You are on page 1of 17

Landscape and Urban Planning 228 (2022) 104555

Contents lists available at ScienceDirect

Landscape and Urban Planning


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

Research Paper

Using agent-based modeling to assess multiple strategy options and


trade-offs for the sustainable urbanization of cultural landscapes: A case in
Nansha, China
Jingyi Liu a, Menghan Zhang b, Yu Xia a, Huisen Zheng a, Chongxian Chen a, *
a
College of Forestry and Landscape Architecture, South China Agricultural University, No. 483 Wushan Road, Tianhe District, Guangzhou 510642, China
b
School of Landscape Architecture, Beijing Forestry University, No. 35 Qinghua East Road, Haidian District, Beijing 100083, China

A R T I C L E I N F O A B S T R A C T

Keywords: Many cultural landscapes are undergoing inevitable urbanization. For these landscapes to attain sustainability,
Sustainability participatory planning processes are often required. However, different stakeholders may prefer different stra­
Urbanization strategies tegies and goals, making decision-making extremely difficult. This paper explores how three urbanization
Trade-offs
strategy options (based on lay stakeholders’ willingness, historical landscape pattern, and land-use planning) and
Stakeholder participation
Agent-Based Modeling
their possible trade-offs can affect cultural landscapes’ sustainability. Specifically, it examines 1) how these
Temporal patterns strategies’ varying adoption levels affect different landscape functions, and 2) how their different trade-offs
affect sustainability perceived by different stakeholders. We conducted a case study in the Nansha District of
Guangzhou, China. Based on interviews with lay and expert stakeholders, we developed an agent-based model to
experiment with a range of scenarios reflecting different trade-offs of the three strategies. With 11 indicators and
their stakeholder-weighted aggregation assessed in every simulation step, we analyzed the strategies’ effects and
compared five typical scenarios’ sustainability on two timescales. The results showed that the scenario priori­
tizing lay stakeholders’ willingness outperformed other scenarios in the sustainability assessment. Following
historical patterns performed well in the ecological aspect and in the near term, but it may fall short in intensive
urbanization from non-experts’ perspectives. Land-use planning was plausible as a static destination but inef­
fective during the entire process. The findings suggest that sustainable rural development should prioritize local
adaptations; landscape pattern preservation and land-use planning must consider place-specific processes based
on stakeholder participation. They also suggest the usability of agent-based modeling and time-series analysis in
cultural landscape and sustainability research.

1. Introduction demographic and economic changes. However, rapid urbanization may


irreversibly disrupt the structure, function, feedback, and identities of
Cultural landscapes, i.e., landscapes that have evolved from a long the original cultural landscapes (Rescia et al., 2010; Selman, 2012),
history of human-nature interaction, are appreciated for their high which in turn affects the well-being of local populations (Hasan et al.,
ecological, cultural, and aesthetic values (Agnoletti, 2014; Schulp et al., 2020). In this reality, rather than aiming for conservation or restoration
2019). While the term “cultural landscape” is most commonly associated of their previous states, sustainable urbanization has become a critical
with rural and agricultural contexts, it also applies to the entire land­ goal for these inevitably changing cultural landscapes (Fischer et al.,
scape mosaic with different degrees of human modification (Wu, 2010). 2012; Plieninger et al., 2014).
Many cultural landscapes, despite their agricultural origins, are under­
going profound transformations due to rapid land-use change, especially
1.1. Sustainable urbanization of cultural landscapes
urbanization (Plieninger et al., 2014; Yu et al., 2016). In the current
context, urbanization of cultural landscapes is often inevitable due to
The concept of “sustainable development” or “sustainability,”

* Corresponding author.
E-mail addresses: liujingyi@scau.edu.cn (J. Liu), zmh1993@bjfu.edu.cn (M. Zhang), xiayuscau@scau.edu.cn (Y. Xia), 339278543@qq.com (H. Zheng),
chongxian@scau.edu.cn (C. Chen).

https://doi.org/10.1016/j.landurbplan.2022.104555
Received 20 October 2021; Received in revised form 24 August 2022; Accepted 29 August 2022
Available online 5 September 2022
0169-2046/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

commonly defined as “development that meets the needs of the present Vecchiato, 2015; Yang et al., 2021). Lay stakeholders’ (e.g., non-expert
without compromising the ability of future generations to meet their rural and urban residents) voluntary decisions are internal drivers of
own needs’’ (WCED, 1987), emphasizes a reconciliation of human land-use changes, which depend on individuals’ ability and willingness
development and environmental protection in the long-term (NRC, to engage in optional land-use activities in various ways (Valbuena et al.,
1999). Three pillars or dimensions—environment, economy, and soci­ 2010). Integrating or prioritizing their perspectives in planning (known
ety—are frequently used as a conceptual framework for assessing and as “bottom-up planning”) increases sensitivity to place-specific social
promoting sustainability (Wu, 2013). However, the relationship be­ and environmental conditions (Rovai et al., 2016; Szetey et al., 2021;
tween environmental capital and socio-economic capital has been at the Van Berkel & Verburg, 2012). However, bottom-up planning may be
core of discussions regarding “weak sustainability” versus “strong sus­ limited by lay stakeholders’ immediate demands while neglecting
tainability” (Huang et al., 2015; Wu, 2013; Xie et al., 2020). Weak external influences on landscape change (Butler et al., 2015), rendering
sustainability permits mutual substitutability between natural and landscape-scale sustainability uncertain (Liu et al., 2021). Expert
human-made capitals as long as the total capital increases or remains the stakeholders (e.g., scholars, planners, and policymakers) typically focus
same. By contrast, strong sustainability assumes that the substitutability on goals and strategies at larger scales. Many scholars appreciate his­
of manufactured for natural capital is seriously limited by environ­ torical landscape patterns because they embody essential knowledge
mental capital, which makes a unique contribution to welfare (Daly about sustainable management (Antrop, 2005; Di Fazio & Modica,
et al., 2007; Ekins et al., 2003). Meanwhile, as the concept of ‘‘ecosystem 2018) and inspire new planning concepts (Kučera et al., 2015; Tang
services’’ (ESs), which aims to identify and evaluate the benefits people et al., 2019). Planners and policymakers usually rely on executive-
obtain from ecosystems (MEA, 2005), has been a bridge between envi­ ordered land-use planning for the optimized spatial arrangement of
ronment and society (Braat & de Groot, 2012), the ESs framework has human development and environmental conservation (Cao et al., 2021;
also been widely used for understanding and assessing natural and Gong et al., 2018). However, because historical patterns may lose their
human-made capitals and sustainability (Wiens, 2013). More recently, relationship to contemporary drivers of transformation (Selman, 2012),
the United Nations Sustainable Development Goals (SDGs) have envi­ their sustainability in rapidly changing contexts is not guaranteed.
sioned a global agenda for sustainability (United Nations, 2015). Expert-led and command-and-control planning tends to neglect local
While sustainability is relevant to various spatial scales, its research adaptive activities, thus jeopardizing the multifunctional services they
and application would be more operational at a landscape or regional provide (Agnoletti, 2014; Rovai et al., 2016).
scale (Wu, 2013). At this scale, SDGs are difficult to fully implement and
should be localized (Szetey et al., 2021). To attain sustainability of 1.3. Agent-based modeling (ABM) approach
specific landscapes, scholars have suggested focusing on ESs in area-
specific contexts (also known as “landscape services”) that integrate To facilitate decision-making under uncertainty, previous research
nature and culture, consider spatial patterns, and involve stakeholder has complemented participatory planning with exploratory modeling
values (Termorshuizen & Opdam, 2009; Vallés-Planells et al., 2014). (Moallemi & Malekpour, 2018; Van Berkel & Verburg, 2012). Besides a
From this perspective, Wu (2013) reviewed the development of the participatory backcasting process where goals and possible strategies
concept of “landscape sustainability” and defined it as “the capacity of a are envisioned, future outcomes of the optional strategies are explored
landscape to consistently provide long-term landscape-specific ESs for by model simulations; decisions can be made by identifying the strate­
maintaining and improving human well-being in a regional context and gies that reach closer to the goals (Verburg et al., 2016). So far, myriad
despite environmental and sociocultural changes.” In this sense, the different modeling approaches have been employed to simulate and
sustainability of a landscape would be the sustainability of the human- assess planning strategy options and trade-offs. However, many of these
valued ESs provision by that landscape. For cultural landscapes to approaches would considerably reduce the complexity of the in­
achieve sustainable urbanization, the landscape should maintain the teractions and feedbacks in social–ecological systems and fail to reflect
overall ESs provided by the original landscape despite land-use changes the paths to predefined scenarios in the model structure (Brown et al.,
during the rural–urban transition. From a strong sustainability stand­ 2013; Verburg et al., 2016). ABM is a tool that represents a large number
point, while environmental services might inevitably become compro­ of local decision-making units and their interactions and feedbacks with
mised for economic and social services in the urbanization of rural and their environments and socio-political settings in spatial–temporal
natural areas, sustainable urbanization should not unduly limit the contexts (Valbuena et al., 2010). Some cultural landscape studies have
ecological capacity of the entire landscape. applied the ABM approach to simulate alternative scenarios (involving
rural agents’ decisions in response to the environment under different
1.2. Participatory planning process policy settings) and evaluate their outcomes using indicators that reflect
specific targets (Tieskens et al., 2017; Valbuena et al., 2010; Van Berkel
While landscape-specific ESs provide a framework for comprehend­ & Verburg, 2012).
ing landscape sustainability, decision-making and planning remain Compared with rural landscapes, rural–urban fringe and rapidly
challenging. Complexity and uncertainty are prevalent in cultural urbanizing contexts pose additional challenges to sustainability
landscapes as coupled human-natural systems (Moallemi et al., 2020). decision-making (Dolley et al., 2020). Due to the diversity of stakeholder
The feasibility of planning is influenced by complex factors such as in­ types and administrative arrangements, more strategy options and goals
dividuals’ decisions, societal demands, policy and institutional settings, may arise in these contexts (Marshall et al., 2018). Rather than being
and environmental capacities (Van Berkel & Verburg, 2012). As a result, mutually exclusive, multiple strategy options may coexist with varying
sustainable planning for cultural landscapes often entails the partici­ adoption levels and trade-offs, creating a multidimensional choice space
pation of crucial stakeholders, i.e., the individuals or groups affecting or (Moallemi et al., 2020); diverse sustainability goals may be valued
affected by an issue (Hassenforder et al., 2016). Participatory planning differently among stakeholders (Hölting et al., 2020). As a result, rather
processes enable the co-creation of knowledge by different stakeholders, than just finding optimal strategies based on explicit criteria, decision-
which can increase the chances of more sensible decisions and successful making and planning may benefit from more specific and comprehen­
outcomes (Moallemi et al., 2020; Rovai et al., 2016; Tieskens et al., sive information about, for example, the effects of each strategy’s
2017; Van Berkel & Verburg, 2012). different adoption levels and of multiple strategies’ various trade-offs.
However, trade-offs are ubiquitous in participatory planning (Dolley Achieving this requires systematically experimenting with an array of
et al., 2020), which exist in the divergent objectives as well as the as­ scenarios in the multidimensional choice space (Moallemi et al., 2020).
sumptions about pathways or strategies (Singh et al., 2015), notably Assessment of the scenarios’ sustainability needs to consider diverse
between lay and expert stakeholders (Scott, 2011; Tempesta & landscape-specific ES indicators, as well as their aggregation or

2
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

landscape multifunctionality, based on different stakeholders’ demands traditional cultural landscape facing drastic urbanization. During the
(Hölting et al., 2020). Moreover, temporal patterns of ESs should be past millennium, people progressively migrated to this estuary area and
highlighted (Rouleau & Zupko, 2019; Stürck & Verburg, 2017), for reclaimed the deposited sand flats (Panyu Chorography Editorial Board,
sustainability should be assessed by the change process rather than by 1995, p. 268). Much like the famous Dutch polder landscapes, the
any time point (Haines-Young, 2000). landscape structure of Nansha (among other river delta regions in
This paper aims to address the challenge of cultural landscapes’ southeast China) is also characterized by polders, or diwei (enclosed
sustainable urbanization when various strategy options and goals exist. dikes around farmland or ponds), with canals in between and linear
With a case study conducted in Nansha, China, we explored how three settlements along the elevated banks (Nijhuis et al., 2021). While most
urbanization strategy options (based on lay stakeholders’ willingness, polder landscapes in the Pearl River Delta have vanished due to massive
historical landscape pattern, and land-use planning) and their various urbanization in the last 40 years, Nansha retains this structure and
trade-offs can affect a cultural landscape’s sustainability as perceived by benefits from its functions, such as food production, water delivery, and
different stakeholders (villagers, urban residents, and experts). We flood control, which also represents the cultural identity of this
specifically addressed two research questions: (1) How will varying landscape.
adoption levels of the three strategies affect multiple landscape-specific However, Nansha has been undergoing rapid and extensive urbani­
ESs? (2) How will different trade-offs between these strategies affect zation as a result of government policies (Cheng et al., 2017). With the
landscape sustainability from different stakeholders’ perspectives? development of the Nansha New District elevated to the national stra­
Following this introduction, Section 2 outlines the context of the tegic level in the “12th Five-year Plan,” the Urban Master Planning of
study area. Section 3 details the research methods: an agent-based Nansha New District (2012–2025) was proposed in 2012. According to
model was developed based on stakeholder interviews; a range of sce­ this ambitious plan, Nansha would become a platform for Guang­
narios was parameterized to reflect different strategy options’ adoption dong–Hong Kong–Macao cooperation, an economic growth node in
levels and trade-offs; eleven indicators and a multifunctionality indica­ southern China, and a regional ecological, transportation, and service
tor were used to evaluate the impacts of the strategies on landscape center. From 2012 to 2025, the population was projected to increase
sustainability over different timescales. Section 4 presents the results of from 620,000 to 2.3 million, with the urbanization degree increasing
the study. Finally, Section 5 discusses the implications of the results for from 65 % to 95 %.
sustainable policymaking and planning practice. This rapid urbanization, being an imperative, also faces significant
challenges. Although Nansha’s development “has proceeded at a speed
2. Study area and scale never before seen in the history of the world” (Li, 2020, p.
109), the intended speed would still be difficult to accomplish based on
The study was conducted in the Nansha District (784 km2), which is current trends. According to the Guangzhou Statistical Bureau, Nansha’s
located in the south of Guangzhou City in Guangdong Province, China, population and urbanization degree were only 846,600 and 73.28 % by
and borders the Pearl River estuary (Fig. 1). Nansha is an example of a 2020, with annual growth rates of 5.244 % and 0.19 %, respectively, in

Fig. 1. Study area, including satellite imagery showing the polder landscape structure.

3
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

the previous 5 years. Due to the vast gap between the current situation 3.1. Stakeholder interviews
and the planning goal, reaching the targeted population and urbaniza­
tion degree will primarily depend on the influx of migrants (Yang et al., Stakeholders in this study involved both lay stakeholders (i.e., local
2019). However, most people consider Nansha a rural eco-tourism villagers, local urban residents, and potential urban residents) and
destination (Liu et al., 2019), and only a few people choose to move expert stakeholders (i.e., scholars, planners, and government officials).
there (Yang et al., 2019). Consequently, the latest “14th Five-year Plan” To collect data on their perspectives and knowledge, a total of 137
for the Nansha District released in May 2021 had to reduce the popu­ questionnaire-based interviews were conducted with 127 lay stake­
lation target for 2025 to 1.2 million, postponing the previously projected holders (63 villagers and 64 urban residents) and 10 experts in February
2.3 million as a long-term goal. and March 2022. Before the main fieldwork, pretest interviews were
Nansha’s current planning, based on zoning principles, promotes conducted to help refine the questionnaire and interview process by
intensive land use in both urban and rural areas. While the master plan gathering information and response options (allowing interviewees to
preserves natural land patches and the river-canal system, many low- add options) and identifying possible misunderstandings (Buschle et al.,
lying farmlands and ponds were occupied by the urban development 2021).
zones, and the traditional linear settlement structure was ignored. This Table 1 outlines the questionnaire structure, including three parts
urbanization approach has jeopardized many ESs that the traditional (questionnaire details in Supplementary file - Appendix A). Part I
landscape provided, leading to ecological land degradation (Zhang collected lay stakeholders’ basic information and behavior options and
et al., 2020), increased flood risk (Chen et al., 2020), land subsidence preferences in urbanization. Part II gathered their perceptions of suit­
(Ma et al., 2019), and a loss of aesthetic and tourism values (Xu et al., able urban development locations. The results of the two parts were used
2021). Moreover, implementing this top-down planning relies on to parameterize an agent-based model. Part III focused on the demands
expropriating farmland and village assets while compensating affected and objectives of both lay and expert stakeholders in future urbaniza­
villagers. In the last decade, a majority of villages in Nansha have had tion, which helped select indicators and assign weights in sustainability
virtually all of their agricultural land expropriated (Yuan, 2017, p. 14); assessment.
most inhabitants of such villages are rural residents only in legal terms The lay stakeholders were interviewed face to face in different vil­
and have depended primarily on industry and commerce for a living lages and towns across Nansha’s township areas (excluding Longxue, a
rather than farming (Zhou & Zhou, 2015). Some of these villages are in newly reclaimed port area with no villages or urban residence). The 63
the process of transitioning to cities, while the current planning en­ villagers were selected through a stratified sampling method. As the
courages relocating urbanized villagers to nearby towns, which may pretests revealed that villagers’ answers could be influenced by their
uproot village communities, affect former villagers’ income sources, and proximity to existing urbanized areas, the total rural population was
harm local cultural diversity, social relations, and sense of place (Wong divided into three groups based on their villages’ distances to the nearest
et al., 2021). urbanized areas: adjoining (<0.5 km), near (0.5–2 km), and far (2–5
km). In each of the eight township areas, we chose one spot for each
3. Methods group (adding up to 21 spots). In each chosen spot, we interviewed three
randomly encountered persons whom we confirmed to be native vil­
To address the uncertainty of the future, we employed a “research lagers. Among the 64 urban residents interviewed, 32 were locals, with
through design” process (Liu et al., 2021; Maher et al., 2018; Stremke & four encountered at random in each town and confirmed to be local
Schöbel, 2019), which draws on design—envisioning landscape changes urban residents; the other 32 were randomly chosen on the streets of
for sustainably providing ESs while meeting societal needs and downtown Guangzhou and confirmed to be interested in moving to
respecting societal values (Nassauer & Opdam, 2008)—to conduct sus­ Nansha (interviewee selection details in Supplementary file - Appendix
tainability experiments (Wu, 2013). This research included three pha­ A). To promote mutual understanding in the interviews with lay stake­
ses: stakeholder interviews, scenario design, and sustainability holders, we asked questions orally in plain language, providing relevant
assessment (Fig. 2). An ABM approach was applied to simulate and background information (e.g., urbanization and compensation policy)
assess the urbanization scenarios over time. Stakeholders’ perspectives and explanations (e.g., meanings of factors and services); interviewees
and demands obtained through interviews were incorporated into all were encouraged to elaborate on their answers.
stages in various forms. Furthermore, based on public information online, we identified 15
expert stakeholders (five each of scholars, professional planners, and
government officials) who have been involved in the research, planning,

Fig. 2. Diagram of the method integrating the ABM approach into a “research through design” process.

4
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Table 1 urbanization, suitability analyses for urbanization were conducted


Brief structure of the questionnaire. based on villagers’ and urban residents’ spatial preferences. For the
Interviewees Questions Options impacting factors identified by each of the two stakeholder types, 10
scores from 10 to 100 were assigned spatially by reclassifying the cor­
Part I
responding factor maps to reflect the stakeholders’ qualitative judg­
Villagers Confirming identity as ments obtained through interviews. We used the Analytic Hierarchy
native villager
Whether they own - None or few
Process (AHP) to assign weights for the factors (Saaty, 1990). Although
agricultural lands - More than 1 mu (0.067 several more objective methods exist for determining weights, including
ha) regression analysis, principal component analysis, data envelopment
(Informing them of - Yes, within 1 year analysis, and projection pursuit (Krajnc & Glavič, 2005; Lu et al., 2021),
urbanization and - Yes, within a decade
we used the AHP because it combines quantification with qualitative
compensation policy) - No
Whether and how soon they aspects of problems (stakeholders’ subjective perceptions), which play
wish to become urban an important role in ABM. During the AHP, different stakeholders’
residents opinions can be systematically extracted by assessing the relative
(If willing to urbanize, then) - Resettle in current importance of factors in a pairwise comparison matrix with 1–9 scales,
Whether they prefer to location
resettle in place or relocate - Relocate to a nearby town
and numerical weights were generated by matrix operation. Meanwhile,
- Depends on future in cases where conflicting factors may be valued similarly, the AHP can
condition help reflect the underlying trade-offs between these factors (Nardo et al.,
Urban residents Confirming identity as local - Local urban resident 2005) by comparing them within a limited range while being enough
or potential urban residents - Not local but interested in
sensitive to make a distinction (Krajnc & Glavič, 2005). For each
moving to Nansha
Whether they prefer to live - Mixed residence with stakeholder type, the perceived urbanization suitability (S) was then
with resettled villagers resettled villagers calculated by the weighted overlay of scored factor maps:
- Residence with only ∑n
urban residents S= (wi × si )
i=1
Part II
Villagers and urban How and how much - 12 factors chosen/
where n is the number of factors influencing urbanization suitability, wi
residents different factors influence provided in pretests
their location choice - Qualitative judgments for is the weight of factor i, and si is the score of factor i.
“far/near,” “high/low,” or The AHP was conducted after the interviews by the authors using the
“steep/flat” average Likert scale scores rather than by the interviewers directly
- Rating importance on 1–5 because pairwise comparisons of multiple factors are time-consuming
Likert scales
Part III
and possibly inconsistent. Furthermore, since the impacts of land-use
Villagers, urban How much they demand - - 11 services chosen/ change may affect future land-use changes as a result of feedback
residents, and different services from the provided in pretests mechanisms (Verburg, 2006), the model iterated the suitability calcu­
experts landscape - Rating importance on 1–5 lation at every time step to reflect this.
Likert scales

3.2.2. Autonomous agent behaviors


or management of Nansha. We interviewed them via email or phone and The modeling began with an endogenous urbanization process, in
received responses from five scholars, four planners, and one govern­ which agents make decisions about whether and how to urbanize and
ment official (anonymous list of respondents in Supplementary file - where to settle based on their own willingness, without external re­
Appendix A). strictions from historical landscape pattern or land-use planning
(Table 2).
3.2. Agent-based modeling In each time step before the rural population reaches the planning
target, all villagers first determine whether or not to become urban res­
Based on the interviews, we developed an agent-based model to idents with probabilities based on interviews. Villagers who decide to
explore the impacts of alternative urbanization strategies and scenarios become urban residents then have two options: 1) rebuilding dwellings
reflecting their trade-offs. The modeling software used was NetLogo. on the original plot and 2) relocating to undeveloped plots adjoining
The study area (784 km2) was modeled at 100-m resolution. Due to the nearby towns with the highest perceived urbanization suitability. Based
large population (846,600 in 2020) and the high proportion of unreg­ on interviews, the choice of a villager was assumed to have a probability
istered residents in the population (42 %), we were unable to obtain the determined by the urbanization suitability of their current location (the
exact spatial distribution of all individuals in the whole district. As a higher the suitability, the more likely they will stay put). By making this
simplification, we assumed that the urban population (620,388) is choice, the villager becomes a rebuilder or relocator, who then converts
evenly distributed across the urban plots, as is the rural population the chosen location into an urban plot. In either case, the land-use
(226,212) across the village plots. Since rural land in China is collec­ transition and resettlement process were assumed to face little resis­
tively owned, land-use decisions are typically made by village commu­ tance because virtually all undeveloped lands are currently state-owned
nities or collective households rather than by individuals (Tang, 2015). (Yuan, 2017, p. 14) and the government covers all development costs
As a result, an agent in this model represents a collective of individuals according to current policy.
in a 100-m-by-100-m plot, which would be an appropriate size for a Based on the urbanization process of Guangzhou (Liu et al., 2019),
collective decision-making unit in a rural village (containing an average we assumed that incomers (migrants from surrounding areas) first fill in
of 38 inhabitants) or an urban development project. The agent typol­ the already urbanized plots before more urban plots are developed.
ogies were identified based on observations and interviews, being vil­ There are two types of these urbanized plots: 1) the resettlement housing
lagers, rebuilders, relocators, and incomers. The agents perform optional plots, where only a small part is taken by urbanized villagers and the rest
behaviors with probabilities by time steps, with one step representing becomes commercial housing (Tang, 2015), and 2) the existing urban
one year (model details in Supplementary file - Appendix B). areas that emerged during the previous decade’s extensive construction,
which are still underpopulated (142 m2/person, on average) compared
3.2.1. Stakeholder-perceived urbanization suitability with the planned density (90 m2/person). The choices of incomers in a
To parameterize the probabilities of agents’ location choices in step are determined by interview-based proportions. When all the

5
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Table 2 except those in planned industrial zones; one urban plot can have mixed
The rules for voluntary agent behaviors without external restrictions. functions except that industry and tourism are incompatible; any urban
Agent How it Behavior How it is How it plot’s economic functions will be influenced by its neighbors based on
appears options determined affects plots the agglomeration effect (Wang et al., 2015).
Villagers Existing in Urbanize or not Whether rural –
village plots population 3.2.3. Three optional urbanization strategies
meets In addition to the endogenous urbanization process based on lay
projection; stakeholders’ willingness, we modeled the strategies of guiding urban­
probabilities
based on
ization with historical landscape patterns and land-use planning by
proportions in imposing restrictions on agents’ autonomous behaviors, yielding three
interviews optional strategies (Fig. 3):
(Those who Probabilities –
choose to based on
1) Willingness-based strategy (Fig. 3a):
urbanize) villager-viewed
rebuild or urbanization
relocate suitability of This strategy assumes that the endogenous urbanization process
their current described above is unrestricted. All agents make decisions based on their
locations willingness about whether to urbanize, whether to rebuild in place or
Rebuilders Sourced Rebuild houses Change land
relocate, and where to live based on their perceived suitability. In

from villagers use;
who decide population following this strategy, future urban areas are likely to gather organi­
to urbanize remains cally around or along existing landscape elements or service facilities.
in place
Relocators Sourced Where to Villager-viewed Change land
2) History-based strategy (Fig. 3b):
from villagers relocate urbanization use and
who decide suitability; increase
to urbanize neighboring population This strategy assumes that urban growth follows the inhabiting
and move existing urban process in response to the historical landscape pattern. Villagers can still
areas choose whether to urbanize and whether to rebuild or relocate. How­
Incomers Appearing Settle in Whether existing
ever, the location choices of relocators and incomers, based on their

with current existing or new urban areas meet
growth rate urban plots planned density perceived suitability, are restricted by the polder structure. Specifically,
(Settle in Proportions in Increase agents first choose the undeveloped plots immediately beside existing
existing plots) interviews population canals; when all these plots are occupied, they then choose locations one
live in
plot away from canals, and so on. In this way, new urban areas extend
resettlement
plots or original linearly, beginning beside canals before gradually reaching the interiors
urban plots of polders. It is worth noting that the history-based strategy differs from
(Settle in new Urban resident- Change land the willingness-based strategy because local population are not neces­
plots) where to viewed use and sarily willing to adopt the traditional settlement development approach
settle urbanization increase
in the context of future urban development. In addition, incomers from
suitability, population
neighboring other regions may not value the historical landscape pattern when
existing urban making willingness-based decisions.
areas
3) Planning-based strategy (Fig. 3c):
urbanized areas meet the planned population density, additional urban
This strategy assumes that urbanization completely follows the cur­
areas will be developed for incomers on undeveloped plots next to
rent land-use planning, which promotes intensive development and
already urbanized areas with the highest perceived suitability. The land-
natural land conservation through zoning. Villagers can still choose
use development process was also assumed to be unhindered, as urban
whether to urbanize. However, as the planning requires “actively
development is encouraged and profitable.
guiding an aggregation of villagers to central cities and towns,” urban­
Additionally, to simulate the layout and richness of urban functions
ized villagers must relocate rather than rebuild in place. Relocators and
(the main categories being residence, commerce, industry, and tourism),
incomers can only choose locations, albeit based on their perceived
the model assumed that all urban plots have a residential function,

Fig. 3. Schematic diagrams of the three optional urbanization strategies based on a) lay stakeholders’ willingness, b) historical landscape pattern, and c) land-
use planning.

6
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

suitability, within the urban development zones designated by the Land scenarios’ performances with regard to sustainability. Eleven ES in­
Use Master Plan of Nansha New District, adjoining already urbanized dicators were selected to assess the optional strategies’ impacts in
plots. different aspects. The indicators were then aggregated using a multi­
functionality index to measure the overall sustainability of different
3.2.4. Scenarios reflecting trade-offs between strategies scenarios. These indices were calculated at every simulation step to
The above strategies may coexist with different combinations of analyze the time pattern of sustainability.
trade-offs. To simulate this, we introduced two variable parameters,
Willingness-History (W-H) and Willingness-Planning (W-P), to represent 3.3.1. Indicator selection, calculation, and aggregation
spectrums of potential external influences of the history- and planning- It has been suggested that landscape sustainability assessment should
based strategies on the endogenous process. Both parameters have five be relevant to human well-being and land management (Cao et al., 2021;
adjustable integer values ranging from 0 to 4. Larger values represent Wu, 2013). Therefore, indicators in this study were selected and
increased probabilities (0, 25 %, 50 %, 75 %, and 100 %) of an agent developed based on the demands and knowledge of stakeholders (lay
complying with the historical or planning constraints or decreased and expert) acquired in the interviews, reflecting the environmental,
probabilities (100 %, 75 %, 50 %, 25 %, and 0) of acting autonomously economic, and social dimensions of sustainability (Table 4). While there
(except on the newly reclaimed Longxue Island, where incomers follow have been suggestions of a cultural dimension for assessing the sus­
current planning regardless of the variable values). tainability of cultural landscapes (Selman, 2007), cultural ESs were
The two parameters are independent but work together to influence considered in the social dimension in this study.
each agent’s choices with combined probabilities. Compared with sim­ As the assessment aimed to compare the alternative strategies’
ply adjusting the proportions of choices, this approach may better reflect relative performances rather than provide a thorough ES valuation, the
the three strategies’ interactions in reality. For example, if W-H = 1 and indicators measured ESs at the landscape level and were kept pragmatic.
W-P = 1, i.e., both the history- and planning-based strategies have a 25 Some indicators were landscape metrics or other established indices.
% chance of influencing or a 75 % chance of not influencing agent be­ Meanwhile, some place-specific demands, which were difficult to
haviors, the probabilities of different choices for any particular agent quantify directly, were estimates or proxies based on existing metrics or
would be 56.25 % acting voluntarily, 18.75 % following traditional other available data. Most of the indicators were calculated directly by
pattern, 18.75 % conforming to planning, and 6.25 % adhering to both the agent-based model, while those based on landscape metrics (CIf,
tradition and planning. SEDw, PCIu, and IJIu) were calculated with Fragstats v4 using landscape
In total, there are 25 possible combinations of the values of W-H and pattern datasets exported during the simulation process. For a majority
W-P, covering all the conceivable scenarios relating to the research of the indicators, higher values can be interpreted as increased perfor­
questions. Although some of the scenarios may not be realistic (i.e., W-H mance, except for DDu, EVu, DMu, and DRu (measuring distance or
or W-P equals 0 or 4 when all agents follow the same strategy), all of cost), of which smaller values signify better performance.
these scenarios were simulated in order to systematically analyze the To measure and compare the overall provision of the multiple
potential effects of the varying adoption levels of the three strategies. landscape-specific ESs or landscape sustainability, we employed the
Further comparisons were then limited to five relatively realistic sce­ Multiple Ecosystem Services Landscape Index (MESLI) to aggregate the
narios that reflect typical trade-offs between the strategies: willingness- above ES indicators, which is calculated as the sum of normalized in­
first, history-first, planning-first, balanced, and low-willingness (Table 3). dicators (Rodríguez-Loinaz et al., 2015; Stürck & Verburg, 2017). In
Considering that the development of Nansha, required by state and addition, to reflect the diverging demands of different stakeholder
local governments, would be an unavoidable reality (Cheng et al., groups (villagers, urban residents, and experts), we modified the MESLI
2017), the modeling took the achievement of the government’s urban­ with a weighted approach, in which ES indicators were weighted ac­
ization goal as a prerequisite. The population growth rate and ultimate cording to their stakeholder-perceived importance (Hölting et al., 2020;
goal were control variables for all scenarios, and their simulated out­ Manning et al., 2018). Although more objective methods to assign
comes were only influenced by W-H and W-P, i.e., the adoption levels of weights for developing aggregated sustainability indices exist (e.g., Lu
the three strategies. Specifically, the population growth rate was a et al., 2021), this study used AHP based on the average Likert scale
constant (5.244 %) based on the previous 5-year average, the villagers ratings in interviews in order to reflect different stakeholders’ opinions
stopped participating in urbanization when the target rural population and trade-offs. The weighted MESLI is then calculated as:
(115,000) was met, and simulations ended when the total population ∑n ( ESi − ESimin )
reached 2.3 million. weighted MESLI = × wi
i=1 ES
imax − ESimin

3.3. Sustainability assessment


where n is the number of ES indicators, ESi is the indicator i, wi is the
weight for indicator i, and ESimin and ESimax are the minimum and
As landscape sustainability is defined as the consistent provision of
maximum observed values of ESi, respectively. The indicators with
landscape-specific ESs, the ES approach was applied to assess the

Table 3
The five focused scenarios that reflect typical trade-offs between the strategies.
Scenario Variable Description Probabilities of an agent’s behavior options
values Willingness- History- Planning- History- &
based based based planning-based

Willingness- W-H = 1, W-P Low constraint levels of history and planning on voluntary 56.25 % 18.75 % 18.75 % 6.25 %
first =1 behaviors
History-first W-H = 3, W-P High constraint level of historical pattern on voluntary 18.75 % 56.25 % 6.25 % 18.75 %
=1 behaviors, low adoption of planning
Planning-first W-H = 1, W-P High constraint level of planning on voluntary behaviors, low 18.75 % 6.25 % 56.25 % 18.75 %
=3 adoption of history strategy
Balanced W-H = 2, W-P Medium constraint levels of history and planning on voluntary 25 % 25 % 25 % 25 %
=2 behaviors
Low- W-H = 3, W-P High constraint levels of history and planning on voluntary 6.25 % 18.75 % 18.75 % 56.25 %
willingness =3 behaviors

7
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Table 4
The ES indicators for sustainability assessment and their calculation methods and implications.
Service Indicator Calculation Explanation Implication

Environmental Dimension
∑n
Water Rainwater Storage (RS) RS = i=1 (1 − ri ) × a × p n is the number of plots; ri is the runoff Higher values indicate increased water
management coefficient of plot i; a is the area of a plot; p storage, underground water recharging and
is the mean annual precipitation land subsidence mitigation (Ao et al., 2015)
∑n
Mean Drainage Distance i=1 dwi n is the number of urban plots; dwi is the Lower values indicate better drainage,
DDu =
of urban plots (DDu) n distance to the nearest waterbody of urban reducing flood risk
plot i
∑n
Biodiversity Connectance Index of i=1 ci n is the number of farmland patches; ci Higher values indicate increased
CIf = × 100
farmland patches (CIf) n × (n − 1) equals 1 or 0 dependent on whether connectivity of farmland patches (potential
2 connections (within 200 m) exist between habitat changing), promoting biodiversity
farmland patch i and other farmland
patches
∑n
Similarity-weighted i=1 (ei × si ) n is the number of patch types; ei is the Higher values indicate increased amount
SEDw =
Edge Density of A length of edge between patch type i and and quality of riparian areas, promoting
waterbody patches water patches; si is the similarity weight biodiversity
(SEDw) between patch i and water; A is the total
landscape area
Economic Dimension
∑n
Crop production Mean Normalized i=1 NDVIi n is the number of farmland plots; NDVIi is Higher values indicate increased crop
NDVIf =
Differential Vegetation n the NDVI of farmland plot i productivity (Benedetti & Rossini, 1993)
Index of farmland plots
(NDVIf)
∑n
Construction cost Earthwork Volumes of EVu = i=1 (h − hi ) × a n is the number of new urban plots;‾h is Lower values indicate less earthwork
savings new urban plots (EVu) the average elevation of existing urban required to reach safe altitude (main source
plots of the same town; hi is the elevation of construction cost variation)
of new urban plot i; a is the area of a plot
∑n
Mixed urban Mean Function Types in i=1 fi n is the number of new urban plots; fi is the Higher values indicate increased diversity
FTu =
functions new urban plots (FTu) n number of function types in new urban of urban functions, promoting diverse
plot i economic activity, income streams, and
employment (Raman & Roy, 2019)
Urban cohesion Patch Cohesion Index of PCIu = n is the number of urban plots; pi and ai are Higher values indicate increased
( ∑n )( )
urban patches (PCIu) pi 1 − 1 the perimeter and area of urban patch i; A aggregation of urban patches, promoting
1 − ∑n i=1√̅̅̅̅ 1 − √̅̅̅̅ × is the total landscape area efficient economic processes (He et al.,
i=1 pi ai A
100 2020)
Social Dimension
∑n
Regional Mean Distance to i=1 dmi n is the number of urban plots; dmi is the Lower values indicate increased availability
DMu =
connection Motorways of urban n distance to the nearest motorways of to motorway, estimating connection to
plots (DMu) urban plot i adjacent regions
∑n
Transportation Mean Distance to Roads i=1 dri n is the number of urban plots; dri is the Lower values indicate increased availability
DRu =
convenience of urban plots (DRu) n distance to the nearest roads of the urban to main roads, estimating transportation
plot i convenience
( )
Aesthetics and Interspersion & ∑n e ei n is the number of patch types; ei is the Higher values indicate more contacts of
− i=1 ∑n i × ln ∑n
social relations Juxtaposition Index i=1 ei i=1 ei total length of edge between patch type i urban patches with natural and village
IJIu =
between urban and other ln(n − 1) and urban patches patches, estimating potentials for good
patches (IJIu) views, ecotourism, cultural diversity, and
social relations

Note. In this table, “patch” refers to its landscape ecological meaning, i.e., a landscape area different from surrounding areas, while the pixel grid in the agent-based
model is referred to as “plot” instead of “patch”.

smaller values indicating better performance were subtracted from 1 To make the predictors’ impacts on different indicators more compa­
after normalization. The weighted MESLI ranges from 0 to 1, with higher rable, the indicators were normalized to their minimum and maximum
values indicating increased sustainability, which can only be interpreted values in the simulation outcomes of the 25 scenarios (those with
in relative terms in this study. smaller values indicating better performance were subtracted from 1).
The regressions were conducted on two timescales with multi-year av­
3.3.2. Time-series data analysis erages so that the different urbanization processes, rather than just the
To assess the sustainability of the dramatically changing landscape, final state, affected the result.
all indicators were calculated at every simulation step, with the resultant The MESLI weighted by different stakeholder groups’ opinions was
time-series data analyzed on two timescales (a short and a long time­ also calculated for every time step using the 11 ES indicator outputs
scale). Since the model involves randomness to reflect probabilistic (with the median of 100 simulations). We conducted trend surface
decisions, simulations of the same scenario would not produce identical analysis (using quadratic polynomials) to analyze the relationship be­
outcomes. As a result, each of the 25 scenarios was simulated 100 times, tween the overall landscape sustainability (multi-year average MESLI)
yielding a data group (n = 100) for each indicator in each step. On both and different W-H and W-P combinations. The overall sustainability of
timescales, ANOVA and Kruskal-Wallis H tests were conducted (with the the five focused scenarios was then compared. The analysis and com­
multi-year average) to identity whether there was a significant differ­ parison were also conducted on the two timescales.
ence between each indicator generated by simulations of the 25
scenarios. 4. Results
To explore the influences of different adoption levels of the strategies
on the ES indicators, we used multiple linear regression with ordinary 4.1. Interview result analysis
least squares to analyze the correlations between each indicator (with
the median of the 100 simulations) and two predictors of W-H and W-P. Part I of the interviews revealed that, among the 63 villager

8
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

interviewees, up to 89 % own<1 mu (0.067 ha) of farmland due to land 4.2. Land-use pattern and amount change
expropriation. Seventy-three percent were willing to progressively
become urban residents to improve their living standards and social Simulations of different scenarios produced varied spatial pattern
positions, of which 6.5 % were willing to urbanize within a year. Among changing processes and outcomes. Three single-strategy scenarios are
the willing-to-urbanize villagers, 61 % made a choice between shown in Fig. 6 to highlight the disparities between the optional stra­
rebuilding and relocating, in which the proportion of rebuilding declines tegies. Despite difference in patterns, since the population growth rate
as the distance from urbanized areas increases (84.6 %, 77.8 %, and 50 and target were control variables, simulations of all scenarios lasted the
% in adjoining, near, and far groups, respectively). This supported our same amount of time (19 steps), and their land-use area change curves
assumption that urbanized villagers are more inclined to rebuild than were similar (Fig. 7). It is worth noting that an abrupt changing point in
relocate when their current locations are more suitable for urbanization. the curves of new urban and farmland areas appeared in the ninth year.
Meanwhile, 39 % of villagers failed to express an explicit preference and Before this point, new urban areas increased slowly because the popu­
chose “depending on the future surroundings of my current location.” lation density of existing urban plots had not reached the planned ca­
Among the 64 urban interviewees, choices for living with resettled vil­ pacity. During this period, local villagers’ decisions were the main
lagers or only urban incomers were 62.5 % against 37.5 %, respectively. driving force for the land-use change. Thereafter, the land-use change
In the spatial factors for suitability analysis (Part II of the interview), accelerated, with the influx of incomers becoming the primary driver.
the adjacency to existing villages was valued (average Likert scale above Considering this change, the model outcomes were analyzed on two
2) only by villagers, while proximity to regional entrances/exits was timescales: a short scale covering years 1–9 and a long scale covering the
only appreciated by city dwellers. Among the same stakeholder group, entire period (years 1–19).
the qualitative judgments of the 11 valued factors were largely-one-
sided, with minorities accounting for relatively small proportions. The
4.3. Impacts of optional strategies on different ESs and sustainability
average number of any interviewee’s preferences for the 11 factors that
were identical with the majorities’ choices was 9.4 (85.2 %) for villagers
On both timescales, significant differences (p < 0.005) existed in all
and 9.5 (86.2 %) for urban residents. As a simplification, we used the
ES indicators between the 25 scenarios, and the multiple linear re­
majorities’ qualitative judgments and perceived importance (Fig. 4) in
gressions indicated strong relationships between all indicators and the
suitability analysis and modeling (otherwise there could be 121 very
W-H/W-P values, with Prob (F-statistic) < 0.005 (Table 5). For most of
similar suitability options for agents).
these relationships, the trends were identical on the two timescales, but
Fig. 5 displays the weights for the 11 ES indicators based on the three
the magnitudes were greater on the long timescale. Generally, higher W-
stakeholder groups’ ratings (Part III of the interview). The villagers paid
H values (increased adoption of the history-based strategy) positively
particular attention to the mixed urban functions (FTu) because most of
affected all environmental performances while negatively impacting
them rely on diverse income streams other than agricultural production.
most economic and social services. Higher W-P values (increased
They also emphasized the drainage distance (DDu) and rainwater stor­
adoption of the planning-based strategy), while increasing farmland
age (RS) because of a deep understanding of flood risk in this region. The
connectivity (CIf), urban cohesion (PCIu), and traffic convenience
urban residents prioritized many social and economic services, espe­
(DRu), diminished all other services. Both W-H and W-P negatively
cially the proximity to motorways (DMu) and roads (DRu) due to the
impacted crop production (NDVIf), urban function mixture (FTu), and
demand for connection to adjacent cities and efficient traffic. While lay
aesthetics and social relations (IJIu), which in turn suggested the
stakeholders mainly focused on their direct needs, the experts’ weight­
willingness-based strategy’s strengths. Meanwhile, their positive in­
ing scheme was relatively balanced, with a slight emphasis on the
fluences on farmland connectivity (CIf) indicated the willingness-based
environmental dimension.
strategy’s shortcoming.

Fig. 4. Preferences and weights for factors affecting urbanization suitability based on the majority of villagers’ and urban residents’ qualitative judgements and
perceived importance in the interviews (negative weights indicate an opposite preference).

9
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Fig. 5. Weighting schemes for the indicators based on the perceived importance expressed by different stakeholder groups (villagers, urban residents, and experts) in
the interviews.

The trend surface analysis demonstrates the relationship between the planning-first scenario still significantly underperformed the history-first
overall sustainability (indicated by stakeholder-weighted MESLI) and scenario. Referring to the regression results, an explanation for the
different strategy combinations (Fig. 8). In all subplots, conforming to divergence between the non-experts and experts could be that adhering
lay stakeholders’ willingness achieves the highest MESLI. As the adop­ to historical landscape pattern negatively affected many economic and
tion levels of the history-based strategy (W-H) and the planning-based social indicators that the non-experts emphasized unilaterally (notably
strategy (W-P) increase, the MESLI decreases to varying degrees. As FTu and DRu); however, it positively impacted environmental indicators
shown by the contour lines in Fig. 8, on the short-term timescale, valued by the experts (especially DDu and SEDw).
increasing W-P causes a sharper decline of the MESLI than increasing W- In this sense, the MESLI weighted by non-experts, particularly by
H. A similar trend can be observed for the villager- and expert-weighted urban residents, measured weak sustainability, whereas expert-
MESLI on the long-term scale, but not the urban resident-weighted weighted MESLI measured strong sustainability. This is further
MESLI. On both timescales, increasing W-H has a less negative effect demonstrated in Fig. 10. While the overall performances of willingness-
on the MESLI weighted by experts than that weighted by non-experts, and planning-first urbanization scenarios appeared to be more sustain­
because experts attached more values to the environmental indicators able according to the overall level of the 11 normalized indicators
that are positively correlated with W-H. (which equals the unweighted MESLI), the aggregated level of the
environmental indicators declined dramatically after the first decade.
4.4. Sustainability of focused scenarios Although the overall ES level in the history-first scenario dropped more
sharply in the last few years, the environmental indicators remained
Fig. 9 shows the temporal patterns of the stakeholder-perceived much higher levels.
overall sustainability (indicated by stakeholder-weighted MESLI) of It is also worth noting that after around the 15th–17th year, most
the five focused scenarios. During the entire process, most MESLI curves MESLI levels exhibited a rapid decline. Many of them decreased to levels
first rose before falling. On the short timescale, there was an agreement much lower than the current situation, which could indicate that the
among all stakeholder groups, despite their varied weighting schemes, massive urbanization goal has exceeded the ecosystem’s capacity. In this
on the five scenarios’ relative performances: the willingness-first scenario regard, rather than the currently projected population of 2.3 million, a
performed best, immediately followed by the history-first scenario. The total population of around 1.8–2 million and an urbanized area of
balanced scenario, trying to reconcile the three strategies, produced around 155 km2 that appeared in the 15th–17th year would be a more
average results. The planning-first scenario performed poorly, and the appropriate objective.
low-willingness scenario adhering to both the historical pattern and land-
use planning underperformed all others. 5. Discussion
On the long timescale, while the willingness-first scenario remained
the best and low-willingness the worst, disagreement appeared between We explored how the optional urbanization strategies based on lay
the rankings of the history-first and planning-first scenarios. As the curves stakeholders’ willingness, historical landscape patterns, and land-use
showed, the performance of the history-first scenario declined more planning, as well as their trade-offs, could affect the sustainability of
quickly in the second decade. Meanwhile, the planning-first scenario, the Nansha landscape. Based on the research findings, some implications
which performed poorly in the short term, improved to varying degrees. for sustainable policymaking and planning practice, especially in the
As shown by the long-term boxplots, the level of the planning-first sce­ Chinese context, can be discussed.
nario became very close to that of the history-first scenario based on
villagers’ weighting scheme and even surpassed the latter based on
urban residents’ weights. However, according to experts’ weights, the

10
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Fig. 6. Spatial patterns in steps 9 and 19 of three single-strategy scenarios (among all the 19 steps in the 25 scenarios simulated), illustrating the varied spatial
pattern changing processes and results affected by different urbanization strategies.

5.1. Deficiency of top-down land-use planning intensively cultivated land. The land-use plan also reduced the inter­
mixing of urban patches with natural patches (waterbodies, farmland,
Land-use planning based on functional zoning is a prevalent instru­ and forests) and traditional villages. This could be beneficial from a
ment in current Chinese urban and rural planning. It pursues intensive conservation-focused viewpoint, but it may jeopardize cultural services
land-resource utilization by spatially arranging human development and such as ecotourism, cultural diversity, and urban–rural integration.
natural protection based on landscape ecological principles (Wu et al., Furthermore, as the MESLI curves demonstrated, while land-use plan­
2020). Despite criticisms of its top-down manner, land-use planning is ning may perform well as a static goal, it limits the overall ES provision
favored for providing optimized spatial patterns, which are often proved during the entire process, which is what sustainability entails. As a
by landscape metrics (Gong et al., 2018). However, landscape analysis result, we suggest that land-use planning, despite its necessity as macro-
should not treat quantification of spatial patterns as an end and ignore level guidance, should be applied with caution in rural urbanization to
how patterns affect processes (Li & Wu, 2004). avoid compromising spontaneous and vibrant local processes.
This study found that land-use planning could be insufficient for
sustainable urbanization of cultural landscapes due to the difficulty of
adapting it to specific landscape processes. While increased adoption of 5.2. Advantage and future uncertainty of historical pattern
land-use planning enhanced urban patch cohesion, it also limited urban
function mixture, which is crucial for local economic activities and With the promotion of national policies such as “Rural Revitaliza­
employment generation. Although the planning improved farmland tion” and “New Urbanization,” the preservation of historical-cultural
connectivity, it lowered the amount and quality of riparian habitats, values has been highlighted in Chinese rural planning. However, poli­
which would be more important for biodiversity in this reclaimed and cymakers and planners in China often focus on iconic elements or relics
while ignoring the relatively “ordinary” landscape-scale context, which

11
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

moderate development stage and in ecologically sensitive areas rather


than in a one-size-fits-all manner. Adhering to historical patterns while
neglecting the practical needs of the general public could also be an
example of expert-led elitism. Meanwhile, from a strong sustainability
perspective, because the history-based strategy outperforms the others
in the environmental aspect, there is a need to promote communication
between experts and non-experts on the values of ecological and cultural
services to long-term human well-being.

5.3. Necessity of respecting bottom-up willingness

Since 2014, the “New Urbanization” policy has been promoting a


shift from “land-oriented” to “people-oriented” urbanization, with the
market rather than the government taking the lead. However, the pre­
vailing practice still relies on zoning plans enacted through land
expropriation on the one hand (Tang, 2015) and tax cuts, land subsidies,
and fiscal investment on the other hand (Shen & Shen, 2019). In this
case study, while the government’s urbanization agenda has been
accepted by most stakeholders, an impediment to “people-oriented”
urbanization would be restrictions in current planning on the urbanized
villagers’ and incomers’ choices of (re)settlement methods and locations
Fig. 7. Land-use change curves in the 25 scenarios are similar. An abrupt based on their willingness.
changing point exists in year 9 in the new urban and farmland areas, indicating The results showed that the willingness-first scenario, enabling more
different phases of the urbanization process. micro-scale adaptations than any one-size-fits-all approach, had the best
performance in the overall sustainability assessments from all stake­
could be an example of “expert-led elitism” (Scott, 2011). Nevertheless, holders’ perspectives. Although this scenario resulted in a less coherent
historical landscape patterns, with their values, spatiotemporal changes, urban morphology, it maximized urban function diversity, urban-nature
and drivers in regional and territorial contexts, have received attention contact and urban–rural integration, among other difficult-to-quantify
in Chinese academia in recent years (Guo et al., 2021; Tang et al., 2019; social benefits. As a result, we suggest that urbanization policies and
Wang & Lin, 2018), which has the potential to influence future policy­ planning, with necessary guidance regarding land-use amounts and
making and planning. spatial configuration, should give more credit to voluntary mechanisms.
Our results suggest that development based on the polder structure Alongside the prevailing method of relocating urbanized villagers to
can benefit continuous ecological ES provision and short-term overall towns, urbanizing in original locations can also be encouraged. Apart
performance. In addition, inheriting the historical pattern itself can from zoning, development can rely on market-based mechanisms that
provide many cultural ESs that are difficult to quantify (e.g., knowledge reflect stakeholders’ actual demands. To identify and integrate these
systems, educational values, and sense of place). If the goal is moderate bottom-up perspectives, policymaking and planning processes should
urbanization, the history-first scenario would be desirable in mitigating put more emphasis on stakeholder participation.
trade-offs between ecological protection, human development, and
cultural inheritance. However, inherited patterns may not guarantee 5.4. Weak versus strong sustainability of urbanization
sustainability in the face of drastic changes (Selman, 2012). As the re­
sults revealed, the history-first scenario may be insufficient for massive Huang et al. (2015) suggested that the measurement of urban sus­
urbanization due to some weaknesses in the socioeconomic aspects. In tainability should be clear about whether weak or strong sustainability is
this sense, the history-based strategy would be better to adopt in the being gauged, and that at least one strong sustainability indicator should
be included in the assessment by combining indicator sets and composite

Table 5
Relationships between ES indicators and W-H/W-P revealed by multiple linear regressions on two timescales.
Indicator Short timescale (years 1–9) Long timescale (years 1–19)
Prob (F- Coef W-H (p < Coef W-P (p < Prob (F- Coef W-H (p < Coef W-P (p <
statistic) 0.005) 0.005) statistic) 0.005) 0.005)

Environmental dimension
Rainwater Storage (RS) 0.0000 0.0057 − 0.0037 0.0000 0.0159 − 0.0143
Mean Drainage Distance of urban plots (DDu) 0.0000 0.0196 − 0.0202 0.0000 0.0302 − 0.0575
Connectance Index of farmland patches (CIf) 0.0000 0.0073 0.0095 0.0000 0.0232 0.0454
Similarity-weighted Edge Density of waterbody patches 0.0000 0.0194 − 0.0119 0.0000 0.0280 − 0.0428
(SEDw)
Economic dimension
Mean NDVI of farmland plots (NDVIf) 0.0000 − 0.0052 − 0.0105 0.0000 − 0.0252 − 0.0290
Earthwork Volumes of new urban plots (EVu) 0.0001 − 0.0013 0.0014 0.0028 − 0.0040 − 0.0031
Mean Function Types in new urban plots (FTu) 0.0000 − 0.0503 − 0.0187 0.0000 − 0.0598 − 0.0144
Patch Cohesion Index of urban patches (PCIu) 0.0000 − 0.0173 0.0339 0.0000 − 0.0446 0.0605
Social dimension
Mean Distance to Motorways of urban plots (DMu) 0.0000 0.0206 − 0.0470 0.0000 0.0037 − 0.0834
Mean Distance to Roads of urban plots (DRu) 0.0000 − 0.0182 0.0273 0.0000 − 0.0718 0.0251
Interspersion & Juxtaposition Index between urban and 0.0000 − 0.0493 − 0.0861 0.0000 − 0.0791 − 0.0524
other patches (IJIu)

Notes. For all indicators, positive coefficients indicate positive impacts, and vice versa. The coefficient values should be interpreted in relative terms, with larger
absolute values suggesting greater impacts, allowing for horizontal and vertical comparisons.

12
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Fig. 8. Trend surface analysis of the impacts of different urbanization strategies (different combinations of W-H and W-P values) on the landscape sustainability
perceived by different stakeholders (indicated by the multi-year average stakeholder-weighted MESLI) on two timescales. The same colors correspond to the same
MESLI values in all subplots.

indices. This study used stakeholder-weighted MESLI that aggregated 11 their evolution. In discussing planning options and trade-offs, agent-
ES indicators to measure the sustainability of the urbanization process. based models that reflect local-level reactions and evolution paths to the
Among the different stakeholders’ weighting schemes, the total weights objectives can shed light on the emergent and unintended consequences
for environmental, economic, and social dimensions were 0.41, 0.35, of macro-level policies (Brown et al., 2013; Verburg et al., 2016).
and 0.24 assigned by villagers, 0.27, 0.35, and 0.37 assigned by urban Moreover, the ABM approach can provide practical insights into
residents, and 0.46, 0.29, and 0.25 assigned by experts. Villagers and cultural landscapes’ sustainable transformation through a combination
experts assigned higher weights to the environmental dimension, which of qualitative and quantitative methods. In cultural landscape research,
indicates that in the urbanization process of the cultural landscape, any qualitative methods such as field observations and interviews on land­
loss of natural capital has to be complemented by significantly more scape morphology and local activities and demands have played an
human-made capital. In this regard, the villager- and expert-weighted essential role (Di Fazio & Modica, 2018; Scott, 2011). Meanwhile,
MESLI (especially the expert-weighted MESLI with more balanced simulation modeling and statistical analysis enable experimental
weights for different environmental indicators) can be considered as research to explore and assess future outcomes in response to alternative
measuring strong sustainability of urbanization, while the urban interventions (Van Berkel & Verburg, 2012). Integrating human be­
resident-weighted MESLI measured weak sustainability. As shown in haviors and exploratory scenarios, ABM can bridge the two methods,
Fig. 10, the willingness- and planning-first scenarios saw a dramatic making research more transferable to planning and policymaking
decline in the environmental services during the second decade, despite (Rouleau & Zupko, 2019).
high overall sustainability levels. In other words, the high overall per­ In response to the notion that sustainability is essentially a process
formance was contributed by the growth of the economic and social rather than a state (Haines-Young, 2000; Wu, 2013), we used the
capitals during the urbanization process at the cost of environmental time-series data generated by ABM to evaluate the temporal pattern of
capital. This indicates weak sustainability of urbanization, which is the performance rather than that at a single or a few time nodes.
essentially unsustainable from the perspective of long-term develop­ Including temporal patterns in assessment could have a substantial
ment (Wilson & Wu, 2017). In light of this, apart from respecting lay impact on the results. For example, while the current planning may
stakeholders’ willingness, the results based on experts’ weighting attain acceptable outcomes in the end, we found its performance during
schemes, which support the merit of the historical landscape pattern, the whole urbanization process unsatisfactory. Furthermore, the tem­
should also be emphasized in decision-making. poral patterns allow decision-making to adapt to specific urbanization
stages on finer timescales, which may facilitate continuous learning in
5.5. ABM approach and temporal pattern analysis the future (Bond et al., 2012).

Many models have been used in prior research to discuss strategy 6. Conclusion
options and trade-offs, such as the System Dynamics model and the
Cellular Automata model (Lauf et al., 2012; Liu et al., 2017). However, As rapid and massive urbanization has become an inevitable reality
these models can hardly reflect local-level human decisions and feed­ for many cultural landscapes, it is critical to understand how potential
backs in response to changes. By contrast, the structure of ABM mirrors urbanization strategies would affect the sustainability of these land­
that of complex adaptive systems in the real world, which are under­ scapes. By modeling different adoption levels and trade-offs of three
pinned by individuals’ interactions and feedbacks with each other, their typical strategies and testing their outcomes over time, we found that a
environments, and macro-level factors and changes. This structure sustainable urbanization process in Nansha can be achieved by allowing
would be essential for modeling cultural landscapes, because local-level villagers and migrants to make their own decisions with minimal
human decisions and human–natural interactions play central roles in necessary supervision. While history- or planning-based strategies may

13
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Fig. 9. Temporal patterns of weighted MESLI (integration of 11 normalized ES indicators) were used to assess the landscape sustainability of the five focused
scenarios from stakeholders’ perspectives. The curves in the middle row show the changing processes of weighted MESLI during the simulation period. The dis­
tributions of the time-series data are compared with boxplots on 9-year (top row) and 19-year (bottom row) timescales. In all graphs, higher values indicate increased
sustainability, which can only be interpreted in relative terms.

deliver acceptable results in certain aspects and over certain periods, transformation. In further research, more place-specific case studies,
they would be less effective in the whole urbanization process due to applying and advancing exploratory modeling and temporal pattern
their limiting local adaptations and hence many context-specific ESs. assessment, are needed to shed more light on the unprecedented chal­
However, historical pattern-based strategy produced better environ­ lenges that contemporary landscapes face. New forms of public
mental performance during the urbanization process, which can engagement and feedback are also required to address the inherent
contribute to a stronger sustainability. Combining these results, we uncertainty in modeling.
suggest that sustainable rural development should place lay stake­
holders’ willingness first; landscape pattern preservation and land-use Declaration of Competing Interest
planning are necessary, but they must consider the place-specific pro­
cesses behind spatial patterns based on communications between lay The authors declare that they have no known competing financial
and expert stakeholders. interests or personal relationships that could have appeared to influence
Exploring and assessing scenarios over time with ABM is a practical the work reported in this paper.
method in decision-making for a cultural landscape’s sustainable

14
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Fig. 10. Stack plots of the 11 normalized ES indicators in the willingness-first, history-first, and planning-first scenarios show the contributions of different ESs to the
overall sustainability of urbanization during the simulation period. Environmental indicators are represented in green, while other indicators are in red.

Data availability study of China’s land urbanization from 2000 to 2015. Land Use Policy, 108. https://
doi.org/10.1016/j.landusepol.2021.105536
Chen, C., Xia, Y., & Yang, X. (2020). Resilience strategies to impacts of sea level rise on
The authors have shared the link to the data in the supplementary the agricultural areas in Nansha District of Guangzhou. Landscape Architecture
materials. Frontier, 8(03), 10–25. https://doi.org/10.15302/J-LAF-1-020027
Cheng, H., Liu, Y., He, S., & Shaw, D. (2017). From development zones to edge urban
areas in China: A case study of Nansha, Guangzhou City. Cities, 71, 110–122. https://
Acknowledgments doi.org/10.1016/j.cities.2017.07.015
Daly, H., Jacobs, M., & Skolimowski, H. (2007). Discussion of Beckerman’s critique of
This research is funded by the National Natural Science Foundation sustainable developemnt. Environmental Values, 4(1), 49–70. https://doi.org/
10.3197/096327195776679583
of China (52108051) and Natural Science Foundation of Guangdong Di Fazio, S., & Modica, G. (2018). Historic rural landscapes: Sustainable planning
Province, China (2021A1515010924; 2018A0303130017). We would strategies and action criteria. The Italian experience in the Global and European
also like to thank the editor and the anonymous reviewers for their Context. Sustainability, 10(11), 3834. https://doi.org/10.3390/su10113834
Dolley, J., Marshall, F., Butcher, B., Reffin, J., Robinson, J. A., Eray, B., & Quadrianto, N.
constructive comments on early drafts of this paper. (2020). Analysing trade-offs and synergies between SDGs for urban development,
food security and poverty alleviation in rapidly changing peri-urban areas: A tool to
Supplementary data support inclusive urban planning. Sustainability Science, 15(6), 1601–1619. https://
doi.org/10.1007/s11625-020-00802-0
Ekins, P., Simon, S., Deutsch, L., Folke, C., & de Groot, R. (2003). A framework for the
Supplementary data to this article can be found online at https://doi. practical application of the concepts of critical natural capital and strong
org/10.1016/j.landurbplan.2022.104555. sustainability. Ecological Economics, 44(2–3), 165–185. https://doi.org/10.1016/
S0921-8009(02)00272-0
Fischer, J., Hartel, T., & Kuemmerle, T. (2012). Conservation policy in traditional
References farming landscapes. Conservation letters, 5(3), 167–175. https://doi.org/10.1111/
j.1755-263X.2012.00227.x
Agnoletti, M. (2014). Rural landscape, nature conservation and culture: Some notes on Gong, J., Jiang, C., Chen, W., Chen, X., & Liu, Y. (2018). Spatiotemporal dynamics in the
research trends and management approaches from a (southern) European cultivated and built-up land of Guangzhou: Insights from zoning. Habitat
perspective. Landscape and Urban Planning, 126, 66–73. https://doi.org/10.1016/j. International, 82, 104–112. https://doi.org/10.1016/j.habitatint.2018.10.004
landurbplan.2014.02.012 Guo, W., Hou, X., & Cui, Z. (2021). Research on traditional landscape construction of
Antrop, M. (2005). Why landscapes of the past are important for the future. Landscape Ningbo Sun Lake and Moon Lake under the perspective of polder landscape.
and urban planning, 70(1–2), 21–34. https://doi.org/10.1016/j. Landscape Architecture, 28(05), 120–125. https://doi.org/10.14085/j.
landurbplan.2003.10.002 fjyl.2021.05.0120.06
Ao, M., Wang, C., Xie, R., Zhang, X., Hu, J., Du, Y., … Kuang, C. (2015). Monitoring the Haines-Young, R. (2000). Sustainable development and sustainable landscapes: defining
land subsidence with persistent scatterer interferometry in Nansha District, a new paradigm for landscape ecology. Fennia, 178(1), 7–14.
Guangdong, China. Natural Hazards, 75(3), 2947–2964. https://doi.org/10.1007/ Hasan, S. S., Zhen, L., Miah, M. G., Ahamed, T., & Samie, A. (2020). Impact of land use
s11069-014-1471-2 change on ecosystem services: A review. Environmental Development, 34. https://doi.
Benedetti, R., & Rossini, P. (1993). On the use of NDVI profiles as a tool for agricultural org/10.1016/j.envdev.2020.100527
statistics: The case study of wheat yield estimate and forecast in Emilia Romagna. Hassenforder, E., Pittock, J., Barreteau, O., Daniell, K. A., & Ferrand, N. (2016). The
Remote Sensing of Environment, 45(3), 311–326. https://doi.org/10.1016/0034-4257 MEPPP Framework: A Framework for Monitoring and Evaluating Participatory
(93)90113-C Planning Processes. Environmental Management, 57(1), 79–96. https://doi.org/
Bond, A., Morrison-Saunders, A., & Pope, J. (2012). Sustainability assessment: The state 10.1007/s00267-015-0599-5
of the art. Impact Assessment and Project Appraisal, 30(1), 53–62. https://doi.org/ He, S., Yu, S., Li, G., & Zhang, J. (2020). Exploring the influence of urban form on land-
10.1080/14615517.2012.661974 use efficiency from a spatiotemporal heterogeneity perspective: Evidence from 336
Braat, L. C., & De Groot, R. (2012). The ecosystem services agenda: Bridging the worlds Chinese cities. Land Use Policy, 95, Article 104576. https://doi.org/10.1016/j.
of natural science and economics, conservation and development, and public and landusepol.2020.104576
private policy. Ecosystem services, 1(1), 4–15. https://doi.org/10.1016/j. Hölting, L., Komossa, F., Filyushkina, A., Gastinger, M. M., Verburg, P. H., Beckmann, M.,
ecoser.2012.07.011 … Cord, A. F. (2020). Including stakeholders’ perspectives on ecosystem services in
Brown, D. G., Verburg, P. H., Pontius, R. G., & Lange, M. D. (2013). Opportunities to multifunctionality assessments. Ecosystems and People, 16(1), 354–368. https://doi.
improve impact, integration, and evaluation of land change models. Current Opinion org/10.1080/26395916.2020.1833986
in Environmental Sustainability. https://doi.org/10.1016/j.cosust.2013.07.012 Huang, L., Wu, J., & Yan, L. (2015). Defining and measuring urban sustainability: a
Buschle, C., Reiter, H., & Bethmann, A. (2021). The qualitative pretest interview for review of indicators. Landscape Ecology, 30(7), 1175–1193. https://doi.org/
questionnaire development: Outline of programme and practice. Quality and 10.1007/s10980-015-0208-2
Quantity. https://doi.org/10.1007/s11135-021-01156-0 Krajnc, D., & Glavič, P. (2005). A model for integrated assessment of sustainable
Butler, J. R. A., Wise, R. M., Skewes, T. D., Bohensky, E. L., Peterson, N., Suadnya, W., … development. Resources, Conservation and Recycling, 43(2), 189–208. https://doi.
Rochester, W. (2015). Integrating Top-Down and Bottom-Up Adaptation Planning to org/10.1016/S0921-3449(04)00120-X
Build Adaptive Capacity: A Structured Learning Approach. Coastal Management, 43 Kučera, P., Novák, J., Löw, J., & Dohnal, T. (2015). Historical landscape structures and
(4), 346–364. https://doi.org/10.1080/08920753.2015.1046802 their importance for landscape character. Acta Universitatis Agriculturae et
Cao, Y., Kong, L., Zhang, L., & Ouyang, Z. (2021). The balance between economic Silviculturae Mendelianae Brunensis, 63(1), 49–57. https://doi.org/10.11118/
development and ecosystem service value in the process of land urbanization: A case actaun201563010049

15
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Lauf, S., Haase, D., Hostert, P., Lakes, T., & Kleinschmit, B. (2012). Uncovering land-use Scott, A. (2011). Beyond the conventional: Meeting the challenges of landscape
dynamics driven by human decision-making - A combined model approach using governance within the European Landscape Convention? Journal of Environmental
cellular automata and system dynamics. Environmental Modelling and Software, Management, 92(10), 2754–2762. https://doi.org/10.1016/j.jenvman.2011.06.017
27–28, 71–82. https://doi.org/10.1016/j.envsoft.2011.09.005 Selman, P. (2007). Landscape and sustainability at the national and regional scales. In
Li, H., & Wu, J. (2004). Use and misuse of landscape indices. Landscape Ecology, 19(4), J. F. Benson, & M. Roe (Eds.), Landscape and Sustainability (pp. 104–117). Routledge.
389–399. https://doi.org/10.1023/B:LAND.0000030441.15628.d6 Selman, P. (2012). Landscapes as integrating frameworks for human, environmental and
Li, Q. (2020). China’s Development Under a Differential Urbanization Model. Springer policy processes. In T. Plieninger, & C. Bieling (Eds.), Resilience and the cultural
Singapore. landscape: Understanding and Managing Change In Human-Shaped Environments (pp.
Liu, J., Zhang, M., & Nikita, N. (2021). Agent-based design research to explore the 27–48). Cambridge: Cambridge University Press. https://doi.org/10.1017/
effectiveness of bottom-up organizational design in shaping sustainable vernacular CBO9781139107778.
landscapes: A case in Hailar, China. Landscape and Urban Planning, 205, 103961. Shen, M., & Shen, J. (2019). State-led commodification of rural China and the sustainable
https://doi.org/10.1016/j.landurbplan.2020.103961 provision of public goods in question: A case study of Tangjiajia, Nanjing. Journal of
Liu, S., Yu, Q., & Wei, C. (2019). Spatial-temporal dynamic analysis of land use and Rural Studies.. https://doi.org/10.1016/j.jrurstud.2019.03.004
landscape pattern in Guangzhou, China: Exploring the driving forces from an urban Singh, R., Reed, P. M., & Keller, K. (2015). Many-objective robust decision making for
sustainability perspective. Sustainability, 11(23), 6675. https://doi.org/10.3390/ managing an ecosystem with a deeply uncertain threshold response. Ecology and
su11236675 Society, 20(3). https://doi.org/10.5751/ES-07687-200312
Liu, X., Liang, X., Li, X., Xu, X., Ou, J., Chen, Y., … Pei, F. (2017). A future land use Stremke, S., & Schöbel, S. (2019). Research through design for energy transition: Two
simulation model (FLUS) for simulating multiple land use scenarios by coupling case studies in Germany and The Netherlands. Smart and Sustainable Built
human and natural effects. Landscape and Urban Planning, 168, 94–116. https://doi. Environment, 8(1), 16–33. https://doi.org/10.1108/SASBE-02-2018-0010
org/10.1016/j.landurbplan.2017.09.019 Stürck, J., & Verburg, P. H. (2017). Multifunctionality at what scale? A landscape
Lu, X., Zhang, Y., Lin, C., & Wu, F. (2021). Analysis and comprehensive evaluation of multifunctionality assessment for the European Union under conditions of land use
sustainable land use in China: Based on sustainable development goals framework. change. Landscape Ecology, 32(3), 481–500. https://doi.org/10.1007/s10980-016-
Journal of Cleaner Production, 310. https://doi.org/10.1016/j.jclepro.2021.127205 0459-6
Ma, P., Wang, W., Zhang, B., Wang, J., Shi, G., Huang, G., … Lin, H. (2019). Remotely Szetey, K., Moallemi, E. A., Ashton, E., Butcher, M., Sprunt, B., & Bryan, B. A. (2021).
sensing large- and small-scale ground subsidence: A case study of the Guangdong- Participatory planning for local sustainability guided by the Sustainable
Hong Kong–Macao Greater Bay Area of China. Remote Sensing of Environment, 232, Development Goals. Ecology and Society, 26(3). https://doi.org/10.5751/es-12566-
Article 111282. https://doi.org/10.1016/j.rse.2019.111282 260316
Maher, R., Maher, M., Mann, S., & McAlpine, C. A. (2018). Integrating design thinking Tang, B. (2015). Not Rural but Not Urban: Community Governance in China’s Urban
with sustainability science: A Research through Design approach. Sustainability Villages. China Quarterly, 223, 724–744. https://doi.org/10.1017/
Science, 13(6), 1565–1587. https://doi.org/10.1007/s11625-018-0618-6 S0305741015000843
Manning, P., van der Plas, F., Soliveres, S., Allan, E., Maestre, F. T., Mace, G., … Tang, B., Zhan, X., & Zhang, J. (2019). The Agricultural Heritage Value of the Polder
Fischer, M. (2018). Redefining ecosystem multifunctionality. Nature Ecology & System in Gaochun, Nanjing. Built Heritage, 3(1), 47–58. https://doi.org/10.1186/
Evolution, 2(3), 427–436. https://doi.org/10.1038/s41559-017-0461-7 bf03545735
Marshall, F., Dolley, J., Bisht, R., Priya, R., Waldman, L., Amerasinghe, P., & Tempesta, T., & Vecchiato, D. (2015). Testing the difference between experts’ and lay
Randhawa, P. (2018). Ecosystem services and poverty alleviation in urbanising people’s landscape preferences. Aestimum, 66(1), 1–41. https://doi.org/10.13128/
contexts. In Ecosystem Services and Poverty Alleviation: Trade-Offs and Governance (pp. Aestimum-16481
111–125). Taylor and Francis. https://doi.org/10.4324/9780429507090. Termorshuizen, J. W., & Opdam, P. (2009). Landscape services as a bridge between
MEA. (2005). Ecosystems and Human Well-Being: Current State and Trends. Washington, landscape ecology and sustainable development. Landscape Ecology, 24(8),
DC: Island Press. 1037–1052. https://doi.org/10.1007/s10980-008-9314-8
Moallemi, E. A., Kwakkel, J., de Haan, F. J., & Bryan, B. A. (2020). Exploratory modeling Tieskens, K. F., Shaw, B. J., Haer, T., Schulp, C. J. E., & Verburg, P. H. (2017). Cultural
for analyzing coupled human-natural systems under uncertainty. Global landscapes of the future: Using agent-based modeling to discuss and develop the use
Environmental Change, 65, Article 102186. https://doi.org/10.1016/j. and management of the cultural landscape of South West Devon. Landscape Ecology,
gloenvcha.2020.102186 32(11), 2113–2132. https://doi.org/10.1007/s10980-017-0502-2
Moallemi, E. A., & Malekpour, S. (2018). A participatory exploratory modelling approach United Nations. (2015). Transforming our world: The 2030 agenda for sustainable
for long-term planning in energy transitions. Energy Research and Social Science, 35, development. United Nations General Assembly. Retrieved from https://
205–216. https://doi.org/10.1016/j.erss.2017.10.022 sustainabledevelopment.un.org/post2015/transformingourworld/publication.
Nardo, M., Saisana, M., Saltelli, A., & Tarantola, S. (2005). Tools for Composite Indicators Valbuena, D., Bregt, A. K., McAlpine, C., Verburg, P. H., & Seabrook, L. (2010). An agent-
Building. Retrieved from https://publications.jrc.ec.europa.eu/repository/handle/ based approach to explore the effect of voluntary mechanisms on land use change: A
JRC31473. case in rural Queensland, Australia. Journal of Environmental Management, 91(12),
Nassauer, J. I., & Opdam, P. (2008). Design in science: Extending the landscape ecology 2615–2625. https://doi.org/10.1016/j.jenvman.2010.07.041
paradigm. Landscape Ecology, 23(6), 633–644. https://doi.org/10.1007/s10980-008- Vallés-Planells, M., Galiana, F., & Van Eetvelde, V. (2014). A classification of landscape
9226-7 services to support local landscape planning. Ecology and Society, 19(1). https://doi.
Nijhuis, S., Xiong, L., & Cannatella, D. (2021). Towards a Landscape-based Regional org/10.5751/ES-06251-190144
Design Approach for Adaptive Transformation in Urbanizing Deltas. Research in Van Berkel, D. B., & Verburg, P. H. (2012). Combining exploratory scenarios and
Urbanism Series, 6, 55–80. https://doi.org/10.7480/rius.6.94 participatory backcasting: Using an agent-based model in participatory policy design
NRC. (1999). Our Common Journey: A Transition Toward Sustainability. Washington, DC: for a multi-functional landscape. Landscape Ecology, 27(5), 641–658. https://doi.
National Academy Press. org/10.1007/s10980-012-9730-7
Panyu Chorography Editorial Board. (1995). The Chorography of Panyu.. Guangdong Verburg, P. H. (2006). Simulating feedbacks in land use and land cover change models.
People’s Publishing House. Landscape Ecology, 21(8), 1171–1183. https://doi.org/10.1007/s10980-006-0029-4
Plieninger, T., Van der Horst, D., Schleyer, C., & Bieling, C. (2014). Sustaining ecosystem Verburg, P. H., Dearing, J. A., Dyke, J. G., van der Leeuw, S., Seitzinger, S., Steffen, W., &
services in cultural landscapes. Ecology and Society, 19(2), 59. https://doi.org/ Syvitski, J. (2016). Methods and approaches to modelling the Anthropocene. Global
10.5751/ES-06159-190259 Environmental Change, 39, 328–340. https://doi.org/10.1016/j.
Raman, R., & Roy, U. K. (2019). Taxonomy of urban mixed land use planning. Land Use gloenvcha.2015.08.007
Policy, 88, Article 104102. https://doi.org/10.1016/j.landusepol.2019.104102 Wang, X., & Lin, Q. (2018). Traditional Chinese mountain-water-field-city system from
Rescia, A. J., Pérez-Corona, M. E., Arribas-Ureña, P., & Dover, J. W. (2010). Cultural the perspective of territorial landscape. Landscape Architecture, 25(09), 10–20.
landscapes as complex adaptive systems: The cases of northern Spain and Northern https://doi.org/10.14085/j.fjyl.2018.09.0010.11
Argentina. Resilience and the cultural landscape: Understanding and managing change in Wang, Z., Ren, P., & Ren, H. (2015). Research of contemporary urban complex from the
human-shaped environments (pp. 126–145). Cambridge University Press. angle of industry- city integration. Open Cybernetics and Systemics Journal, 9(1),
Rodríguez-Loinaz, G., Alday, J. G., & Onaindia, M. (2015). Multiple ecosystem services 2820–2825. https://doi.org/10.2174/1874110X01509012820
landscape index: A tool for multifunctional landscapes conservation. Journal of WCED. (1987). Our Common Future. New York: Oxford University Press.
Environmental Management, 147, 152–163. https://doi.org/10.1016/j. Wiens, J. A. (2013). Is landscape sustainability a useful concept in a changing world?
jenvman.2014.09.001 Landscape Ecology, 28(6), 1047–1052. https://doi.org/10.1007/s10980-012-9801-9
Rouleau, M., & Zupko, R. (2019). Agent-Based Modeling for bioenergy sustainability Wilson, M. C., & Wu, J. (2017). The problems of weak sustainability and associated
assessment. Landscape and Urban Planning, 188, 54–63. https://doi.org/10.1016/j. indicators. International Journal of Sustainable Development and World Ecology, 24(1),
landurbplan.2019.04.019 44–51. https://doi.org/10.1080/13504509.2015.1136360
Rovai, M., Andreoli, M., Gorelli, S., & Jussila, H. (2016). A DSS model for the governance Wu, J. (2010). Landscape of culture and culture of landscape: Does landscape ecology
of sustainable rural landscape: A first application to the cultural landscape of Orcia need culture? Landscape Ecology, 25(8), 1147–1150. https://doi.org/10.1007/
Valley (Tuscany, Italy). Land Use Policy, 56, 217–237. https://doi.org/10.1016/j. s10980-010-9524-8
landusepol.2016.04.038 Wu, J. (2013). Landscape sustainability science: Ecosystem services and human well-
Saaty, T. L. (1990). How to make a decision: The analytic hierarchy process. European being in changing landscapes. Landscape Ecology, 28(6), 999–1023. https://doi.org/
Journal of Operational Research, 48(1), 9–26. https://doi.org/10.1016/0377-2217 10.1007/s10980-013-9894-9
(90)90057-I Wu, Y., Shan, L., Zheng, S., Lai, S.-K., & Xia, B. (2020). Regional planning reconfiguration
Schulp, C. J. E., Levers, C., Kuemmerle, T., Tieskens, K. F., & Verburg, P. H. (2019). in China based on inclusiveness: Examining development and control orientation.
Mapping and modelling past and future land use change in Europe’s cultural Journal of Urban Planning and Development, 146(3), 05020012. https://doi.org/
landscapes. Land Use Policy, 80, 332–344. https://doi.org/10.1016/j. 10.1061/(asce)up.1943-5444.0000578
landusepol.2018.04.030

16
J. Liu et al. Landscape and Urban Planning 228 (2022) 104555

Xie, H., Zhang, Y., Zeng, X., & He, Y. (2020). Sustainable land use and management Environmental Management, 57(6), 1304–1318. https://doi.org/10.1007/s00267-
research: A scientometric review. Landscape Ecology, 35(11), 2381–2411. https:// 016-0683-5
doi.org/10.1007/s10980-020-01002-y Wong, S. W., Dai, Y., Tang, B. Sin, & Liu, J. (2021). A new model of village urbanization?
Xu, Q., Yang, R., Zhuang, D., & Lu, Z. (2021). Spatial gradient differences of ecosystem Coordinative governance of state-village relations in Guangzhou City, China. Land
services supply and demand in the Pearl River Delta region. Journal of Cleaner Use Policy, 109, Article 105500. https://doi.org/10.1016/j.landusepol.2021.105500
Production, 279. https://doi.org/10.1016/j.jclepro.2020.123849 Yuan, R. (2017). Study on the Development of Rural Land Circulation in Nansha District,
Yang, H., Qiu, L., & Fu, X. (2021). Toward cultural heritage sustainability through Guangzhou (Master’s thesis, Zhongkai University of Agriculture and Engineering).
participatory planning based on investigation of the value perceptions and Zhang, Y., Hu, Y., & Zhuang, D. (2020). A highly integrated, expansible, and
preservation attitudes: Qing Mu Chuan, China. Sustainability, 13(3), 1–14. https:// comprehensive analytical framework for urban ecological land: A case study in
doi.org/10.3390/su13031171 Guangzhou, China. Journal of Cleaner Production, 268, Article 122360. https://doi.
Yang, Z., Wang, Y., & Liu, Z. (2019). Improving socially inclusive development in fast org/10.1016/j.jclepro.2020.122360
urbanized area: Investigate livelihoods of immigrants and non-immigrants in Nansha Zhou, Y., & Zhou, J. (2015). Rural planning compilation practices in urban planning
Special Economic Zone in China. Habitat International, 86, 10–18. https://doi.org/ areas: A case study of Luwan village in Nansha district of Guangzhou city. City
10.1016/j.habitatint.2019.02.005 Planning Review, 39(08), 92–100. https://doi.org/10.11819/cpr20150814a
Yu, H., Verburg, P. H., Liu, L., & Eitelberg, D. A. (2016). Spatial analysis of cultural
heritage landscapes in rural China: Land use change and its risks for conservation.

17

You might also like