embankments is often overgrown with spontaneous vegetation [7], and is managed only to a limited extent (e.g., vegetation removal to protect power lines from overgrowth). They share ambiguities in land tenure, conservation status, maintenance regimes, use, regulation and legitimacy [9], and are best characterized as liminal spaces. Even street verges and suburban lawns can be liminal. While they may have been planted originally, they are oftentimes a mix of intentionally planted and opportunistic species. Their maintenance level is similar to that of backyard gardens, and depends upon many factors, such as feelings of ownership, cultural beliefs, age, and level of neighbor's surveillance [10], [11]. The concept of liminality is derived from several disciplines, but is well-established in the urban geography literature [12], [13]. It refers to a state of ‘betweenness’, intermediacy, or ambiguity of being – the ‘indeterminacy of loose space’, as Franck and Stevens call it [14]. Liminal spaces are ‘at the margins’, characterized by emergence and flux, fluidity and malleability, and are neither segregated nor uncontained[15].

This liminality presents a challenge for quantitatively surveying such spaces, which we aim to address by proposing a provisional, non-exclusive definition and typology of a form of liminal green spaces we term ‘informal urban green space’ (IGS). For the purpose of this study, we have defined IGS as an explicitly socio-ecological entity, rather than a solely biological or cultural object. IGS consists of any urban space with a history of strong anthropogenic disturbance that is covered at least partly with non-remnant, spontaneous vegetation [5][7] and has a history of strong anthropogenic disturbance. It is neither formally recognized by governing institutions or property owners as greenspace designated for agriculture, forestry, gardening, recreation (either as parks or gardens) or for environmental protection (the typical purposes of most greenspace). Nor is the vegetation contained therein managed for any of these purposes. Any use for recreation is typically informal and transitional (e.g. unsanctioned verge gardening).

IGSs differ in their management (e.g. access, vegetation removal, stewardship), land use and site history, their scale and shape, soil characteristics and local urban context. For example, a small brownfield may be similar to a vacant lot in appearance and size, but their different land use history, vegetation removal periods, and urban context distinguish them. We identified nine potentially different subtypes of IGS: street verge, lot, gap, railway, brownfield, waterside, structural, microsite and power line IGS (Table 1, Figure 1, Figure 2). The subtypes are not exclusive, thus an IGS area may be categorized as multiple subtypes (e.g. street verge and gap). Because this typology recognizes the variety of non-traditional greenspace, it provides a better basis to analyze the implications of IGS for planning and conservation than broad terms such as “wasteland” or “derelict land”. The distinction between IGS and formal greenspace is not binary, but rather characterized by a gradient of informality: formal recognition as recreational space by the owner provides a criterion to identify a local-government owned vacant lot covered with mowed lawn as IGS, but a low maintenance “wild” private garden as formal greenspace. Secondary-growth urban forests (rather than e.g. small patches of woody vegetation on a brownfield) represent a borderline case, but in most cases such forests are recognized for silvicultural or recreational value and thus excluded from the definition of IGS used in this article. For the IGS area survey we only recorded IGS larger than one square meter and therefore also excluded microsite IGS.

thumbnail
Figure 1. Photos of informal greenspace types following typology in Table 1.

Street verges: A) Spontaneous herbal vegetation on sidewalk (Sapporo, Japan), B) Unused, highly maintained nature strip with mix of planted and spontaneous vegetation (Brisbane, Australia), C) Spont. herbal vegetation between street and sidewalk (Sapporo). Lots: D) Former residential vacant lot, remains of garden structure still present (Sapporo), E) Long-term vacant lot in residential area (Brisbane), F) Former residential, long-term vacant lot, “no trespassing” sign (Nagoya, Japan). Gap: G) Space with spontaneous herbal vegetation between two buildings, informal storage use (Sapporo), H) Gap with rudimentarily blocked access in front of building (Sapporo), I) Vegetated gap in sealed surface around fence in industrial zone (Brisbane). Railway: J) Annual grass in verge between rail track and street (Sapporo), K) Vegetated cliff next to rail track (Brisbane), L) Vegetated verge and inter-track space (Sapporo). Brownfield: M) Publicly-owned, large vacant tract with grassland and single trees (Sapporo), N) Old city quarter, overgrown former ceramics factory lot (Tokoname, Japan), O) Vegetated area on municipal land for disaster preparation material storage in urban fringe (Sapporo).

http://dx.doi.org/10.1371/journal.pone.0099784.g001

thumbnail
Figure 2. Photos of informal greenspace types following typology in Table 1 (cont.).

Waterside: A) Vegetation on soil deposits in concreted river bed (Nagoya), B) Spontaneous vegetation and informal agricultural use of flood-protection stream banks (Sapporo), C) Spontaneously vegetated anthropogenic river banks (Sapporo). Structural: D) Creeping vines on industrial building (Nagoya), E) Overgrown bridge (Nagoya), F) Concrete soil retention wall completely covered in ivy (Sapporo). Microsite: G) Vegetated crack in asphalt on parking lot (Sapporo), H) Vegetation between two sidewalk plates (Brisbane), I) Plant growing out of degraded traffic cone remains (Nagoya). Powerline: J) Powerline reserve in industrial zone (Brisbane), K) Vegetated area around powerline pylon (near Osaka, Japan), L) Vegetated area around powerline pylon (Sapporo).

http://dx.doi.org/10.1371/journal.pone.0099784.g002

Study locations

Brisbane (Queensland, Australia) and Sapporo (Hokkaidō, Japan) were chosen as case study cities, because research that examines IGS outside of Europe and the USA is relatively scarce. The two case study cities have similarities and differences that lend them well to comparison (Table 2); they thus provide excellent opportunities for a cross-cultural research. Both cities are relatively young (being founded in the 19th century) and they saw most of their growth during the 20th century, especially in the post-second world-war period. Their close geographical size is complemented by a similar urban morphology. Both cities are built around a dense central business district, are situated near to the coast and upland regions, and are intersected by a central river (Figure 3). These similarities contrast with differences in population density, population growth forecasts, and available parks and other greenspaces.

thumbnail
Figure 3. Study locations including sampling sites: Brisbane, Australia (left) and Sapporo, Japan (right).

http://dx.doi.org/10.1371/journal.pone.0099784.g003

While Sapporo has seen rapid growth throughout the second half of the 20th century and now has a population of about 1.9 million, its population is now stagnating and is predicted to decline in the future. In contrast, Brisbane has a population of around 1 million but is still growing quickly (Table 2). This difference in population development is of particular interest as both expanding cities [16] and shrinking cities [17] have important impacts on urban greenspace provision.

In both cities, formal greenspace consists of networks of over 2,000 public parks, most of them small local parks. Brisbane has 3,290 ha of local parkland (32 m2/capita), whereas Sapporo has 2,345 ha (12.3 m2/capita) (Table 2). All parks in Brisbane form an area of 11,840 ha (115 m2/capita), while all greenspace in Sapporo forms an area of 5,508 ha (28.9 m2/capita). These areas include forested hillsides in the southwest of both cities, providing recreational benefits to residents and habitat to wildlife.

Research design

To be able to measure the proportion of land use consisting of IGS and compare it between the survey areas in Brisbane and Sapporo, we used a systematic grid sampling design[18], [19]. We placed 121 sampling sites of 50 m by 50 m each on the intersecting lines of a 10 km by 10 km grid, centered on the city centers (Figure 4, File S1 Sapporo sampling sites, File S2 Brisbane sampling sites). Surveying only the central area of a city rather than the whole allowed us to assess a large area despite limited resources, while still covering most of the densely populated areas where access to greenspace may be difficult[16] (Figure 5). This kind of rapid assessment technique can provide an efficient estimate of land use proportions, and can later be followed up with a more detailed, finer resolution assessment if necessary. The General Post Office (Brisbane) and Sapporo City Office (city hall) were chosen as city centers, following common practice in Australia[20], [21] and Japan[22], [23]. There is no internationally accepted method for determining a city center. There was a one-kilometer distance between any two adjacent sampling sites. Each 2,500 m2 sampling site was divided into 25 sub-sites of 10 m by 10 m for a total of 3,025 sub-sites to facilitate land use assessment (total surveyed area 302,500 m2 or 0.299% of the square enclosing all sampling sites (101,002,500 m2)).

thumbnail
Figure 4. Research design: sampling sites on gridline intersections, with sub-sites and example of IGS percentage calculation.

http://dx.doi.org/10.1371/journal.pone.0099784.g004

thumbnail
Figure 5. Survey areas and population density of study locations: A) Sapporo, B) Brisbane.

http://dx.doi.org/10.1371/journal.pone.0099784.g005

Land use assessment

We used a three-step process to measure the percentage of IGS and other land uses. First, we created a geographic information system (GIS) layer with site locations and projected it on publicly available high-resolution aerial photography data (Google Earth in Brisbane, see http://www.google.com/earth/; Microsoft Bing Maps in Sapporo, see http://www.bing.com/maps/). Second, we surveyed land use type in the field for each sub-site and recorded land use percentage for small land use areas assessable on the ground. This was conducted using a measuring tape and visual estimation for inaccessible site parts (physically or marked with entry-forbidden signs). Sub-sites (25 sub-sites 10×10 m each) for all 121 sampling sites per city were created, as smaller sites allow both easier tape measurement and easier visual estimation. Only land use of one square meter or more was recorded, smaller areas were included in adjacent land use. Land use types, changes in land use since production of the aerial photography, building and land use borders were added to printed field data entry sheets containing GIS-layer and aerial photography. Land use was categorized using the IGS typology (Table 1, except microsite IGS) and a customized land use category system (Table S1. Land use category system). This was produced loosely based on the Brisbane land use code system[24], [25] and adapted to suit the project after pilot tests. It was further amended in the field if land uses were encountered that could not be properly recorded with the existing categories (e.g. mixed multi-story land use such as bridges or commercial/residential mixed buildings). Extremely rare land uses were filed under the sub-category name “Other” in the category “Other” (category nomenclature followed a category/sub-category system, e.g. “Private greenspace - garden”), recorded and described in a comment field. We documented the site and its surroundings with photographs. Location data was recorded using a handheld GPS device (Trimble Juno ST) at an accessible part of the site edge or at up to 20 m from a site edge. No permission to carry out this study was necessary, as the survey method was designed to work without direct site access, and was conducted on publicly accessible land only. Sites or parts of sites located on private, military or access-restricted conservation land were surveyed visually from publicly accessibly land if possible (likewise for vegetation structure and accessibility), or surveyed via aerial photography only (see above). Data collection for this paper did not involve endangered or protected species.

For the final step, we individually estimated percentages on paper in each 10×10 m sub-site for each land use category present in the sub-site. One percent of land use in each of those sub-sites equals one square meter. For complex sites, additional support lines were drawn across the aerial photo, dividing each sub-site into four 5×5 m sites that each represented 25%. Where necessary, these were further divided into 12.5% or 6.25% blocks. To improve the quality of percentage estimates we used non-GIS-compatible high-resolution aerial photography by NearMap (Brisbane, see https://www.nearmap.com/), the photographic collection produced in the field, and Google Street View (for orientation purposes, see https://www.google.com/maps/views/streetview). In one case a sampling site had to be revisited to re-assess a present IGS. Automating this percentage calculation using software was considered but deemed not feasible for the limited number of sampling sites, as the variable quality and nature of the aerial photography (e.g. perspective distortions of higher buildings) used would have required sophisticated software and labor-intensive checks. Future studies of a larger number of sampling sites or cities should, however, consider the use of such software tools.

Vegetation structure assessment

For all IGS types, we visually estimated (with the help of measuring tape) vegetation cover percentage of four different vegetation strata (Figure 6): 1) tree layer cover (all vegetation >2 m height), 2) bush layer cover (all vegetation between 1 m–2 m height), 3) herb layer cover (all vegetation under 1 m, if vegetation between 30 cm and 1 m height is present), and 4) ground cover (all vegetation under 30 cm, if vegetation between 30 cm and 1 m is not present). Tree layer and herb layer cover are independent, while herb and ground cover are mutually exclusive and thus cannot exceed 100% combined coverage in one IGS. While IGS was defined as vegetated space, ground cover percentage does not have to reach 100% if ground vegetation cover is patchy and includes bare ground (e.g. 10 m2 IGS area covered to 50% by herbal layer vegetation, 30% ground layer vegetation, and 20% patchy 1∶1 ground layer vegetation/bare ground mix was recorded as 50% herb cover, 40% ground cover). Vegetation cover height of (possibly vertical) structural IGS was measured in a 90° angle from the substrate.

thumbnail
Figure 6. Vegetation structure assessment: tree layer, bush layer, herb layer and ground layer heights.

http://dx.doi.org/10.1371/journal.pone.0099784.g006

Accessibility assessment

We assessed how accessible IGS areas were on a three-level scale derived from prior research into vacant lot accessibility[26], based on the amount of physical or psychological effort necessary to overcome access barriers. As access barriers we included physical barriers such as fences, walls, chains or barbed wire, as well as symbolic barriers such as signs (e.g. “private ground”, “entry forbidden”, “no child play”, Figure 7). IGS were classified as: accessible, if there were no barriers to access, or very low barriers that required only minimal effort to overcome; partially accessible, if a low fence, a “no entry” sign was present, or space was restricted but not too narrow or high to enter and thus required some effort to overcome the barriers; and not accessible, if a high fence, sign warning of injury or other barriers were present that required considerable effort to overcome them.

thumbnail
Figure 7. Barriers to IGS access.

Example photographs: a) IGS inaccessible due to height and missing ladder; b) IGS completely fenced off; c) IGS access restricted by physical (wire) and symbolic barriers (sign).

http://dx.doi.org/10.1371/journal.pone.0099784.g007

Data analysis and statistical methods

We used SPSS (v. 21 and 22, OS X) and R (v. 3.02, OS X) to perform descriptive and inferential statistical analyses. Frequency tables were used to describe quantity of IGS, quantity of IGS types, and IGS characteristics (vegetation structure, accessibility). Initial analysis indicated that the sample data was not normally distributed (P-P plots, skewness and kurtosis tests). We therefore used non-parametric tests, namely a Mann-Whitney U test to test for differences in IGS proportion, and a PERMANOVA test with Euclidean distance matrix to test for differences in IGS type proportions between the two survey areas (using R function adonis, see http://cc.oulu.fi/~jarioksa/softhelp/vegan/html/adonis.html). All statistical tests were performed on sampling site scale (N = 121 per survey area) with data aggregated from the sub-sites. To test whether distance of the sampling site from the city center were linked with IGS quantity or IGS type quantity, we used a Pearson correlation. A p-value of 0.05 or smaller was interpreted as statistically significant.

Method accuracy assessment

To test how accurate the IGS and land use survey method used in this paper was, we compared the results to land use data from GIS data sets supplied by the local city governments. For this purpose, we first combined the geographic features (e.g. polygons representing residential or green space land use) in the city-supplied data sets (ArcGIS 10, UNION), then removed all features outside the smallest possible square containing all sampling sites (ArcGIS 10, CLIP) to calculate the total land use of the features we wanted to compare. In Brisbane, we compared total combined land use percentages of parks (FGSPK), conservation areas (FGSCN), and sports and recreation areas (FGSSR) from our survey with the total greenspace land use percentage from two Brisbane council greenspace data sets (see Table S1 for land use categories). Additionally, we compared total combined residential (all RES categories), garden (PGSGD) and shared greenspace (PGSSG) land use percentages from our survey with the total residential land use percentage of a Brisbane council general land use data set. In Sapporo, we compared park (FGSPK) and sports and recreation area (FGSSR) percentages from our survey with the total non-conservation greenspace land use percentage from a Sapporo City greenspace land use data set. Conservation greenspace was excluded from the comparison in Sapporo because its definition and included greenspace differed substantially between our land use survey and the supplied data set. We then calculated how much the percentages of land use deviate between our land use survey and the supplied GIS data sets.

To check for an accumulation curve and observe the change in land use percentage as sample size increased, we plotted the land use percentage over the number of sites surveyed. Additionally, we plotted the deviation of our land use percentage results from city-supplied datasets (formal greenspace and residential land use in Brisbane, formal greenspace in Sapporo) against the number of sites surveyed. This allowed us to observe what sample size is necessary to achieve a certain level of deviation from the city-supplied datasets.

Results

The surveyed area in Brisbane consisted of 6.3% (19,027 m2) IGS (Table 3), while the surveyed area in Sapporo consisted of IGS to 4.8% (14,559 m2) (Table 4). This difference in IGS proportion was not significant when comparing between the survey areas on site-level (p = .495, N = 242 (121 per survey area), U = 6953.0, z = −.683; for distribution see Figure 8). In the Brisbane survey area, the mean IGS area per site was 157.25 (Mdn = 95, SD = 214.17). In the Sapporo survey area, the mean was 120.32 (Mdn = 41, SD = 186.44). Street verges made up over 80% of IGS in the Brisbane survey area (Table 3), while lots (42.2%) and gaps (19.2%) were the two largest IGS types in the Sapporo survey area (Table 4). In the Sapporo survey area, IGS consisted of more different IGS types and the proportion of individual IGS types showed IGS was more diverse than in the Brisbane survey area (Table 3, 4).The distribution of IGS types between the two survey areas was significantly different (p = .0001, df = 1, Pseudo-F = 19.121, MS = 825762, based on 9999 permutations). In comparison to formal greenspace (parks, sports and recreation, conservation greenspace, and planted verges such as flower-beds) and private greenspace (gardens, shared greenspace, community land, and commercial and industrial greenspace), the area surveyed in Brisbane consisted of more than half as much IGS (6.3%) as formal greenspace (11.6%) and more than one fifth as much IGS than private greenspace (27.4%)(Table 5). In Sapporo, IGS area (4.8%) was almost a third of formal greenspace area (15.4%) and private greenspace (15.0%)(Table 5). Most common non-IGS land use types were small streets (13.3%, INFSS), conservation greenspace (10.7%, FGSCN), and car parks (10.1%, INFCP) in Sapporo, and private gardens (20.7%, PGSGD), small streets (13.6%, INFSS) and residential land use (12.4%, RESLD) in Brisbane (Table S2. Non-IGS land use in Sapporo and Brisbane).

thumbnail
Figure 8. Frequency distribution of IGS land use percentage in sampling sites.

http://dx.doi.org/10.1371/journal.pone.0099784.g008

We found IGS was present in most of the sampling sites in both cities (Figure 9), with obvious exceptions of sites located in areas with other large-scale land use types (e.g. Brisbane river, Mt. Moiwa in the South-West of Sapporo). For vegetation structure, Brisbane IGS had 27.8% mean tree cover, 7.9% mean bush cover, 21.3% mean herb cover, and 72.8% mean ground cover (Table 6). Sapporo IGS had less mean tree cover (6.5%), similar mean bush cover (7.8%), higher herb cover (43.0%) and subsequently lower ground cover (46.0%) (Table 6). For accessibility in Brisbane, 78% of IGS area was accessible, 7% partially accessible and 15% not accessible (Table 7). In Sapporo, the accessible IGS area (68%) and not accessible area (10%) was smaller, compensated by a larger partially accessible IGS area (21%) (Table 6). Lot, gap and brownfield IGS was more accessible in Sapporo, but street verge and waterside IGS in Brisbane (Table 6). When testing whether the total amount of IGS or the amount of an individual IGS type was correlated with distance to the city center, we found no significant correlations between either total IGS or individual IGS type and distance in either surveyed area.

thumbnail
Figure 9. Spatial IGS distribution: percentage of IGS per sampling site in Sapporo (top) and Brisbane (bottom).

http://dx.doi.org/10.1371/journal.pone.0099784.g009

When measuring the accuracy of our land use survey method, we found the combined percentage of parks, conservation and sports and recreation land use in our survey (11.2%) deviated 8.4% from the combined greenspace land use percentage in Brisbane Council datasets (10.4%). The combined percentage of residential, garden and private shared greenspace in our survey (39.5%) deviated -4.2% from the residential land use percentage in the Brisbane Council dataset (41.3%). In Sapporo, the combined percentage of non-conservation greenspace in our survey (4.8%) deviated −6.9% from the non-conservation greenspace percentage in the Sapporo City dataset (5.1%). As a result, the accuracy of our method was over 90% in both cities when comparing land use percentages of around 5% or more with those of official datasets. A visualization of the change in land use percentage with increasing sample size showed that for a common land use type (residential), good accuracy was reached at a sample size of around 70, while for the rare land use types a sample size of around 90 was necessary (Figure 10). A deviation from the city-supplied datasets of less than 10% is reached at 60 sampled sites for the residential land use type, but only at 120 sampled sites for the formal greenspace land use types (Figure 11).

thumbnail
Figure 11. Decrease in deviation of total land use percentage from city-supplied datasets with increasing sample size.

http://dx.doi.org/10.1371/journal.pone.0099784.g011

Discussion and Conclusions

This study has found similar proportions of IGS in both survey areas. While this could indicate other urban areas may contain a similar percentage of IGS, the conclusions we can draw are limited by the sampling design used. The similarity of the study cases (e.g. age and spatial structures of the cities, size and shape of the survey areas; see Methods) may be partly responsible for the similarity in IGS proportions, so results may vary across survey areas with different characteristics. The survey areas we compared differ in their population density (Figure 5) and cultural context. These two factors seemed to have little influence on the proportion of IGS in the survey areas and its accessibility. However, they may explain the differences in IGS types and vegetation structure. For example, higher population density may influence the amount of land dedicated to street verges through planning policy. The rapid growth Brisbane is experiencing may limit the proportion of lot type IGS, as a high demand for land available for development possibly reduces the time land remains vacant before redevelopment. IGS was also widely, but not equally distributed throughout the survey areas (Figure 9). To better understand the factors driving IGS occurrence across different cities, future research should seek to compare IGS in multiple locations. Additionally, possible links between IGS occurrence and factors such as other land use types (e.g. formal green space, industrial, residential) or average income around the site could be explored.

The sampling design used in this study has limitations that make it unsuitable for assessing factors of importance to urban conservation, such as the size of individual IGS or the distribution of IGS sizes. Such information is valuable and has been recorded by prior research into roundabouts[27] and vacant lots[26], but these studies require a different sampling design to studies aiming at measuring the percentage of IGS land use in a survey area. However, for research on the size of individual IGS, linear IGS (e.g. railway verges) and microsites represent challenges as they lack clear boundaries.

Most prior research on IGS has pointed out its potential without knowing what proportion of cities consists of IGS. Having this knowledge allows us to identify some potential policy implications that IGS has for recreational use and conservation, by considering its area compared to other greenspace types. IGS accounts for about 14% of total greenspace in the survey areas (Table 5). This suggests IGS may represent an important source of green space, an aspect further emphasized by the fact that the proportion of IGS present in both survey areas was similar. Furthermore, in both survey areas over 80% of IGS was accessible or partly accessible. However, the limitations of the sampling design discussed above also apply here. Additionally, these findings represent just the first necessary step to understand the potential of IGS for recreation and IGS. Further research is needed to clarify the degree to which IGS is used by residents, whether accessible sites are actually accessed, and which factors may influence residents' perception and appreciation of IGS. These questions are particularly important, because IGS is different from formal greenspace. It lacks common park facilities such as seating or toilets, its informal nature can cause especially adults to perceive it as unsafe or dangerous[28], and the liminal nature of these spaces may limit the degree to which they can be planned – a challenge for urban planners. Yet, being different can also be an advantage. IGS can offer residents an alternative experience to formal greenspace[2], [29], [30], such as opportunities for children to test themselves in a non-controlled environment[31].

The results also have implications for urban conservation. Research has shown IGS plays a role in providing habitat to fauna and flora[32], [33] as well as in connecting habitat in and between cities[34], [35]. The amount of IGS we found (three times that of conservation greenspace in Brisbane and close to 50% in Sapporo survey areas), its distribution throughout the survey areas, and the relatively complex vegetation structure (Table 6) suggest IGS' role for urban conservation may be more important than previously assumed. Sites with IGS completely surrounded by sites without IGS were rare, suggesting a potential connectivity for species present in IGS (Figure 9). The lack of difference in IGS proportion between the survey areas raises the possibility that a similar percentage of IGS may also be available for conservation in other cities, although the limits of the sampling design discussed above need to be taken into account. The differences in vegetation structure and individual IGS types also imply its actual conservation potential likely depends on its local characteristics. There are concerns about the opportunistic type of species IGS tends to favor[36], and the possible function of IGS as a reservoir or corridor for biological invasions[37]. But research has not always confirmed such connectivity for invasive species[35], and many opportunistic species are adapted to their local environment – they represent what could be called the “de facto native vegetation of the city” [7], [38]. As Dunn et al. suggest, the opportunities for nature experience IGS can provide to residents may also be vital for conservation efforts, even beyond urban areas[1].

The potential importance of IGS for urban recreational use and urban conservation has implications for urban and environmental planning. Planners may need to re-think their negative view of ‘vacancy’ in the urban landscape[39] and acknowledge the benefits of residents' informal creativeness[40]. Expectations for parks, conservation greenspace and private greenspace such as gardens as sole providers of recreational and conservation benefits might need to be reevaluated. Reducing barriers to IGS access[26] or reusing IGS as community gardens[41] are ways to improve IGS utility for residents, but planners should refrain from too much intervention. Some scholars assert that IGS does not necessarily need to be tamed, but should be valued for its informality, ambivalence and special aesthetic [14], [29]. In the case of some IGS, freedom of purpose can mean freedom from purpose (e.g. abandoned lots), while other IGS (e.g. utility corridors, railway verges) have purpose but may have to capacity to accommodate additional, informal use. On the other hand, IGS may have negative effects for residents, such as damage to structures caused by spontaneous vegetation [42], [43]. The liminality and legally non-public status of many IGS may also present a challenge to planners, as their influence on IGS characteristics such as accessibility is likely limited. It depends not only on the cooperation of the formal space owners, but issues such as liability for injury involve third parties such as insurance companies[44].

The results of our study emphasize more research on IGS is needed to unlock its full potential. Examining how IGS is influenced by its socio-ecological context would be a valuable starting point for future studies. A quantitative examination of how residents use and perceive IGS could provide into the social aspects. A cross-cultural comparison seems particularly promising, as concepts such as public space may be interpreted differently depending on the cultural context[45]. A possible direction for research on the ecological side could be a comprehensive examination of the value of IGS for biodiversity, either in the form of a systematic literature review providing a synthesis of the many studies focusing on only one or a few IGS types, or in the form of field studies taking into account all IGS types identified in this paper. Finally, replications of the study conducted in this paper in geographic locations around the world would improve our global knowledge of global IGS distribution and provide valuable input for planning policy.

Supporting Information

Table S1.

Land use category system.

doi:10.1371/journal.pone.0099784.s001

(XLSX)

Table S2.

Non-IGS land use in Sapporo and Brisbane.

doi:10.1371/journal.pone.0099784.s002

(XLSX)

File S1.

Sapporo sampling sites.

doi:10.1371/journal.pone.0099784.s003

(KMZ)

File S2.

Brisbane sampling sites.

doi:10.1371/journal.pone.0099784.s004

(KMZ)

Acknowledgments

We are deeply grateful to Yumi Nakagawa for her invaluable support with data collection and data entry. We thank Jennifer Garden and Jean-Marc Hero for their advice on research design, Alex Lo for helpful comments on the manuscript, Mariola Rafanowicz for assistance in GIS analysis and obtaining data sets, and two anonymous reviewers for their detailed and helpful comments. We also thank the Brisbane City Council and Sapporo City Department of Environment Section Greenspace for providing green space and land use data sets.

Author Contributions

Conceived and designed the experiments: CR JB. Performed the experiments: CR. Analyzed the data: CR. Wrote the paper: CR JB.

References

  1. 1. Dunn RR, Gavin MC, Sanchez MC, Solomon JN (2006) The Pigeon Paradox: Dependence of global conservation on urban nature. Conserv Biol 20: 1814–1816 doi:–10.1111/j.1523–1739.2006.00533.x.
  2. 2. Campo D (2013) The accidental playground. Fordham University Press.
  3. 3. Unt A-L, Travlou P, Bell S (2013) Blank Space: Exploring the sublime qualities of urban wilderness at the former fishing harbour in Tallinn, Estonia. Landscape Research: 1–20. doi:10.1080/01426397.2012.742046.
  4. 4. Jorgensen A, Keenan R, editors (2012) Urban wildscapes. Routledge.
  5. 5. Kowarik I (2011) Novel urban ecosystems, biodiversity, and conservation. Environmental Pollution 159: 1974–1983 doi:10.1016/j.envpol.2011.02.022.
  6. 6. Lososová Z, Horsák M, Chytrý M, Cejka T, Danihelka J, et al. (2011) Diversity of Central European urban biota: effects of human-made habitat types on plants and land snails. J Biogeogr 38: 1152–1163 doi:–10.1111/j.1365–2699.2011.02475.x.
  7. 7. Del Tredici P (2010) Spontaneous urban vegetation: Reflections of change in a globalized world. Nat Cult 5: 299–315 doi:10.3167/nc.2010.050305.
  8. 8. Jorgensen A, Tylecote M (2007) Ambivalent landscapes—wilderness in the urban interstices. Landscape Research 32: 443–462 doi:10.1080/01426390701449802.
  9. 9. McLain RJ, Hurley PT, Emery MR, Poe MR (2014) Gathering “wild” food in the city: Rethinking the role of foraging in urban ecosystem planning and management. Local Environment 19: 220–240 doi:10.1080/13549839.2013.841659.
  10. 10. Head L, Muir P (2006) Suburban life and the boundaries of nature: Resilience and rupture in Australian backyard gardens. Trans Inst Br Geog 31: 505–524 doi:–10.1111/j.1475–5661.2006.00228.x.
  11. 11. Trigger DS, Head L (2010) Restored nature, familiar culture: Contesting visions for preferred environments in Australian cities. Nat Cult 5: 231–250 doi:10.3167/nc.2010.050302.
  12. 12. Howitt R (2001) Frontiers, borders, edges: Liminal challenges to the hegemony of exclusion. Australian Geographical Studies 39: 233–245 doi:10.1111/1467-8470.00142.
  13. 13. Davison A (2008) The trouble with nature: Ambivalence in the lives of urban Australian environmentalists. Geoforum 39: 1284–1295 doi:10.1016/j.geoforum.2007.06.011.
  14. 14. Franck KA, Stevens Q, editors (2007) Loose space: Possibility and diversity in urban life. Routledge.
  15. 15. Moran D (2011) Between outside and inside? Prison visiting rooms as liminal carceral spaces. GeoJournal 78: 339–351 doi:10.1007/s10708-011-9442-6.
  16. 16. Byrne J, Sipe N, Searle G (2010) Green around the gills? The challenge of density for urban greenspace planning in SEQ. Australian Planner 47: 162–177 doi:10.1080/07293682.2010.508204.
  17. 17. Haase D (2008) Urban ecology of shrinking cities: An unrecognized opportunity? Nat Cult 3: 1–8 doi:10.3167/nc.2008.030101.
  18. 18. Royal Geographic Society (n.d.) Royal Geographic Society. Available: http://www.rgs.org. Accessed 2014 May 6.
  19. 19. Hirzel A, Guisan A (2002) Which is the optimal sampling strategy for habitat suitability modelling. Ecological Modelling 157: 331–341 doi:10.1016/S0304-3800(02)00203-X.
  20. 20. Phinn S, Stanford M, Scarth P, Murray AT, Shyy PT (2002) Monitoring the composition of urban environments based on the vegetation-impervious surface-soil (VIS) model by subpixel analysis techniques. Int J of Remote Sensing 23: 4131–4153. doi: 10.1080/01431160110114998
  21. 21. Haby MM, Peat JK, Mellis CM, Anderson SD, Woolcock AJ (1995) An exercise challenge for epidemiological studies of childhood asthma: Validity and repeatability. European Respiratory Journal 8: 729–736. doi: 10.1183/09031936.94.07010043
  22. 22. Glickman NJ, Oguri Y (1978) Modeling the urban land market: The case of Japan. Journal of Urban Economics 5: 505–525 doi:10.1016/0094-1190(78)90006-2.
  23. 23. Tobita T, Iai S, Kang GC, Konishi Y (2012) Observed damage of wastewater pipelines and estimated manhole uplifts during the 2004 Niigataken Chuetsu, Japan, Earthquake Reston, VA: American Society of Civil Engineers. pp. 1–12. doi:10.1061/41050(357)75.
  24. 24. Brisbane City Council (2010) Brisbane council land use codes.
  25. 25. Brisbane City Council (2010) City Plan 2000. Brisbane City Council.
  26. 26. Hayashi M, Tashiro Y, Kinoshita T (1999) A study on vacant lots enclosed by fences in relation to urbanization. Landscape Research Japan 63: 667–670 doi:10.5632/jila.63.667.
  27. 27. Helden AJ, Leather SR (2004) Biodiversity on urban roundabouts—Hemiptera, management and the species-area relationship. Basic and Applied Ecology 5: 367–377 doi:10.1016/j.baae.2004.06.004.
  28. 28. Rink D, Herbst H (2011) From wasteland to wilderness - aspects of a new form of urban nature. In: Richter M, Weiland U, editors. Applied urban ecology: A global framework. Chichester, UK: Wiley-Blackwell. pp. 82–92. doi:10.1002/9781444345025.ch7.
  29. 29. Foster J (2013) Hiding in plain view: Vacancy and prospect in Paris' Petite Ceinture. Cities. doi:10.1016/j.cities.2013.09.002.
  30. 30. Nohl W (1990) Gedankenskizze einer Naturästhetik der Stadt. Landschaft und Stadt 22: 57–67.
  31. 31. Mugford K (2012) Nature, nurture; danger, adventure; junkyard, paradise; the role of wildscapes in children's literature. In: Jorgensen A, Keenan R, editors. Urban Wildscapes. Routledge. pp. 80–96.
  32. 32. Robinson SL, Lundholm JT (2012) Ecosystem services provided by urban spontaneous vegetation. Urban Ecosystems 15: 545–557 doi:10.1007/s11252-012-0225-8.
  33. 33. Meffert PJ, Dziock F (2012) What determines occurrence of threatened bird species on urban wastelands? Biological Conservation 153: 87–96 doi:10.1016/j.biocon.2012.04.018.
  34. 34. Francis RA, Hoggart SPG (2012) The flora of urban river wallscapes. River Res Applic 28: 1200–1216 doi:10.1002/rra.1497.
  35. 35. Penone C, Machon N, Julliard R, Le Viol I (2012) Do railway edges provide functional connectivity for plant communities in an urban context? Biological Conservation 148: 126–133 doi:10.1016/j.biocon.2012.01.041.
  36. 36. Wittig R, Becker U (2010) The spontaneous flora around street trees in cities—A striking example for the worldwide homogenization of the flora of urban habitats. Flora 205: 704–709 doi:10.1016/j.flora.2009.09.001.
  37. 37. Pyšek P, Hulme PE (2005) Spatio-temporal dynamics of plant invasions: Linking pattern to process. Ecoscience 12: 302–315 doi:10.2980/i1195-6860-12-3-302.1.
  38. 38. Del Tredici P, Pickett STA (2010) Wild urban plants of the Northeast: a field guide. Cornell University Press.
  39. 39. Corbin CI (2003) Vacancy and the landscape: Cultural context and design response. Landscape Journal 22: 12–24 doi:10.3368/lj.22.1.12.
  40. 40. Jonas MC (2007) Private use of public open space in Tokyo - A study of the hybrid landscape of Tokyo's informal gardens. Journal of Landscape Architecture 2: 18–29. doi: 10.1080/18626033.2007.9723385
  41. 41. Acres: About Us (n.d.) 596acresorg. Available: http://www.596acres.org/en/about/about-596-acres/. Accessed 13 August 2013.
  42. 42. Melander B, Holst N, Grundy AC, Kempenaar C, Riemens MM, et al. (2009) Weed occurrence on pavements in five North European towns. Weed Res 49: 516–525 doi:10.1111/j.1365-3180.2009.00713.x.
  43. 43. Lisci M, Monte M, Pacini E (2003) Lichens and higher plants on stone: a review. Int Biodeter Biodegr 51: 1–17 doi:10.1016/S0964-8305(02)00071-9.
  44. 44. De Sousa CA (2006) Unearthing the benefits of brownfield to green space projects: An examination of project use and quality of life impacts. Local Environment 11: 577–600 doi:10.1080/13549830600853510.
  45. 45. Dimmer C (2012) Re-imagining pubic space. In: Brumann C, Schulz E, editors. Urban Spaces in Japan. Urban Spaces in Japan. pp. 74–105.
  46. 46. Byrne J, Sipe N (2010) Green and open space planning for urban consolidation – A review of the literature and best practice. Urban Research Program, Griffith University.
  47. 47. Brisbane City Council (2007) CityShape Implementation Strategy. Brisbane City Council.
  48. 48. Australian Bureau of Statistics (2013) National Regional Profile: Brisbane (C) (Local Government Area). Australian Bureau of Statistics. Available: +http://www.abs.gov.au/ausstats/abs@nrp.nsf/lookup/LGA31000Main+Features12007-2011. Accessed 2013 Oct 7.
  49. 49. Sapporo Kanky?kyoku Midori No Suishinbu (2011) Sapporoshi midori no kihon keikaku. (Sapporo Basic Green Plan.) City of Sapporo.
  50. 50. Sapporo Shich? Seisakujitsu Seisaku Kikakubu (2012) Sapporoshi machizukuri senryaku bishon kanren de-ta. (Sapporo City Town Development Strategic Vision Related Data.) City of Sapporo. Available: http://www.city.sapporo.jp/kikaku/vision/data/newdata.html. Accessed 2013 Oct 20.
  51. 51. Sapporo Kanky?kyoku Midori No Suishinbu (2012) Sapporoshi no kōen, ryokuchi. (Sapporo City parks and green space.) City of Sapporo. Available: http://www.city.sapporo.jp/ryokuka/shiryo/toukei/documents/1-1.pdf. Accessed 2013 Oct 20.