ArticlePDF Available

Mapping wild nature areas to identify priority areas for urban rewilding in cities: A process-oriented approach

October 2024
Urban Forestry & Urban Greening 101(7):128549

DOI:10.1016/j.ufug.2024.128549

License
CC BY 4.0

Authors:

Brenda Maria Zoderer

BOKU University

Francesco Vuolo

BOKU University

Urban rewilding initiatives are gaining traction in cities, yet little remains known about the spatial opportunities for transitioning greenspaces into rewilding sites. This paper advances a process-oriented framing of urban wild nature areas (WNAs) and proposes a novel methodological approach for mapping existing WNAs at the city-scale, thereby shedding light on the potentials for urban rewilding across greenspace types and urban environments. Using Vienna as a case study, we identify WNAs as greenspaces predominantly shaped by natural succession processes and devoid of vegetation management. NDVI timeseries derived from Sentinel-2-satellite data for the period 2017 to 2022 are processed using a Random Forest algorithm to distinguish between unmanaged and managed vegetation annually and integrated into a multi-year composite map of ‘urban vegetation management intensity’. Based on this map and a set of objective cut-off values, 1298 WNAs are identified, representing 29.5% of the city’s total urban greenspace and 14.9% of the city’s total area. WNAs are distributed across diverse urban settings and greenspace types, with the majority being found in formal greenspaces such as forests, meadows and parks, and low-density built-up areas. The key strength of the process-oriented approach lies in its ability to detect WNAs dominated by natural succession, regardless of historic origin, greenspace type, and governance structure, thereby providing a baseline against which the full potential for urban rewilding can be evaluated. We demonstrate the methods’ utility in identifying potentials for urban rewilding, emphasising the importance of directing rewilding efforts towards residential greenspaces, urban parks, and street greenery, especially in densely built inner-city districts.

Map of the study site and its coverage by different urban greenspace types (UGS types, own calculation). Data sources: Federal Office of Metrology and Surveying, City of Vienna -https://data.wien.gv.at. Note: Arable land in the southern and eastward part of Vienna was excluded from our analysis and is therefore not shown.

…

continued )

…

Overview of methodological approach.

…

a) and c) RGB composites of the three first PCA components, and b) and d) RGB composites of the 4th, 5th and 6th PCA components. Panel a) and c) highlight the most clear differences in land cover types (i.e., forests, built-up area), while Panel b) and d) show more subtle differenes in land cover types, with a good spatial consistency with actual patterns.

…

Management intensity of urban vegetation in Vienna for the period 2017 to 2022.

…

Figures - uploaded by Brenda Maria Zoderer

Content may be subject to copyright.

Content uploaded by Brenda Maria Zoderer

Content may be subject to copyright.

Original article

Mapping wild nature areas to identify priority areas for urban rewilding in

cities: A process-oriented approach

Brenda Maria Zoderer

a,*

, Christa Hainz-Renetzeder

, Francesco Vuolo

Institute of Landscape Development, Recreation and Conservation Planning, University of Natural Resources and Life Sciences, Peter-Jordan-Straße, 65, Vienna 1180,

Austria

Institute of Geomatics, University of Natural Resources and Life Sciences, Peter-Jordan-Straße, 82, Vienna 1190, Austria

ARTICLE INFO

Keywords:

Wild space

Urban wilderness

Rewilding

Informal greenspace

Urban greenspace

Mapping

ABSTRACT

Urban rewilding initiatives are gaining traction in cities, yet little remains known about the spatial opportunities

for transitioning greenspaces into rewilding sites. This paper advances a process-oriented framing of urban wild

nature areas (WNAs) and proposes a novel methodological approach for mapping existing WNAs at the city-scale,

thereby shedding light on the potentials for urban rewilding across greenspace types and urban environments.

Using Vienna as a case study, we identify WNAs as greenspaces predominantly shaped by natural succession

processes and devoid of vegetation management. NDVI timeseries derived from Sentinel-2-satellite data for the

period 2017–2022 are processed using a Random Forest algorithm to distinguish between unmanaged and

managed vegetation annually and integrated into a multi-year composite map of ‘urban vegetation management

intensity’. Based on this map and a set of objective cut-off values, 1298 WNAs are identied, representing 29.5 %

of the city’s total urban greenspace and 14.9 % of the city’s total area. WNAs are distributed across diverse urban

settings and greenspace types, with the majority being found in formal greenspaces such as forests, meadows and

parks, and low-density built-up areas. The key strength of the process-oriented approach lies in its ability to

detect WNAs dominated by natural succession, regardless of historic origin, greenspace type, and governance

structure, thereby providing a baseline against which the full potential for urban rewilding can be evaluated. We

demonstrate the methods’ utility in identifying potentials for urban rewilding, emphasising the importance of

directing rewilding efforts towards residential greenspaces, urban parks, and street greenery, especially in

densely built inner-city districts.

1. Introduction

The conservation and restoration of urban wild nature is increasingly

recognised as an important cornerstone in the transition towards more

sustainable, resilient, and equitable cities (Pineda-Pinto et al., 2023;

Randrup et al., 2020). Characterised by the dominance of ecological

processes (Kowarik, 2018), urban ‘wild nature areas’ (WNAs) can offer

multiple benets for urban residents and non-human species. They can

play a crucial role in climate change adaptation and mitigation efforts by

cooling urban environments, supporting stormwater management, and

improving carbon sequestration rates (Luo and Patuano, 2023; Sikorski

et al., 2021); they can contribute to urban biodiversity conservation by

providing habitats for plant and animal species and fostering ecological

connectivity within urban landscapes (Bonthoux and Chollet, 2024;

Müller et al., 2018); and promote urban residents’ mental and physical

well-being (Allard-Poesi et al., 2022; Maller et al., 2019) while also (re)

connecting people with nature near their homes (Cheesbrough et al.,

2019; Noe and Stolte, 2023).

Recognising these benets, a growing number of initiatives have

recently been launched worldwide to conserve and enhance wild nature

within urban environments. For instance, city partnerships such as

Biophilic Cities,

C40,

and the National Park City Foundation

have

been established to raise awareness about the value of different forms of

nature, including WNAs in an urban context. In addition, there has been

* Corresponding author.

E-mail address: brenda.zoderer@boku.ac.at (B.M. Zoderer).

https://www.biophiliccities.org/our-vision

https://www.c40knowledgehub.org/s/article/Urban-rewilding-the-value-and-co-benets-of-nature-in-urban-spaces?language=en_US

https://www.nationalparkcity.org/

Contents lists available at ScienceDirect

Urban Forestry & Urban Greening

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

https://doi.org/10.1016/j.ufug.2024.128549

Received 8 May 2024; Received in revised form 19 July 2024; Accepted 15 October 2024

Urban Forestry & Urban Greening 101 (2024) 128549

Available online 18 October 2024

(

http://creativecommons.org/licenses/by/4.0/ ).

a noticeable rise in ‘urban rewilding’ initiatives in practice, largely

spurred by the ‘Rewilding London’ taskforce established in 2021

(Greater London Authority, 2023). Those initiatives endeavour to

translate rewilding principles originally developed for rural landscapes

(Carver et al., 2021; Perino et al., 2019) to urban settings and identify

diverse opportunities for rewilding interventions throughout the urban

fabric (C40 & Arup, 2023; Greater London Authority, 2023). Similar to

passive rewilding approaches in rural areas (Massenberg et al., 2023),

urban rewilding aims to transition urban ecosystems to a state of greater

ecological complexity and resilience by restoring ecological processes

and reducing direct human control in the form of maintenance or

planting interventions (adapted from Bonthoux and Chollet, 2024 and

Pettorelli et al., 2022). Thus, urban rewilding initatives typically target

two distinct ecological processes: assembly and functioning spontaneity

(Bonthoux and Chollet, 2024). While assembly spontaneity can be

increased by reducing human intervention in species composition (e.g.,

by reducing planting or sowing), functioning spontaneity can be pro-

moted by reducing vegetation maintenance efforts (e.g., mowing or

pruning).

Despite the rising interest in wild nature and rewilding in cities

among policy-makers and practitioners, limited information is available

concerning the current prevalence and distributional characteristics of

WNAs at the city-scale. Previous research on urban wild nature has

predominantly focused on either the ecological function and biodiver-

sity of WNAs (e.g., Bonthoux et al., 2014; Sikorski et al., 2021) or on

people’s attitudes towards these spaces (e.g., Mathey et al., 2018;

Rupprecht et al., 2015) and their user experiences (e.g., Cheesbrough

et al., 2019; Watson et al., 2023). Although these studies provide

detailed accounts of the social-ecological value of specic local mani-

festations of wild spaces, little remains known about the

social-ecological potential of these spaces across the entire city. How-

ever, mapping the extent and distribution of WNAs at the city-scale can

facilitate the identication of opportunities to utilise these spaces for

recreational purposes, the conservation of biodiversity, or the promo-

tion of ecological connectivity across the city. Furthermore, providing

information on the prevalence and distribution of WNAs to urban

planners and greenspace managers can support them in identifying

suitable locations for urban rewilding and understanding how such en-

deavours could complement the conservation of existing WNAs.

In this paper, we adopt a process-oriented framing of WNAs to

explore their spatial distribution and to identify potentials for urban

rewilding using Vienna, the capital of Austria, as a case study. Following

this framing, we dene WNAs as all greenspaces in a city already

characterised by high levels of self-regulating capacities in ecosystem

processes and a corresponding lack of human maintenance (cf. Kowarik,

2018; Threlfall and Kendal, 2018). In contrast to alternative framings of

urban wild nature that focus on specic types of ecosystems or land-use

categories (see Section 2), a process-oriented framing emphasises the

presence of autonomous ecological processes (i.e., high levels of wild-

ness), irrespective of the origins, levels of formality, or types of green-

spaces in which they occur (Kowarik, 2018). This framing is thus closely

aligned with the idea of ‘urban rewilding’, which considers opportu-

nities for establishing WNAs by restoring and augmenting wildness

qualities in diverse settings such as parks, residential gardens, ceme-

teries, or railway tracks (Hwang, 2020; Pettorelli et al., 2022).

The diversity and spatio-temporal variability of WNAs (especially

when adopting a process-oriented framing) make them particularly

difcult to quantify and map. WNAs often remain blind spots on urban

greenspace maps because they do not t into standard land-use cate-

gories or are not captured in annual greenspace inventories

(Feltynowski et al., 2018; Preston et al., 2023). Furthermore, the pre-

dominant emphasis on mapping the quantity and distribution of urban

greenspaces using ready-to-use land-cover data often neglects internal

variations in management practices (Wellmann et al., 2018), likely

overlooking spatially heterogeneous patterns of greenspace qualities

such as wildness (Aznarez et al., 2022). Hence, for the mapping of

existing WNAs alternative methods are needed that not only capture the

quantity but also the quality of urban greenspaces (such as variations in

management intensity) at high spatiotemporal resolutions and across

xed land-use categories.

Here, we propose a novel methodological approach to map and

quantify WNAs adopting a process-oriented framing, and demonstrate

its applicability for identifying urban rewilding potentials in different

urban greenspace types and urban environments. We build on the gen-

eral idea proposed by Sikorska et al. (2021) of using a remotely-sensed

Normalised Difference Vegetation Index (NDVI) to identify

spatio-temporal patterns of unmanaged vegetation, and advance this

approach to identify WNAs as all those greenspaces with low levels of

vegetation management intensity and a corresponding high degree of

natural succession processes (i.e., high levels of functioning spontaneity,

Bonthoux and Chollet, 2024). We demonstrate the applicability and

robustness of the approach using the city of Vienna, and provide insights

into the distribution of WNAs across diverse urban greenspace types,

informal and formal greenspaces, and urban environments of varying

built-up density. We anticipate that the prevalence of WNAs varies

across the city, with wild spaces being less prevalent in greenspaces of

densely built urban environments as they tend to face higher use pres-

sures. We conclude by discussing possible priority areas for urban

rewilding initiatives and identifying related research needs that could

help advancing the emerging eld of urban rewilding research and

practice.

2. Urban wild nature: different framings and methodological

approaches

To date, only few studies have mapped WNAs and their central

qualities within urban environments. Proposed methodological ap-

proaches include various geospatial analytics (including GIS software)

and eld surveying techniques, reecting differences in how wild nature

and its key attributes are dened. The different assumptions and criteria

used in mapping studies lead to varying results with limited compara-

bility. To make sure that the results correspond to stakeholder needs in

greenspace management and urban planning, it is therefore paramount

to clarify the purpose for which maps are developed and underlying

assumptions. Here, we group the different denitions, emphasised at-

tributes of WNAs, and associated methodological approaches into three

framings of urban wild nature, namely the ‘novel ecosystem’, ‘land-use’,

and ‘process-oriented’ framing (Table 1), and briey describe their main

differences.

The novel ecosystem framing, which is rooted in urban ecology, as-

sociates WNAs with urban novel ecosystems dominated by spontaneous

vegetation (Pineda-Pinto et al., 2023; Riley et al., 2018). Novel eco-

systems often arise on urban land that was previously built-up or

otherwise heavily modied by human activity, resulting in signicant

changes in abiotic conditions (Kowarik, 2011). Within these settings and

in the absence of further human management, unique combinations of

native and non-native species can establish and evolve through natural

succession processes (Hobbs et al., 2013). Operationalising this con-

ceptual understanding, Planchuelo et al. (2019) created an ’ecosystem

novelty map’ of Berlin that distinguishes novel, hybrid, and remnant

ecosystems by spatially combining data on biotope types with current

land-use data.

The land-use framing complements the aforementioned framing by

highlighting key features in the governance and use of WNAs alongside

ecological elements. According to this view, WNAs are primarily iden-

tied as informal greenspaces, characterised by ecological novelty and

spontaneous vegetation, as well as informality in use and governance

processes (Luo and Patuano, 2023; Rupprecht and Byrne, 2014a; Stan-

ford et al., 2022). Scholars in this tradition typically associate WNAs

with distinct land-use units that vary in size, form, land-use history, and

other social-ecological characteristics (Rupprecht and Byrne, 2014a;

Stanford et al., 2022), such as brownelds, vacant lots, or railway

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

verges. Previous attempts to map WNAs using this framing have either

localised urban brownelds using register data on land-use (Preston

et al., 2023), mapped a broad spectrum of informal greenspaces varying

in shape and vegetation structure using systematic eld surveying (Kim

et al., 2020; Rupprecht and Byrne, 2014b), or mapped potential informal

greenspaces considering variation in social-ecological characteristics

using spatial overlay analysis (Stanford et al., 2024).

The process-oriented framing adopted in this study differs from the

previous framings in placing the dominance of ecological processes and

non-human agency at the centre of descriptions of wild nature, without

further considering ecological novelty or informality as additional pre-

conditions (Kowarik, 2018; Threlfall and Kendal, 2018). This framing is

the most comprehensive of the three, encompassing both novel ecosys-

tems and informal greenspaces highlighted in the previous two framings

as well as wild spaces located in formal greenspaces or remnants of

pristine ecosystems where ecological processes are dominant. To fully

capture this wide range of WNAs, it becomes crucial to closely examine

the prevalence of ecological processes as a shared landscape quality

across different contexts. To date, only few studies have attempted to

operationalise this conceptual understanding. Aznarez et al. (2022) and

Müller et al. (2018) used fuzzy GIS-based wildness mapping to deter-

mine the degree of wildness for each pixel cell of the cities

Vitoria-Gasteiz (Spain) and Aarhus (Denmark), considering variations in

naturalness, remoteness, ruggedness, and in one case also human impact

(similar to wildness mapping studies in rural landscapes, e.g., Carver

et al., 2012; Zoderer et al., 2020). While this approach is useful for easily

identifying WNAs based on publicly available data, it tends to over-

simplify variations in the landscapes’ wildness values by assigning ho-

mogenous naturalness values to entire land cover units. Addressing this

limitation, Jin et al. (2024) proposed an alternative method for identi-

fying ’urban rewilding opportunity spaces’ in Chongqing, China. In

addition to human disturbances and ruggedness, they considered also

habitat quality as an indicator for biodiversity and the landscape’s

suitability to accommodate ecological processes, such as trophic com-

plexities, that ultimately support the transition of an ecosystem to a

wilder state.

Focusing on urban spontaneous vegetation, Sikorska et al. (2021)

offers an even more dynamic perspective by identifying natural suc-

cession processes through an analysis of greenspace management over a

period of three years in Warsaw, Poland. While neglecting other po-

tential indicators, the approach provides a more robust analysis of

WNAs’ naturalness, accounting for both irregularities in human main-

tenance over time such as mowing or pruning and for variations within

land cover units. In this study, we advance their approach of using NDVI

time series to capture the spatial extent and temporal persistence of

unmanaged vegetation shaped by natural succession processes. While

Sikorska et al. (2021) could show that unmanaged vegetation is char-

acterised by higher NDVI values during the vegetation season than

cultivated vegetation, we were unable to replicate these results for

Table 1

Overview of different framings of urban WNAs in literature and their associated

denitions, attributes, and mapping approaches.

Framings of

urban

WNAs

Terms used

to refer to

urban WNAs

Denition of

urban WNAs

Key

attributes of

urban WNAs

Approaches

to map

urban WNAs

’Novel

ecosystem

framing’:

WNAs as

novel

ecosystems

Wild novel

ecosystems (

Pineda-Pinto

et al., 2023);

urban

spontaneous

vegetation (

Del Tredici,

2010; Riley

et al., 2018)

“a system of

abiotic, biotic,

and social

components

(and their

interactions)

that, by virtue

of human

inuence, differ

from those that

prevailed

historically,

having a

tendency to self-

organize and

manifest novel

qualities

without

intensive

human

management.”

(Hobbs et al.,

2013, p. 58)

- ecological

novelty

- presence of

spontaneous

vegetation

- lack of

ongoing

human

management

Ecosystem

novelty

mapping (

Planchuelo

et al., 2019)

’Land-use

framing’:

WNAs as

distinct

land-uses

Informal

greenspaces (

Luo and

Patuano,

2023;

Rupprecht

and Byrne,

2014a;

Stanford

et al., 2022)

“Any urban

space with a

history of strong

anthropogenic

disturbance

that is covered

at least partly

with non-

remnant,

spontaneous

vegetation. It is

neither formally

recognized by

governing

institutions or

property owners

as greenspace

[…]. Nor is the

vegetation

contained

therein

managed for

any of these by

the ofcial

owner. Any use

for recreational

purposes is

informal and

transitional

[…]”

(Rupprecht

and Byrne,

2014a)

- ecological

novelty

- presence of

spontaneous

vegetation

- lack of

ongoing

human

management

- informality

greenspace

governance

- informality

recreational

use

Systematic

eld

surveying of

informal

greenspaces (

Kim et al.,

2020;

Rupprecht &

Byrne,

2014b);

mapping

informal

greenspaces

using spatial

overlay

analysis (

Stanford

et al., 2024);

urban

browneld

mapping (

Biernacka

et al., 2023;

Preston et al.,

2023)

’Process-

oriented

framing’:

WNAs as

dominated

ecological

processes

Urban

wilderness

areas (

Kowarik,

2018), wild

spaces (

Threlfall &

Kendal,

2018), urban

rewilding

opportunity

spaces (Jin

et al., 2024)

"Urban

wilderness

areas can thus

be dened,

from an

ecological

perspective, as

places

characterized

by a high level

of self-

regulation in

ecosystem

processes,

including

- dominance

of ecological

processes (e.

g., natural

succession)

- lack of

ongoing

human

management

- varying

degrees of

ecological

novelty

- varying

degrees of

Relative

wildness

mapping (

Aznarez

et al., 2022;

Müller et al.,

2018); urban

rewilding

opportunity

index (Jin

et al., 2024);

urban

spontaneous

vegetation

algorithm (

Table 1 (continued )

Framings of

urban

WNAs

Terms used

to refer to

urban WNAs

Denition of

urban WNAs

Key

attributes of

urban WNAs

Approaches

to map

urban WNAs

population

dynamics of

native and

nonnative

species with

open-ended

community

assembly,

where direct

human impacts

are negligible." (

Kowarik,

2018; p. 339)

formality of

greenspaces

Sikorska

et al., 2021)

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

grasslands in Vienna (SI, Fig. C.1). In response, we propose using a

machine learning algorithm (based on random forest) that considers

idiosyncratic changes in time series of NDVI values of both types of

vegetation over the course of one year. This can provide a more robust

and ne-scaled distinction of unmanaged from managed vegetation per

year that goes beyond the use of a simplied binary approach.

Furthermore, we extend the aforementioned approach by employing

these annual maps of urban vegetation managment to map the level of

management intensity of vegetation over a period of six years. The latter

is used as a proxy for the degree of functioning spontaneity (cf. Bonthoux

and Chollet, 2024), upon which we spatially delineate WNAs at the

city-scale.

3. Materials and methods

3.1. Study site

Encompassing an area of 415 km

, the city of Vienna (Fig. 1) ac-

commodates close to 2 million residents (Statistik Austria, 2024).

Approximately half of Vienna’s territory is comprised of urban green-

spaces, with forests (38.4 %) and private gardens (20.3 %) constituting

the predominant greenspace types, followed by meadows (9.2 %), urban

parks (4.8 %), and greenery on agricultural elds (4.5 %). Of these

greenspaces, 61 % are formally designated as greenspaces with recrea-

tional functions. The largest share of urban greenspaces lies at the out-

skirts of the city, including the Vienna Woods in the north-west and the

Donau-Auen National Park in the south-east of the city. More centrally

located public greenspaces include large urban parks such as the Prater,

Sch¨

onbrunn Palace Park, Augarten or the recreational area Danube

Island.

3.2. Overview of methodological approach

The methodological approach consists of four main steps (Fig. 2).

First, we mapped and quantied the management intensity of urban

vegetation in Vienna using the remotely-sensed NDVI index as proxy.

We used Copernicus Sentinel-2 (S2) Level-2A (L2A) data (based on

sen2cor) and the Random Forest (RF) algorithm (Breiman, 2001) to

obtain a binary classication that distinguishes between unmanaged

and managed vegetation on a yearly basis. To this end, training data

from eld observations and additional random observations extracted

from publicly available geographic layers were used. A composite map

of ’urban vegetation management intensity’ was calculated from the

yearly maps to assess the management intensity of vegetation between

2017 and 2022. Using this map, we identied ’wild nature areas’

(WNAs) as greenspaces with low levels of vegetation management in-

tensity (i.e., high degree of functioning spontaneity, Bonthoux and

Chollet, 2024). We then investigated their spatial distribution across

urban areas with varying levels of built-up density, formal and informal

greenspaces, and different types of urban greenspaces. Finally, we

assessed the ‘theoretical potential’ for rewilding in existing urban

greenspaces, considering current distributions of WNAs across green-

spaces and their respective extent. The following sections describe each

step and the data sources used.

3.3. Mapping the management intensity of urban vegetation

To assess the management intensity of vegetation across Vienna, four

methodological steps were undertaken: 1) Collection of reference data to

train a Random Forest (RF) algorithm, 2) calculation of NDVI values

using remotely-sensed Sentinel-2 time series data, 3) application of RF

modelling to differentiate between unmanaged and managed vegetation

on a yearly basis based on the NDVI values and reference data, and 4)

creating a composite multi-year ’urban vegetation management in-

tensity’ map.

3.3.1. Collecting reference data

The reference data consists of both eld observations and reference

Fig. 1. Map of the study site and its coverage by different urban greenspace types (UGS types, own calculation). Data sources: Federal Ofce of Metrology and

Surveying, City of Vienna - https://data.wien.gv.at. Note: Arable land in the southern and eastward part of Vienna was excluded from our analysis and is therefore

not shown.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

points obtained from publicly available spatial data. The eld observa-

tions were conducted during the vegetation season of 2022 (May-

September) in different urban greenspaces either dominated by

managed or unmanaged vegetation across Vienna. The selected sites

were chosen based on expert interviews, a stakeholder workshop held in

2021 with members of the city councils responsible for nature conser-

vation, forestry and agriculture, and greenspace management, and the

consultation of local guidebooks on urban wilderness (Desbalmes et al.,

2020). Care was taken to select sites where a high persistence of

unmanaged vegetation could be assumed for the entire six-year period.

To this end, we compared different editions of local guidebooks and

examined historical maps from Google Maps.

In all eld sites, we randomly selected 10x10 m plots and recorded

whether the vegetation was managed or unmanaged. Plots with un-

managed vegetation were identied based on the botanical and struc-

tural composition of plants, the presence of tree seedlings, dead wood,

and the absence of signs of mowing, pruning, or mechanical damage

(Sikorska et al., 2021). In contrast, plots with managed vegetation were

Fig. 2. Overview of methodological approach.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

identied only if the same signs of maintenance were clearly visible (see

SI, Fig. A.1 for examples of both managed and unmanaged vegetation

plots). To account for differences in vegetation productivity due to

variations in vegetation structure and environmental conditions, we

collected a sufcient number of reference points in greenspaces domi-

nated by either grassland or tree cover (SI, Table A.1). Additionally,

reference points were selected to represent the three main spatial units

of the city: ’Cisdanubien’ (i.e., hilly areas and former terraces of the

Danube), ’Danube area’ (i.e., alluvial plain/valley oors), and ’Trans-

danubien’, (i.e., older valley oors). These spatial units differ in terms of

their environmental, geomorphological, and urban-historical conditions

(Adler & Mrkvicka, 2003; https://www.wien.gv.at/umweltgut/pu

blic/). A total of 760 eld observations were collected, with 367 refer-

ence points collected for unmanaged and 393 for managed vegetation

(SI, Table A.1).

Additional reference points were sampled for peri-urban forests

using publicly available spatial datasets. Unmanaged vegetation in peri-

urban forests was identied by randomly placing reference points in the

core zones of the city’s national park and biosphere reserve, where a

non-intervention management approach is applied. In contrast, refer-

ence points representing plots dominated by managed vegetation were

placed in the development zone of the two protected areas. In total,

2825 reference points were considered for peri-urban forests, of which

1031 were unmanaged and 1794 were managed (SI, Table A.1).

3.3.2. Processing of Sentinel-2 data

Sentinel-2 (S2) Level 2 A (L2A) data (atmospherically corrected)

with a spatial resolution of 10x10 m and cloud cover below 50 % were

obtained from the BOKU processing portal (Vuolo et al. 2016) for the

period 2017–2022. Based on this data, a time series of NDVI values was

calculated at the smallest possible pixel size of 10 m. Missing data and

potential clouds were identied with the S2 Scene Classication Layer

(considering class values of 0, 3, 8, 9 and 10) and masked. To obtain

monthly composites of NDVI with no data gaps, the masked time series

of NDVI was smoothed and gap-lled using the Whittaker smoother as

implemented in the R package ‘MODIS’ version 1.2.9, and using a

lambda value of 5000 and 1 single iteration. The 12-monthly NDVI

gap-lled image data were further reduced to six components using a

principal components analysis (PCA) for each year (Fig. 3). The PCA was

applied using the R package ‘stats’ version 4.0.3. Both the gap-lled

12-monthly NDVI data and the six PCA components were tested for RF

modelling, with the PCA components providing better results (results

not shown) and therefore they were retained for further modelling.

3.3.3. Random Forest modelling

The six PCA components were used as input to the RF modelling to

classify vegetation pixels as either managed or unmanaged for each year

(i.e., yearly maps of ‘urban vegetation management’). Data on urban

vegetation was derived from the urban greenspace monitoring (City of

Vienna, MA 22, 2018), which represents high resolution land cover data

from aerial imagery (resampled to 10 m pixel size using majority to

make it compatible with the Sentinel-2 satellite data). For our analysis,

only land cover classes ‘trees’, ‘shrubs’, and ‘grassland and herbaceous

vegetation’ were retained and used in the RF. For each year, RF models

were run using the reference data collected as training data, overall

producing six classication maps (one for each year). The RF algorithm

Fig. 3. a) and c) RGB composites of the three rst PCA components, and b) and d) RGB composites of the 4th, 5th and 6th PCA components. Panel a) and c) highlight

the most clear differences in land cover types (i.e., forests, built-up area), while Panel b) and d) show more subtle differences in land cover types, with a good spatial

consistency with actual patterns.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

was applied using the R package ‘randomForest’, version 4.6–14 and

with 1000 tree. The number of variables considered at each split was left

to the default value of 2. A nal composite map of ‘urban vegetation

management intensity’ was produced with a spatial resolution of 10x10

m by combining the six annual maps to quantify the temporal persis-

tence of management patterns over the period 2017–2022. The resulting

map displays the management intensity of vegetation along a seven-item

scale, with values ranging from 1 (always unmanaged) to 7 (always

managed). An additional focal lter was applied considering the modal

value within a moving window of 3×3 pixels.

Performance metrics for classication were summarised from the

confusion matrix that was calculated based on the RF out of bag (OOB)

samples. These are predictions of the RF trees not included in their

respective bootstrap samples (Liaw and Wiener, 2002). Several studies

have shown that the OOB error is comparable to an accuracy assessment

performed by splitting the reference data into training and validation

sets (Vuolo et al., 2018). In this study, using all reference data for

training was a preferred choice due to the relatively small eld reference

dataset and the need to obtain the best possible classication maps.

Performance metrics included Overall Accuracy, Precision, Recall, and

F1-score. Precision (or positive predictive value) measures how many of

the pixels classied as ‘unmanaged’ are ‘unmanaged’, indicating the

accuracy of the retrieved results. Recall (sensitivity) measures how

many of the ‘unmanaged’ pixels were classied as ‘unmanaged’ and thus

indicates the completeness of the classication process. The F1-score is

the harmonic mean of precision and recall, and provides a balanced

single measure of performance (Powers, 2008).

3.4. Identifying wild nature areas

Based on a process-oriented framing of wild nature (see Section 2),

we identied WNAs as all those greenspaces exhibiting low levels of

vegetation management intensity. For this purpose, we selected all

pixels of the composite ‘urban vegetation management intensity’ map

where unmanaged vegetation was recorded for 4 or more years during

the six-year period (i.e., management intensity value of 3 or lower).

After converting them into polygons, we considered all polygons with

more than half of their area within one urban greenspace type to belong

to the same WNA and assigned them a unique ID. This ensured that

discontiguous patches of unmanaged vegetation, although in close

proximity, were considered to be part of the same WNA. Furthermore,

only those areas with at least 100 m

of persistent unmanaged vegeta-

tion (equivalent to one raster pixel and a management intensity value of

1) over the entire six-year period were retained. The cut-off values were

derived through an iterative process of comparing the management in-

tensity map with the reference data obtained from the eld and ortho-

photos representing the surveyed areas.

3.5. Assessing the distribution of wild nature areas across built-up density

areas

The spatial distribution of WNAs was assessed across urban areas of

varying built-up densities. Built-up density was estimated based on the

‘soil sealing footprint’ of each sub-district in Vienna, considering the

percentage of built-up land and other impervious surfaces covering each

sub-district (Tombolini et al., 2016). Data on impervious surfaces were

derived from the high-resolution land cover data generated from aerial

imagery with a resolution of 0.5 m as part of the city’s greenspace

monitoring (City of Vienna, MA 22, 2018). Based on the built-up density

values per sub-district, the total area of Vienna was divided into three

built-up density classes using the ‘Natural Breaks (Jenks)’ algorithm in

ArcGIS Map 10.8.2: low (soil sealing: 0–31.7 %), medium (soil sealing:

31.8–60.4 %), and high built-up density (soil sealing: 60.5 %-100 %).

3.6. Assessing the distribution of wild nature areas across urban

greenspaces types

The distribution of WNAs was assessed for urban greenspace types

and greenspaces differing with regard to their formal recognition in the

city’s zoning plan. With regard to the distribution of WNAs across urban

greenspace types, we rst overlaid spatial data on WNAs with data about

the occurrence of urban greenspace types. The latter was derived by

intersecting data on the distribution of urban vegetation with the land-

use categories distinguished in the land-use map provided by the (City of

Vienna, MA 18, 2022) (SI, Table A.2). This resulted in a total number of

16 distinct urban greenspace types (Fig. 1). In addition, the distribution

of WNAs across formal and informal greenspaces was analysed. To this

end, we overlaid the created urban greenspace types map with the city’s

zoning plan (City of Vienna, MA 21 A, 2023), and identied all urban

greenspaces formally recognised as greenspace with recreational func-

tion in the zoning plan as ‘formal greenspace’ and all other as ‘informal

greenspace (SI, Table A.3).

3.7. Assessing the potential for urban rewilding

To estimate the ‘theoretical’ potential for converting existing urban

greenspaces into WNAs, we compared the total area of WNAs found in

urban greenspace types with the total area covered by these green-

spaces. In addition, the theoretical potential was also assessed at an

aggregated level by comparing the area of WNAs contained within all

greenspaces, formal greenspaces, and informal greenspace with their

respective areas covered. The analysis was carried out on a city-wide

scale and for the three built-up density classes separately. This allows

a rst assessment of the relative contribution of existing urban green-

space (types) in providing space for implementing urban rewilding ini-

tiatives across diverse urban environments.

4. Results

4.1. Distribution of wild nature areas across the city and built-up density

classes

The annual ‘urban vegetation management’ maps showed an average

accuracy of 84.5 % over the period 2017–2022, with class-specic

average accuracies of 81.8 % for unmanaged and 86.3 % for managed

vegetation (see SI, D for more information on the performance of RF

modelling). The multi-year composite (2017–2022) ‘urban vegetation

management intensity’ map reveals distinct spatio-temporal patterns of

management intensity (Fig. 4). Over the six-year period studied, 46.9 %

of urban vegetation was continuously managed (Table 2). In contrast,

11.3 % of the city’s greenery is classied as unmanaged for six consec-

utive years, followed by a further 10.6 % that was managed once and

7.9 % that was managed twice within the six-year period studied. Based

on the composite ‘urban vegetation management intensity’ map, we

identied 1298 WNAs covering a total area of 6165 ha with an average

area per site of 4.7 ha. Overall, these areas constitute 29.5 % of the city’s

total urban green and 14.9 % of the city’s total area.

WNAs are unevenly distributed across the city (Fig. 5a), with

considerable differences between areas of different built-up densities

(Fig. 5b and c). The majority of WNAs (65.3 %) are located in low-

density areas, while comparatively fewer are found in areas of me-

dium (28.4 %) and high built-up density (6.3 %). This uneven distri-

bution is even more pronounced when considering the total area of

WNAs. While 95.9 % of their total area is found in districts of low built-

up density, only 3.8 % is located in areas of medium and 0.3 % in areas

of high built-up density. Relatedly, WNAs are on average smaller in

high- (0.2 ha) and medium-density built-up areas (0.6 ha) than in urban

areas of low built-up density (6.6 ha).

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

4.2. Distribution of wild nature areas across urban greenspace types

WNAs are unevenly distributed across urban greenspaces types.

Forests contain most WNAs (34.8 %), followed by meadows (16.5 %),

private gardens (14.1 %), and parks (10.9 %) (Fig. 6a). In contrast,

WNAs that exist as street verges (3.5 %), railway verges (1.2 %), areas

around technical infrastructure (1.4 %), on industrial and commercial

sites (0.6 %), or in cemeteries (0.7 %) have a lower prevalence. Looking

at the total area of WNAs, this imbalance is again even more pronounced

(Fig. 6b). WNAs in forests cover 93.4 % of the total area of WNAs

Fig. 4. Management intensity of urban vegetation in Vienna for the period 2017–2022.

Table 2

The management intensity of urban vegetation by urban greenspace (UGS) type (in %) and differentiated between formal and informal greenspaces (in %). Note:

Formal greenspaces refer to all urban greenspaces formally recognised as greenspace with recreational function in the city’s zoning plan (City of Vienna, MA 21 A,

2023). Informal greenspaces refer to all other greenspaces.

UGS type Always

unmanaged

One year

managed

Two years

managed

Three years

managed

Four years

managed

Five years

managed

Always

managed

Share of city’s

urban greenery

Forests 27.2 24.6 16.8 10.8 7.7 5.9 7.0 38.4

Greenery on agricultural elds

(orchards. vineyards)

0.5 0.7 1.0 1.7 4.9 19.6 71.7 4.4

Greenery on cemeteries 0.9 1.2 1.6 2.6 4.9 12.3 76.6 2.0

Greenspace along railway

tracks

0.9 1.3 1.9 3.4 7.8 17.4 67.2 1.3

Greenspace on industrial and

commercial sites

0.1 0.3 0.6 1.2 3.5 12.8 81.6 1.1

Greenspace near service areas 0.2 0.2 0.6 1.1 2.5 10.5 84.8 0.3

Greenspace near social

infrastructure

0.4 0.8 1.1 2.0 4.1 12.8 78.9 1.7

Greenspace near technical

infrastructure

1.0 1.8 2.7 3.7 7.3 18.1 65.4 1.1

Greenspace on construction

and mineral extraction sites

0.4 0.9 1.8 2.3 4.9 14.9 74.9 0.4

Greenspace on sports and

leisure sites

1.2 1.6 2.0 2.9 5.1 12.0 75.2 2.5

Meadows 3.3 4.6 5.8 9.1 15.5 24.4 37.4 9.2

Parks 5.0 5.7 6.0 6.8 9.4 16.5 50.7 4.8

Private gardens 0.4 0.6 0.7 1.1 2.0 6.9 88.4 20.3

Shared neighbourhood

greenspace

0.2 0.4 0.7 1.5 3.3 11.7 82.3 5.2

Street greenery 0.6 0.8 1.1 1.8 3.8 11.7 80.2 5.9

Waterside greenspace 7.4 10.5 11.2 11.0 12.1 16.2 31.6 1.5

Formal greenspaces 18.2 17.0 12.2 9.0 8.4 10.0 25.2 61.0

Informal greenspaces 0.6 0.8 1.1 1.8 3.7 11.3 80.7 39.0

Overall 11.3 10.6 7.9 6.2 6.6 10.5 46.9 100.0

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

identied, whereas only a small share of the total area is found in parks

(2.7 %), meadows (1.7 %), private gardens (0.6 %), and other urban

greenspace types (1.6 %). However, the proportion of urban greenspace

types with WNAs varies between the three built-up density classes

(Fig. 6a-b). While in low-density built-up areas WNAs are mainly located

in forests (41.0 % of WNAs, 94.7 % of their total area) and meadows

(17.2 % of WNAs, 1.4 % of their total area), in high-density areas they

are mainly found in parks (33.7 % of WNAs, 34.6 % of their total area)

and shared neighbourhood greenspace (18.6 %, 5 % of their total area).

Although small in number, WNAs such as railway verges or wild vege-

tation near technical infrastructure area are more important in medium-

and high-density built-up areas, where a higher proportion of these

areas can be found.

Fig. 5. Distribution of WNAs across the city of Vienna (a) and areas of varying built-up density (BD) (a, b and c). The city is divided into three built-up density classes

based on the percentage of soil sealing of each subdistrict: low-density (0–31.7 %), medium-density (31.8–60.4 %), and high-density built-up areas (60.5–100 %).

The share of WNAs found across the three built-up density areas is shown in terms of the number of WNAs (b) and the total area covered by WNAs (c).

Fig. 6. Distribution of WNAs across urban greenspaces types shown for the total city and the three built-up density classes. The relative share of WNAs across urban

greenspaces types is displayed considering a) the total number of WNAs and b) the total area covered by WNAs.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

4.3. Distribution of wild nature areas across formal and informal

greenspaces

The majority of identied WNAs (62.3 %) are located in urban

greenspaces formally recognised as greenspaces with recreational

function in the city’s zoning plan (Fig. 7a). Considering that these WNAs

are on average larger (formal: 7.45 ha, informal: 0.29 ha), 96.7 % of the

total area of WNAs is located in formal urban greenspaces (Fig. 7b).

Although both formal and informal WNAs are predominantly located in

low-density built-up areas, the proportion of WNAs located in informal

greenspaces increases with increasing built-up density (Fig. 7a-b). While

informal WNAs account for 27.1 % of all WNAs in areas of low built-up

density, this proportion increases to 45.5 % in medium- and 45.3 % in

high-density built-up areas. This corresponds to a total area of 2.0 % of

informal WNAs in low-density built-up areas, compared to 18.1 % and

32.4 % in medium- and high-density areas respectively.

4.4. Potential for urban rewilding

We nd a high theoretical potential for urban rewilding across urban

greenspace types and urban environments of varying built-up densities

(Fig. 8). Except for forests, where WNAs currently cover 71.8 % of the

total forest area across the city, WNAs cover 0.3 % (street greenery) to

16.7 % (parks) of the total area per urban greenspace type. For most

urban greenspace types, the share of WNAs decreases the more densely

built the urban environment is (e.g., parks: 21.6 % in low-density,

10.5 % in medium-density, and 5.3 % in high-density areas). Simi-

larly, and on an aggregated level, WNAs cover 37.9 % of all urban

greenery in low-density built-up areas, whilst this share decreases to

5.4 % in medium- and to 2.0 % in high-density built-up areas. In all

three built-up density classes, WNAs cover a considerably higher share

of the existing formal (50.9 % in low-density to 7.2 % in high-density

areas) than informal greenspace area. With regard to the latter, WNAs

make up a smaller but relatively more stable share across the three built-

up density classes, ranging from 2.8 % in low- to 0.8 % in high-density

built-up areas.

5. Discussion

5.1. Diversity of urban wild nature areas

Taking natural succession processes rather than specic land-uses or

novel ecosystems as a starting point, we show that wild nature can occur

in a variety of urban environments and urban greenspace types. In line

with previous typologies of ‘urban wilderness’ (Kowarik, 2018) or ‘wild

spaces’ (Threlfall and Kendal, 2018), WNAs can be found in both

traditional greenspaces such as forests, meadows, urban parks, which

are largely included in the formal greenspace network and managed by

the city administration for recreational purposes, and in informal

greenspaces such as residential greenspaces, industrial sites, and road or

railway verges. Although the former have received less attention in

current debates on urban wild spaces, our results indicate a signicant

concentration of WNAs, particularly in forests, but also in meadows and

parks. This reinforces previous empirical mapping studies that have

adopt a process-oriented framing of urban wild nature (Aznarez et al.,

2022; Müller et al., 2018; Sikorska et al., 2021), which show that these

greenspaces can contain substantial areas with high wildness qualities.

Especially in forest areas, the presence of unmanaged vegetation was

persistent over the studied period, conrming previous studies that

assumed a high temporal continuity of natural succession processes in

these greenspaces (Sikorska et al., 2021). While the predominance of

WNAs in urban forests was expected, the relatively high proportion

found in urban parks was rather surprising, suggesting that urban

greenspace types commonly assumed to adhere to intensive manage-

ment patterns can include WNAs of different scales. These mainly

comprise larger wooded areas in historic parks (e.g., Sch¨

onbrunn),

smaller groups of unmanaged trees in otherwise traditionally main-

tained urban parks (e.g., Rathauspark), or spontaneous woodlands that

have developed on abandoned sites and have now been integrated into

formally recognised urban parks (e.g., Freie Mitte).

Our ndings provide rst evidence on the relative importance of

formal and informal WNAs across urban environments of differing built-

up densities. WNAs located in greenspaces formally recognised by the

city administration dominate in all three built-up density areas, both in

terms of area and number. This nding is notable as informal green-

spaces account for more than 70 % of all greenspaces in both medium

and high-density areas in Vienna (SI, Table B.1 and B.2). However, as

Fig. 7. Share of WNAs located in formal and informal urban greenspaces shown

for the total city and the three built-up density areas. Note: Some WNAs are

considered in multiple built-up areas if they cross the border of one or more

built-up density areas. *4 ha of the total area of WNAs identied across the city

(i.e., 6165 ha) could not be attributed to either formal or informal greenspaces

due to No Data values in the land-use map used for UGS mapping.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

indicated by our assessment of vegetation management intensity,

informal greenspaces in particular tend to be continuously managed or

only occasionally not managed, leaving little space for natural succes-

sional processes to drive and shape species community composition over

longer time periods.

The limited prevalence of informal WNAs, although increasing in

relative importance in more densely built inner-city districts, contradicts

prior studies that identied signicant proportions of wild informal

greenspaces such as street verges, railways verges, lots, waterside spaces

and wild vegetation around technical structures in densely populated

areas in cities like Brisbane, Sapporo, or Ichikawa City (Kim et al., 2020;

Rupprecht and Byrne, 2014b). In Vienna, the same particular types of

wild informal greenspaces account for less than 1 % of the overall urban

greenery compared to 14 % found by Rupprecht and Byrne (2014b) for

Brisbane and Sapporo. While this may point to particularly strongly

expressed densication processes in Vienna, differences may also stem

from methodological differences outlined in Section 5.4.

5.2. Spatial distribution and characteristics of urban wild nature areas

The analysis reveals a signicant concentration of WNAs in low-

density built-up areas. While it has been commonly observed that

Vienna’s inner-city of Vienna harbours less urban greenspace compared

to its outskirts (Brenner et al., 2021), our results underline that this

unequal distribution is even more pronounced in the case of wild nature.

Specically, while the coverage of urban greenspaces located in

low-density areas accounts for 61 % of total greenspace areas, the

equivalent share is 96 % in the case of WNAs. The reasons for such stark

differences may be twofold. Firstly, informal greenspaces that were

found to adhere to generally more intensive forms of vegetation man-

agement, such as shared neighbourhood greenspace or street greenery,

are more widespread in medium- and high-density areas than in

low-density areas. Secondly, formal greenspaces, such as forests,

meadows, and parks, where WNAs predominantly exist in low-density

areas, are found to be more intensively managed when located in

densely built-up areas. This may reect a greater necessity for green-

space managers to intensify management efforts to cope with increasing

use pressures arising from an increasing urban population living and

working nearby.

In addition, our ndings indicate that the spatial properties of WNAs

also vary across built-up areas. In low-density areas, WNAs manifest

themselves as large, clustered areas (e.g., Vienna Woods) or linear areas

(e.g., Danube Island). In contrast, WNAs in medium-density areas tend

to be smaller and more scattered, while in high-density areas WNAs are

limited to small patches scattered across a few disconnected locations.

However, both the spatial distribution and characteristics of WNAs hold

implications for the benets these greenspaces offer to humans and non-

human species. As demonstrated by Gao et al., 2021, both patch size and

spatial conguration inuence the number of spontaneous plant species

in an area, with larger, connected areas, and those with higher

perimeter-area ratios expected to harbour more species. Similarly, the

cooling effects of WNAs, crucial for climate adaptation, may be

Fig. 8. Theoretical potential for urban rewilding in existing urban greenspaces considering the proportion of greenspace area (in %) that is not covered by WNAs.

Potentials are shown for the total city and the three built-up density areas.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

inuenced by spatial properties. Previous research has shown a positive

correlation between the size of greenspaces such as forests and parks and

their cooling effects (Jaganmohan et al., 2016). However, these effects

are typically non-linear and limited to certain distances, such as

300–400 m in densely built environments (Andersson et al., 2020), thus

emphasising the importance of an equal distribution of greenspaces to

maximise their potential benets in urban settings.

Hence, complementing the conservation of existing WNAs with

urban rewilding efforts aimed at expanding and connecting existing

WNAs, such as the smaller and more scattered WNAs in medium and

high-density areas with larger ones on the urban fringe, will be imper-

ative. This entails restoring existing greenspaces and transitioning grey

to high-quality greenspaces, particularly in the more densely populated

inner-city districts. This can enhance ecological connectivity for the

benet of biodiversity and contribute to a more equitable distribution of

climate adaptation and other well-being benets among urban residents.

5.3. Opportunities for urban rewilding in existing urban greenspaces

Our research points to a large theoretical potential for urban

rewilding in grey and green areas of the city, both in terms of the total

area that could – in principle – be turned into rewilding sites, and in

terms of the discrepancy that can be found in some greenspace types

between their total size and the size of WNAs contained in them.

Notably, the share of land covered by WNAs is currently low for most

greenspace types, indicating a signicant potential for transforming

parts of these greenspaces into WNAs while retaining their primary

function across the city. Such efforts could also contribute to the real-

isation of the city-specic targets as proposed by the EU Nature Resto-

ration Law adopted in 2024 (European Commission, 2024).

In particular, private gardens and shared residential greenspaces

demonstrate a high rewilding potential due to the large discrepancy

between the area covered (SI, Table B.2) and the amount of WNAs

contained (Fig. 8). Currently, WNAs make up less than 1 % of the total

area of residential greenspaces, whilst the latter account for one third or

more of all greenspaces in medium and high-density built-up areas.

These results support earlier proposals to prioritise rewilding in resi-

dential gardens by reducing management interventions (Moxon et al.,

2023; Pettorelli et al., 2022). Such efforts could enhance the multi-

functionality of these greenspaces, support wildlife conservation

(Goddard et al., 2010), and strengthen people’s connection to nature

(Mumaw et al., 2017).

In addition, promoting rewilding in urban parks could provide a

valuable complementary option for expanding WNAs in the city

(Pettorelli et al., 2022). Currently, urban parks accommodate a higher

proportion of WNAs than residential greenspaces in Vienna. However, it

is evident that the amount of WNAs included in urban parks varies

considerably throughout the city. Urban parks in the suburbs tend to

contain more WNAs than those in the city centre, as the latter are

frequently subject to more intensive management patterns to cope with

increased use pressures. There is also considerable variation in the

amount of WNAs present in inner-city parks, indicating that a thorough

case-by-case basis evaluation is required to identify opportunities for

rewilding in urban parks while also considering current uses and the

diverse needs of park users (Lampinen et al., 2021).

A third, and so far, largely overlooked option for urban rewilding

concerns the increase of wild vegetation alongside street verges. In

particular, in high-density areas, where opportunities for more space-

consuming projects are limited, rewilding streets could be a viable op-

tion (Bonthoux et al., 2019). Currently, street greenery accounts for

18 % of all urban green in high-density areas of Vienna, of which only

0.3 % is considered to be wild. Where safety permits, this percentage

could be increased by applying more extensive management principles

(reduced weeding and mowing) and deliberately promoting sponta-

neous vegetation. Previous research has shown that such efforts can

increase ecological connectivity, wildower richness, and facilitate wild

nature contact also in areas where access to larger (wild) greenspaces is

limited (Bonthoux et al., 2019; Vega and Küffer, 2021).

5.4. Methodological considerations

The methodology developed in this study proved useful for under-

standing the spatial distribution of WNAs at the city-scale. Unlike

common land-use mapping approaches, the proposed methodology

identies WNAs by adopting a continuum approach and dening cut-off

values to separate wild from non-wild areas. The major advantage of the

continuum approach lies in the ability to capture variations in man-

agement intensity of vegetation within and across established land-use

categories. In translating the continuum into binary wild and non-wild

areas, we followed the conventions in the eld of wilderness mapping

(see e.g., Cao et al., 2019 for an overview). The identication of corre-

sponding WNAs facilitates the presentation of the spatial distribution of

WNAs and estimation of the urban rewilding potential across different

greenspace types, allowing urban planners and greenspace managers to

effectively pinpoint locations that require further conservation and

restoration efforts. Further advantages lie in its rapid application, the

use of publicly available satellite and spatial data, and the related

transferability to other cities with different biophysical conditions and

greenspace policies (provided that new reference data can be collected).

While the methodology is transferable to other cities, future research

needs to test the generalisability of the results obtained. For example, a

comparison of the distribution of WNAs in different cities could provide

further insights into the robustness of the relationship between the

occurrence of WNAs and built-up density, and into the factors shaping

the development and persistence of WNAs across varying spatial con-

ditions and greenspace policies.

The methodological approach developed in this study presents

several advancements in comparison to the approach proposed by

Sikorska et al. (2021). First, in contrast to their assumption that un-

managed vegetation consistently exhibits higher yearly averaged NDVI

values compared to cultivated vegetation during the vegetation season,

our ndings did not support this for eld observations collected for

grassland (SI, Fig. C.1). To address this limitation, we analysed data for

the entire year, deriving monthly time series of NDVI (SI, Fig. C.2) as

well as the most informative features obtained through a data reduction

method (using PCA). Second, the nal multi-year ‘urban vegetation

management intensity’ map was based on a longer time period, span-

ning six instead of three years. Third, we considered the management

intensity of forests, using reference data from different zones of pro-

tected areas.

Finally, in comparison to the approach taken by Sikorska et al.

(2021), which used binary thresholds to distinguish unmanaged from

managed vegetation, this study employed a classication approach

based on RF modelling to identify the two vegetation types on an annual

basis. This approach is more robust than a single threshold approach for

several reasons: 1) RF generally provides higher accuracy due to its

ensemble nature, combining the results of multiple decision trees (1000

in this study) to improve predictive performance; 2) the latter makes the

RF approach also more robust to noisy data and outliers by minimising

the inuence of a single poor-performing tree; as well as 3) to overtting

than threshold-based approaches, and 4) RF can also better capture

complex, non-linear relationships between features than threshold ap-

proaches relying on simpler rules.

We also encountered some methodological limitations that deserve

further attention. First, the annual vegetation management maps pro-

duced distinguished between unmanaged and managed vegetation on a

binary basis. Although we employed these maps to delineate the man-

agement intensity on a seven-point scale over a six-year period, future

research could rene the proposed approach by generating a manage-

ment intensity gradient for each individual year. Consequently, rather

than gathering eld data on the basis of whether vegetation is managed

or unmanaged, a wider range of management intensity categories could

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

be evaluated in the eld and incorporated into the RF modelling to map

even more subtle variations in wildness. Second, the Sentinel-2 satellite

data were only available at a spatial resolution of 10x10 m, which

limited the mapping of WNAs to an area threshold of 100 m

. Conse-

quently, smaller WNAs, such as the informal ’gaps’ or ’road verges’

frequently identied in Rupprecht and Byrne (2014b) using eld

surveying, could not be detected. This may have led to an underesti-

mation of the number and total area of WNAs especially in highly

densely built environments, where WNAs tend to be smaller. Third, the

long temporal persistence assumed in this study, namely that WNAs only

qualify as such if they contain at least 100 m

of vegetation not managed

for six years, may have resulted in the predominant detection of WNAs

in their later stages of succession and the exclusion of temporally more

liminal spaces. Despite the application of a rigorous process to identify

cut-off values that best align with the wild spaces surveyed in the eld, it

remains a sensitive step. Further research could experiment with the use

of different cut-off values (e.g., persistence of unmanaged vegetation for

three years) and apply higher resolution satellite data or alternative data

such as LiDAR to more accurately represent also short-lived and smaller

wild nature patches.

6. Conclusions

This study set out to map wild nature areas (WNAs) in Vienna using

Sentinel-2 satellite data and to identify potentials for urban rewilding.

Taking a process-oriented approach, we delineated WNAs as green-

spaces primarily shaped by natural succession processes and assessed

their distribution across urban fabric and greenspace categories. Unlike

the ‘novel ecosystem framing’ and ‘land-use framing’, which aim to map

novel ecosystems or distinct informal greenspaces, our approach iden-

tied already existing WNAs irrespective of historical origin, gover-

nance structure or greenspace type. This information can aid policy-

makers and practitioner to monitor efforts to conserve existing WNAs

and to identify the full potential to expand and connect existing WNAs

through urban rewilding across different urban settings.

In contributing to the growing debate on urban wild nature protec-

tion and restoration, our study offers the following recommendations for

future research and practice. First, our nding that WNAs are primarily

located in formal greenspaces suggests that a more balanced account of

both formal and informal WNAs is needed. While our research un-

derscores the importance of informal WNAs, especially within more

urbanised areas, the prevailing emphasis on informality in delineating

WNAs may inadvertently neglect those located in formal greenspaces.

Recognising the complementary role of these areas in biodiversity

conservation and ecosystem services provision, we encourage future

research to encompass the full spectrum of WNAs, particularly when

assessing the value of wild nature.

Second, our results underscore the importance of directing efforts to

protect and restore wild nature towards more centrally located, medium

to high built-up density areas, where WNAs constitute a small propor-

tion of urban green. Prioritising the protection of WNAs in these areas

can enhance habitat connectivity, climate change adaptation benets,

and nature experiences closer to residents’ home. Given the overall

limited availability of WNAs in densely built-up areas, however, these

efforts will need to be complemented by the rewilding of existing grey

and green areas.

Third, our ndings suggest that residential greenspaces, street

greenery, and urban parks hold signicant potential for conversion into

rewilding sites. While we demonstrate this potential from a quantitative

perspective, future research needs to explore the practical feasibility and

challenges of transforming (part of) these greenspaces into WNAs. This

involves exploring competing land-uses and planning restrictions

potentially hindering rewilding, assessing the acceptability of urban

rewilding initiatives by local communities, understanding the factors

inuencing the behavioural intentions of private and public actors to

implement rewilding, and examining the political, nancial and

administrative conditions required to realise rewilding efforts in cities.

CRediT authorship contribution statement

Brenda Maria Zoderer: Writing – original draft, Visualization,

Methodology, Funding acquisition, Formal analysis, Supervision,

Conceptualization. Christa Hainz-Renetzeder: Writing – review &

editing, Visualization, Formal analysis, Data curation. Francesco

Vuolo: Writing – review & editing, Validation, Methodology, Formal

analysis, Writing – original draft.

Declaration of Competing Interest

The authors declare that they have no known competing nancial

interests or personal relationships that could have appeared to inuence

the work reported in this paper.

Acknowledgements

This work was supported by the City of Vienna Anniversary Fund for

the University of Natural Resources and Life Sciences, Vienna (research

project SUCCESS: Natural Succession as a Solution for Sustainable, Resilient

and Inclusive Cities). We would like to thank the City of Vienna (MA22)

for providing us the greenspace monitoring data, and Franka Mathilde

Fuchs for her help in preparing some of the gures in this manuscript.

We also thank the anonymous reviewers for their valuable and

constructive feedback on an earlier version of this manuscript. Open

access funding was provided by the University of Natural Resources and

Life Sciences, Vienna.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the

online version at doi:10.1016/j.ufug.2024.128549.

References

Adler, W., & Mrkvicka, A. (2003). Die Flora Wiens gestern und heute. Die

wildwachsenden Farn- und Blütenpanzen in der Stadt Wien von der Mitte des 19.

Jahrhunderts bis zur Jahrtausendwende. Wien.

Allard-Poesi, F., Matos, L.B.S., Massu, J., 2022. Not all types of nature have an equal

effect on urban residents’ well-being: A structural equation model approach. Health

Place 74, 102759. https://doi.org/10.1016/j.healthplace.2022.102759.

Andersson, E., Haase, D., Scheuer, S., Wellmann, T., 2020. Neighbourhood character

affects the spatial extent and magnitude of the functional footprint of urban green

infrastructure. Landsc. Ecol. 35 (7), 1605–1618. https://doi.org/10.1007/s10980-

020-01039-z.

Aznarez, C., Svenning, J.-C., Taveira, G., Bar´

o, F., Pascual, U., 2022. Wildness and

habitat quality drive spatial patterns of urban biodiversity. Landsc. Urban Plan. 228,

104570. https://doi.org/10.1016/j.landurbplan.2022.104570.

Biernacka, M., Kronenberg, J., Łaszkiewicz, E., Czembrowski, P., Amini Parsa, V.,

Sikorska, D., 2023. Beyond urban parks: Mapping informal green spaces in an

urban–peri-urban gradient. Land Use Policy 131, 106746. https://doi.org/10.1016/

j.landusepol.2023.106746.

Bonthoux, S., Brun, M., Di Pietro, F., Greulich, S., Bouch´

e-Pillon, S., 2014. How can

wastelands promote biodiversity in cities? A review. Landsc. Urban Plan. 132,

79–88. https://doi.org/10.1016/j.landurbplan.2014.08.010.

Bonthoux, S., Chollet, S., 2024. Wilding cities for biodiversity and people: A

transdisciplinary framework. Biol. Rev. https://doi.org/10.1111/brv.13076.

Bonthoux, S., Voisin, L., Bouch´

e-Pillon, S., Chollet, S., 2019. More than weeds:

Spontaneous vegetation in streets as a neglected element of urban biodiversity.

Landsc. Urban Plan. 185, 163–172. https://doi.org/10.1016/j.

landurbplan.2019.02.009.

Breiman, L., 2001. Random Forests. Mach. Learn. 45 (1), 5–32. https://doi.org/10.1023/

A:1010933404324.

Brenner, A.-K., Mocca, E., Friesenecker, M., 2021. Vienna’s urban green space planning.

In Y. Kazepov & R. Verwiebe, Vienna (1. Au., S. Routledge, pp. 117–130. https://

doi.org/10.4324/9781003133827-12.

2023 C40 & Arup, 2023. Urban rewilding: The value and co-benets of nature in urban

spaces. Available from: https://www.c40knowledgehub.org/s/article/Urban-r

ewilding-the-value-and-co-benets-of-nature-in-urban-spaces?language=en_US

(last accessed 18th of October 2024).

Cao, Y., Carver, S., Yang, R., 2019. Mapping wilderness in China: Comparing and

integrating Boolean and WLC approaches. Landsc. Urban Plan. 192, 103636. https://

doi.org/10.1016/j.landurbplan.2019.103636.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

Carver, S., Comber, A., McMorran, R., Nutter, S., 2012. A GIS model for mapping spatial

patterns and distribution of wild land in Scotland. Landsc. Urban Plan. 104 (3–4),

395–409. https://doi.org/10.1016/j.landurbplan.2011.11.016.

Carver, S., Convery, I., Hawkins, S., Beyers, R., Eagle, A., Kun, Z., Van Maanen, E.,

Cao, Y., Fisher, M., Edwards, S.R., Nelson, C., Gann, G.D., Shurter, S., Aguilar, K.,

Andrade, A., Ripple, W.J., Davis, J., Sinclair, A., Bekoff, M., Soul´

e, M., 2021. Guiding

principles for rewilding. Conserv. Biol., cobi 13730. https://doi.org/10.1111/

cobi.13730.

Cheesbrough, A.E., Garvin, T., Nykiforuk, C.I.J., 2019. Everyday wild: Urban natural

areas, health, and well-being. Health Place 56, 43–52. https://doi.org/10.1016/j.

healthplace.2019.01.005.

City of Vienna, MA 18 (2022). General data on actual land use in Vienna. Available from

〈https://www.data.gv.at/katalog/de/dataset/stadt-wien_realnutzungskartierung

ab200708wien〉(last accessed 8th March, 2024).

City of Vienna, MA 21 A (2023). Generalised illustration of land use and development

plans. Available from 〈https://www.data.gv.at/katalog/de/dataset/stadt-wien_gene

ralisiertechenwidmungwien〉(last accessed 8th March, 2024).

City of Vienna, MA 22 (2018). Greenspace monitoring 2018. Available on request from

the City of Vienna.

Del Tredici, P., 2010. Spontaneous Urban Vegetation: Reections of Change in a

Globalized World. Nat. Cult. 5 (3), 299–315. https://doi.org/10.3167/

nc.2010.050305.

Desbalmes, G., Tichelmann, I., & Doppler, W. (2020). Am Anfang war die Gstett’n.

Wiener Stadtwildnis¨

achen. Available from 〈https://wuawien.at/images/stories/p

ublikationen/gstettnfuehrer-2020.pdf〉(last accessed 8th March, 2024).

European Commission (2024). Nature Restoration Law. Available from: 〈https://www.

consilium.europa.eu/en/meetings/env/2024/06/17/〉(last accessed 15th of July,

2024).

Feltynowski, M., Kronenberg, J., Bergier, T., Kabisch, N., Łaszkiewicz, E., Strohbach, M.

W., 2018. Challenges of urban green space management in the face of using

inadequate data. Urban For. Urban Green. 31, 56–66. https://doi.org/10.1016/j.

ufug.2017.12.003.

Gao, Z., Song, K., Pan, Y., Malkinson, D., Zhang, X., Jia, B., Xia, T., Guo, X., Liang, H.,

Huang, S., Da, L., Van Bodegom, P.M., Cieraad, E., 2021. Drivers of spontaneous

plant richness patterns in urban green space within a biodiversity hotspot. Urban

For. Urban Green. 61, 127098. https://doi.org/10.1016/j.ufug.2021.127098.

Goddard, M.A., Dougill, A.J., Benton, T.G., 2010. Scaling up from gardens: Biodiversity

conservation in urban environments. Trends Ecol. Evol. 25 (2), 90–98. https://doi.

org/10.1016/j.tree.2009.07.016.

Greater London Authority, 2023. Rewilding London. Final Report of the London

Rewilding Taskforce. Available from: https://www.london.gov.uk/progra

mmes-strategies/environment-and-climate-change/parks-green-spaces-and-biodiv

ersity/london-rewilding-taskforce (last accessed 18th of October 2024).

Hobbs, R.J., Higgs, E., & Hall, C. (2013). Novel Ecosystems: Intervening in the New

Ecological World Order. Wiley-Blackwell.

Hwang, Y.H., 2020. Rewilding urban landscapes, 2. Au., S. In: Douglas, I., Anderson, P.

M.L., Goode, D., Houck, M.C., Maddox, D., Nagendra, H., Yok, T.P. (Eds.), The

Routledge Handbook of Urban Ecology. Routledge, pp. 762–772. https://doi.org/

10.4324/9780429506758-65, 2. Au., S.

Jaganmohan, M., Knapp, S., Buchmann, C.M., Schwarz, N., 2016. The Bigger, the Better?

The Inuence of Urban Green Space Design on Cooling Effects for Residential Areas.

J. Environ. Qual. 45 (1), 134–145. https://doi.org/10.2134/jeq2015.01.0062.

Jin, X., Qian, S., Yuan, J., 2024. Identifying urban rewilding opportunity spaces in a

metropolis: Chongqing as an example. Ecol. Indic. 160, 111778. https://doi.org/

10.1016/j.ecolind.2024.111778.

Kim, M., Rupprecht, C.D.D., Furuya, K., 2020. Typology and Perception of Informal

Green Space in Urban Interstices: A case study of Ichikawa City, Japan. Int. Rev.

Spat. Plan. Sustain. Dev. 8 (1), 4–20. https://doi.org/10.14246/irspsd.8.1_4.

Kowarik, I., 2011. Novel urban ecosystems, biodiversity, and conservation. Environ.

Pollut. 159 (8), 1974–1983. https://doi.org/10.1016/j.envpol.2011.02.022.

Kowarik, I., 2018. Urban wilderness: Supply, demand, and access. Urban For. Urban

Green. 29, 336–347. https://doi.org/10.1016/j.ufug.2017.05.017.

Lampinen, J., Tuomi, M., Fischer, L.K., Neuenkamp, L., Alday, J.G., Bucharova, A.,

Cancellieri, L., Casado-Arzuaga, I., ˇ

Ceplov´

a, N., Cerver´

o, L., De´

ak, B., Eriksson, O.,

Fellowes, M.D.E., de Manuel, B.F., Filibeck, G., Gonz´

alez-Guzm´

an, A., Hinojosa, M.

B., Kowarik, I., Lumbierres, B., Klaus, V.H., 2021. Acceptance of near-natural

greenspace management relates to ecological and socio-cultural assigned values

among European urbanites. Basic Appl. Ecol. 50, 119–131. https://doi.org/10.1016/

j.baae.2020.10.006.

Liaw, A., & Wiener, M. (2002). Classication and Regression by RandomForest. R News

2/3. 〈https://www.researchgate.net/publication/228451484〉.

Luo, S., Patuano, A., 2023. Multiple ecosystem services of informal green spaces: A

literature review. Urban For. Urban Green. 81, 127849. https://doi.org/10.1016/j.

ufug.2023.127849.

Maller, C., Mumaw, L., Cooke, B., 2019. Health and social benets of living with ‘wild’

nature. In: Pettorelli, N., Durant, S.M., du Toit, J.T. (Eds.), Rewilding (1. Au., S.

Cambridge University Press, pp. 165–181. https://doi.org/10.1017/

9781108560962.009.

Massenberg, J.R., Schiller, J., Schr¨

oter-Schlaack, C., 2023. Towards a holistic approach

to rewilding in cultural landscapes. People Nat. 5 (1), 45–56. https://doi.org/

10.1002/pan3.10426.

Mathey, D.J., Arndt, T., Banse, J., Rink, D.D., 2018. Public perception of spontaneous

vegetation on brownfields in urban areas—Results from surveys in Dresden and

Leipzig (Germany). Urban For. Urban Green. 29, 384–392.

Moxon, S., Webb, J., Semertzi, A., Samangooei, M., 2023. Wild ways: A scoping review to

understand urban-rewilding behaviour in relation to adaptations to private gardens.

Cities Health 1–15. https://doi.org/10.1080/23748834.2023.2218016.

Müller, A., Bøcher, P.K., Fischer, C., Svenning, J.-C., 2018. ‘Wild’ in the city context: Do

relative wild areas offer opportunities for urban biodiversity? Landsc. Urban Plan.

170, 256–265. https://doi.org/10.1016/j.landurbplan.2017.09.027.

Mumaw, L.M., Maller, C., Bekessy, S., 2017. Strengthening Wellbeing in Urban

Communities Through Wildlife Gardening. Cities Environ. (CATE) 10 (1), 1–18.

Noe, E.E., Stolte, O., 2023. Dwelling in the city: A qualitative exploration of the human-

nature relationship in three types of urban greenspace. Landsc. Urban Plan. 230,

104633. https://doi.org/10.1016/j.landurbplan.2022.104633.

Perino, A., Pereira, H.M., Navarro, L.M., Fern´

andez, N., Bullock, J.M., Ceaușu, S., Cort´

es-

Avizanda, A., van Klink, R., Kuemmerle, T., Lomba, A., Pe’er, G., Plieninger, T., Rey

Benayas, J.M., Sandom, C.J., Svenning, J.-C., Wheeler, H.C., 2019. Rewilding

complex ecosystems. eaav5570 Science 364 (6438). https://doi.org/10.1126/

science.aav5570.

Pettorelli, N., Debney, A., Durant, S.M., Hoffmann, M., Laughlin, B., Pilkington, J.,

Pecorelli, J., Seiffert, S., Shadbolt, T., & Terry, A. (2022). Rewilding our cities. ZSL

report, London, UK.

Pineda-Pinto, M., Kennedy, C., Collier, M., Cooper, C., O’Donnell, M., Nulty, F.,

Casta˜

neda, N.R., 2023. Finding justice in wild, novel ecosystems: A review through a

multispecies lens. Urban For. Urban Green. 83, 127902. https://doi.org/10.1016/j.

ufug.2023.127902.

Planchuelo, G., von Der Lippe, M., Kowarik, I., 2019. Untangling the role of urban

ecosystems as habitats for endangered plant species. Landsc. Urban Plan. 189,

320–334. https://doi.org/10.1016/j.landurbplan.2019.05.007.

Powers, D. (2008). Evaluation: From Precision, Recall and F-Factor to ROC,

Informedness, Markedness & Correlation. Mach. Learn. Technol., 2.

Preston, P.D., Dunk, R.M., Smith, G.R., Cavan, G., 2023. Not all brownelds are equal: A

typological assessment reveals hidden green space in the city. Landsc. Urban Plan.

229, 104590. https://doi.org/10.1016/j.landurbplan.2022.104590.

Randrup, T.B., Buijs, A., Konijnendijk, C.C., Wild, T., 2020. Moving beyond the nature-

based solutions discourse: Introducing nature-based thinking. Urban Ecosyst. 23 (4),

919–926. https://doi.org/10.1007/s11252-020-00964-w.

Riley, C., Perry, K., Ard, K., Gardiner, M., 2018. Asset or Liability? Ecological and

Sociological Tradeoffs of Urban Spontaneous Vegetation on Vacant Land in

Shrinking Cities. Sustainability 10 (7), 2139. https://doi.org/10.3390/su10072139.

Rupprecht, C.D.D., Byrne, J.A., 2014a. Informal urban greenspace: A typology and

trilingual systematic review of its role for urban residents and trends in the

literature. Urban For. Urban Green. 13 (4), 597–611. https://doi.org/10.1016/j.

ufug.2014.09.002.

Rupprecht, C.D.D., Byrne, J.A., Ueda, H., Lo, A.Y., 2015. ‘It’s real, not fake like a park’:

Residents’ perception and use of informal urban green-space in Brisbane, Australia

and Sapporo, Japan. Landsc. Urban Plan. 143, 205–218. https://doi.org/10.1016/j.

landurbplan.2015.07.003.

Rupprecht, C.D.D., Byrne, J.A., 2014b. Informal Urban Green-Space: Comparison of

Quantity and Characteristics in Brisbane, Australia and Sapporo, Japan. PLoS ONE 9

(6), e99784. https://doi.org/10.1371/journal.pone.0099784.

Sikorska, D., Cię˙

zkowski, W., Baba´

nczyk, P., Chorma´

nski, J., Sikorski, P., 2021. Intended

wilderness as a Nature-based Solution: Status, identication and management of

urban spontaneous vegetation in cities. Urban For. Urban Green. 62, 127155.

https://doi.org/10.1016/j.ufug.2021.127155.

Sikorski, P., Gawryszewska, B., Sikorska, D., Chorma´

nski, J., Schwerk, A., Jojczyk, A.,

Cię˙

zkowski, W., Archici´

nski, P., Łepkowski, M., Dymitryszyn, I., Przybysz, A.,

Wi´

nska-Krysiak, M., Zajdel, B., Matusiak, J., Łaszkiewicz, E., 2021. The value of

doing nothing – How informal green spaces can provide comparable ecosystem

services to cultivated urban parks. Ecosyst. Serv. 50, 101339. https://doi.org/

10.1016/j.ecoser.2021.101339.

Stanford, H.R., Garrard, G.E., Kirk, H., Hurley, J., 2022. A social-ecological framework

for identifying and governing informal greenspaces in cities. Landsc. Urban Plan.

221, 104378. https://doi.org/10.1016/j.landurbplan.2022.104378.

Stanford, H.R., Hurley, J., Garrard, G.E., Kirk, H., 2024. Finding the forgotten spaces:

Using a social-ecological framework to map informal green space in Melbourne,

Australia. Land Use Policy 141, 107114. https://doi.org/10.1016/j.

landusepol.2024.107114.

Statistik Austria, 2024: Population at the beginning of the year. Available from 〈https:

//www.statistik.at/statistiken/bevoelkerung-undsoziales/bevoelkerung/bevoelker

ungsstand/bevoelkerung-zu-jahres-/-quartalsanfang〉(last accessed 8th March,

2024).

Threlfall, C.G., Kendal, D., 2018. The distinct ecological and social roles that wild spaces

play in urban ecosystems. Urban For. Urban Green. 29, 348–356. https://doi.org/

10.1016/j.ufug.2017.05.012.

Tombolini, I., Munaf`

o, M., Salvati, L., 2016. Soil sealing footprint as an indicator of

dispersed urban growth: A multivariate statistics approach. Urban Res. Pract. 9 (1),

1–15. https://doi.org/10.1080/17535069.2015.1037340.

Vega, K.A., Küffer, C., 2021. Promoting wildower biodiversity in dense and green cities:

The important role of small vegetation patches. Urban For. Urban Green. 62,

127165. https://doi.org/10.1016/j.ufug.2021.127165.

Vuolo, F., Neuwirth, M., Immitzer, M., Atzberger, C., Ng, W.-T., 2018. How much does

multi-temporal Sentinel-2 data improve crop type classication. Int. J. Appl. Earth

Obs. Geoinf. 72, 122–130. https://doi.org/10.1016/j.jag.2018.06.007.

Vuolo, F., ˙

Z´

ołtak, M., Pipitone, C., Zappa, L., Wenng, H., Immitzer, M., Weiss, M.,

Baret, F., Atzberger, C., 2016. Data Service Platform for Sentinel-2 Surface

Reectance and Value-Added Products: System Use and Examples. Art. 11 Remote

Sens. 8 (11). https://doi.org/10.3390/rs8110938.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

Watson, C.J., Dumont, A., Fortier, J., Miaux, S., 2023. Informal natural greenspaces as

places for urban leisure: Perspectives, uses and values from Quebec, Canada. Urban

For. Urban Green. 90, 128135. https://doi.org/10.1016/j.ufug.2023.128135.

Wellmann, T., Haase, D., Knapp, S., Salbach, C., Selsam, P., Lausch, A., 2018. Urban land

use intensity assessment: The potential of spatio-temporal spectral traits with remote

sensing. Ecol. Indic. 85, 190–203. https://doi.org/10.1016/j.ecolind.2017.10.029.

Zoderer, B.M., Carver, S., Tappeiner, U., Tasser, E., 2020. Ordering „wilderness“:

Variations in public representations of wilderness and their spatial distributions.

Landsc. Urban Plan. 202, 103875. https://doi.org/10.1016/j.

landurbplan.2020.103875.

B.M. Zoderer et al.

Urban Forestry & Urban Greening 101 (2024) 128549

Mapping wild food foraging locations reveals urban green space preferences and avenues for edible city solutions

Article

Full-text available

Nov 2024

Enabling wild nature experiences in cities: A spatial analysis of institutional and physical barriers to using wild nature areas in Vienna, Austria

Article

Full-text available

Feb 2025
LANDSCAPE URBAN PLAN

Wild nature in cities can allow urban residents to experience and connect with nature close to their homes. Previous research has shown that wild nature areas (WNAs) can be found in various urban greenspaces, such as remnant vegetation in forests, wild corners in parks, or spontaneous vegetation on informal sites. However, little is known about their usability for urban residents and their potential to provide wild nature experiences across different urban environments. This paper examines the usability of WNAs in Vienna (Austria) through a spatial analysis of the institutional and physical barriers to use, considering barriers related to the availability, accessibility, and attractiveness of these greenspaces. A city-wide map of WNAs (n=1298) was overlaid with spatial data on the presence of institutional and physical barriers, and complemented with a hierarchical cluster analysis to identify WNAs that share similar usability patterns. The cluster analysis confirms that WNAs vary significantly with regard to usability and that their usability is frequently constrained by the presence of multiple barriers. Furthermore, the usability of WNAs tends to be lower in high-density built-up areas, indicating that the possibilities to experience wild nature in the inner city are limited both in terms of lack of availability and relatively poor usability. We conclude that improving the usability of WNAs demands tailored and integrated approaches to urban planning and greenspace management and provide suggestions of how the usability of each identified WNA cluster can be increased.

Identifying urban rewilding opportunity spaces in a metropolis: Chongqing as an example

Article

Full-text available

Feb 2024
ECOL INDIC

The worldwide urbanization has resulted in a massive loss of natural habitats, which is threatening biodiversity and socio-ecological sustainability. Urban rewilding is a strategy aiming to restore the complexity, autonomy and self-sustainability of urban ecosystems by interlinking ecological processes while minimizing or gradually reducing human disturbances. The benefits of urban rewilding are widely acknowledged; however, there remains a knowledge gap concerning the identification of suitable locations for urban rewilding and the methodologies for assessing the rewilding potential of urban areas. In this study, we proposed an approach that can be used to accurately identify urban rewilding opportunity spaces (UROS, the existing wildness areas and spaces offering ecological conditions conducive to the establishment of new urban wildness), by integrating data on spatial heterogeneity, biodiversity and human disturbance. UROS in Chongqing, China was evaluated and mapped at the city-wide scale, with a subsequent validation of the UROS results through an assessment of species richness of spontaneous herbaceous communities, which serve as the pioneer species in rewilding processes. The results revealed a positive correlation between high-scoring UROS and increased species richness of spontaneous herbs, thus confirming the feasibility of the UROS assessment method. Eleven primary types of UROS were identified, with the majority displaying scattered or linear distributions. Most very high-scoring UROS are characterized by expansive woodland expanses situated on the outskirts of high-density built-up areas and substantial water bodies within the built-up urban areas. Notably, nearly half of the UROS encompass areas of 900 m 2 or less, underscoring the importance of urban rewilding initiatives targeting not only the preservation of extensive UROS with robust connectivity but also the revitalization of fragmented UROS with limited areas situated within central urban regions. The methodology is applicable to offer insights and spatial guidance in the execution of urban rewilding initiatives and biodiversity conservation efforts, contributing to sustainable regional development in the context of rapid urbanization.

Wild ways: a scoping review to understand urban-rewilding behaviour in relation to adaptations to private gardens

Article

Full-text available

Jun 2023

Urbanisation is increasing, while global biodiversity is decreasing. Through ‘urban rewilding’ cities could help tackle this biodiversity crisis, while exploiting the benefits of urban nature for residents. Private residential gardens, which have potential to support significant biodiversity, should be a primary focus. Yet their proportion of vegetated space is decreasing through changes made by residents, negatively impacting biodiversity. Small adaptations to private gardens can turn them into wildlife habitat, but understanding residents’ behaviour is critical to developing intervention strategies for this. This paper presents a scoping review of existing literature on understanding intent-orientated, pro-environmental behaviours with a focus on rewilding in urban gardens. The literature is mapped to assess the state of knowledge; it is then coded, using the ‘COM-B’ model of behaviour, to identify the capability, opportunity and motivation factors forming barriers and facilitators to residents engaging in rewilding activity in their gardens. The results show that all COM-B factors need to be considered to understand urban rewilding behaviour, but that opportunity and motivation factors have more influence, particularly reflective motivation. They indicate that facilitators are more significant than barriers and highlight an important body of work that has implications for practice and policy aimed at influencing urban rewilding.

Beyond urban parks: Mapping informal green spaces in an urban–peri-urban gradient

Article

Full-text available

May 2023
LAND USE POLICY

Urban green space (UGS) is a broad and potentially very inclusive term. Although most discussions on UGS focus on urban parks, there are many other vegetated areas in cities. Many are not formally recognised and protected, hence they are called informal green spaces (IGS), and yet they provide ecosystem services similar to parks. Based on a detailed overview of the different categories of UGS, we shed new light on how the different categories of UGS are included in urban planning and provide a snapshot of the disturbing urban land use transition-from UGS to other uses-which is partly related to the lack of formal recognition of IGS. The novelty in our study is that we mapped and analysed all UGS in our case study area, following two rivers that cut through Lodz (Łódź), Poland-from one peri-urban area (partly rural), through the city centre, to the peri-urban area on the other side of the city. The urban structures and the gradient of buildings in the studied area coincide with the urban structures for the whole of Lodz; therefore, our research area is a representative part of the city and its heterogeneity. Out of the seventeen UGS categories we distinguished, only six are formally recognised and account for 24% of all areas covered by UGS. Only 7% of the area we identified as UGS is designated as different types of vegetated areas according to local zoning plans. The analysis highlights broader challenges related to the planning and management of urban green and blue infrastructures.

Multiple Ecosystem Services of Informal Green Spaces: A Literature Review

Article

Full-text available

Feb 2023

Informal green spaces (IGS) such as overgrown vacant lots and urban brownfields constitute a considerable amount of green resources in the city. Given that the increasingly competitive land use of the urban area, enhancing the potential ecosystem services (ES) of IGS through design and management practices is of critical importance. This literature review paper provides an overview of ES delivered by urban IGS: What types of IGS have been discussed? Which ES have been identified in IGS, what interrelationships among different types of ES have been identified, and how can the potential of IGS be enhanced by urban design practices? 112 scientific papers were analyzed for their 1) IGS terms applied, 2) ES studied, 3) current or potential ES discussed, 4) ES trade-offs, and 5) ES assessment methods. Through the review, we found that although different types of ES have been identified in IGS, most studies did not consider ES synergies and trade-offs. The few studies assessing trade-offs of ES in IGS mostly focused on large-scale IGS such as urban brownfields rather than on small-scale IGS such as vacant lands/lots. The literature review highlights two knowledge gaps for future research: the first one is to explore the design and management knowledge that integrate multiple ES in small-scale IGS based on the assessment of potential ES trade-offs and synergies; the second one is to develop the spatial assessment of ES trade-offs and synergies, which is the key to envision design and management interventions that optimize the benefits of IGS. The literature review promotes the acknowledgement of the term IGS through highlighting their value in ES provisioning and further outlines future research directions on small-scale IGS such as vacant lands/lots and patches of spontaneous vegetation.

Towards a holistic approach to rewilding in cultural landscapes

Article

Full-text available

Dec 2022

1. In recent years, rewilding has been attracting growing attention as novel, process-oriented concept supporting ecosystem restoration and biodiversity conservation. 2. Rewilding aims to strengthen the adaptive capacity of ecosystems by restoring natural processes and withdrawing anthropogenic interventions. Yet, diverging understandings, conceptions and definitions of rewilding result in a somewhat fuzzy concept. 3. So far, the scientific discussion focused primarily on biological and ecological effects and success factors of rewilding. However, particularly in Europe, which is characterised by densely populated areas and a long history of landscape cultivation , rewilding affects also socioeconomic and socio-cultural dimensions 4. Based on a synthesis of current scientific publications, we argue that rewilding should be understood as an increase in wildness, that is restoring the autonomy of natural processes and self-sustaining ecosystems in order to overcome the improper dualistic understanding of human-nature relationships in which humanity is outside of nature. 5. We identify knowledge gaps and emphasise the need for inter-and transdisci-plinary research on rewilding to develop a holistic approach to biodiversity conservation and ecosystem restoration. In this context, we particularly emphasise the temporal dynamics of changes in landscapes, which are often long-term and therefore difficult to monitor, the openness of rewilding processes and the associated uncertainty about end states, as well as the complexity of human-nature relations and the associated value pluralism.

Finding the forgotten spaces: Using a social-ecological framework to map informal green space in Melbourne, Australia

Article

Jun 2024
LAND USE POLICY

REWILDING OUR CITIES

Technical Report

Jan 2022

Wilding cities for biodiversity and people: a transdisciplinary framework

Article

Mar 2024
BIOL REV

Accelerating urbanisation and associated lifestyle changes result in loss of biodiversity and diminished wellbeing of people through fewer direct interactions and experiences with nature. In this review, we propose the notion of urban wilding (the promotion of autonomous ecological processes that are independent of historical land‐use conditions, with minimal direct human maintenance and planting interventions) and investigate its propensity to improve biodiversity and people–nature connections in cities. Through a large interdisciplinary synthesis, we explore the ecological mechanisms through which urban wilding can promote biodiversity in cities, investigate the attitudes and relations of city dwellers towards urban wild spaces, and discuss the integration of urban wilding into the fabric of cities and its governance. We show that favouring assembly spontaneity by reducing planting interventions, and functional spontaneity by limiting maintenance practices, can promote plant diversity and provide ecological resources for numerous organisms at habitat and city scales. These processes could reverse biotic homogenisation, but further studies are needed to understand the effects of wilding on invasive species and their consequences. From a socio‐ecological perspective, the attitudes of city dwellers towards spontaneous vegetation are modulated by successional stages, with grassland and woodland stages preferred, but dense shrubby vegetation stages disliked. Wild spaces can diversify physical interactions with nature, and enrich multi‐sensory, affective and cognitive experiences of nature in cities. However, some aspects of wild spaces can cause anxiety, feeling unsafe, and the perception of abandonment. These negative attitudes could be mitigated by subtle design and maintenance interventions. While nature has long been thought of as ornamental and instrumental in cities, urban wilding could help to develop relational and intrinsic values of nature in the fabric of cities. Wildness and its singular aesthetics should be combined with cultural norms, resident uses and urban functions to plan and design urban spatial configurations promoting human–non‐human cohabitation. For urban wilding to be socially just and adapted to the needs of residents, its implementation should be backed by inclusive governance opening up discussion forums to residents and urban workers. Scientists can support these changes by collaborating with urban actors to design and experiment with new wild spaces promoting biodiversity and wellbeing of people in cities.

Informal natural greenspaces as places for urban leisure: Perspectives, uses and values from Quebec, Canada

Article

Nov 2023

Finding justice in wild, novel ecosystems: A review through a multispecies lens

Article

Mar 2023

Though most cities, particularly in the Global North, have been intensely modified by human activities certain locations still exist in varied forms of abandonment or disinvestment, often allowing for new species assemblages to flourish. These urban novel ecosystems or informal wild spaces are often perceived as in-between or overlooked, calling into question their value and social-ecological role, while also creating tensions amongst different groups and stakeholders who share different visions for their use and management. Within these tensions, issues of justice and equity can be more pronounced and surface historic legacies of environmental contamination, inequitable development, and extraction. Despite this, very little is known about the social-ecological role informal wild spaces play in urban areas, and how best to interrogate and understand the equity and justice dimensions they elicit. To fill this gap in knowledge, this paper critically examines the literature on urban novel ecosystems in relation to justice, with a particular interest in multispecies justice. Through this analysis, gaps in the literature are exposed, while also arguing the informality, neglect and contestation of wild urban spaces provides opportunities to explore issues of access, benefits and harms, particularly in light of global climate and ecological crises. A systematic approach is utilized to search the literature, identifying 45 papers which are thematically analyzed under a justice lens. The study identifies three themes that thread throughout the literature: distributional injustices relate to perceptions and attitudes, which give rise or arise from injustices; the regeneration discourse focuses on a ‘new nature’, which is based on social-ecological displacement and devaluation; and the potential of urban wild spaces to generate new multispecies sensibilities. The paper concludes by discussing trends, gaps, and emerging discourses, and proposing a multispecies justice approach for urban planning through the learnings and engagement with urban wild, novel ecosystems.

Mapping wild nature areas to identify priority areas for urban rewilding in cities: A process-oriented approach

Abstract and Figures

Recommended publications

Enabling wild nature experiences in cities: A spatial analysis of institutional and physical barrier...

Construction of Recreation Opportunity Spectrum of River Beach IGS in Mountain City: A Case Study of...

Working with Wilderness: a promising direction for urban green spaces

Getting to Know Urban Wasteland—A Look at Vacant Lands as Urban Green Space in Japan