Developing China’s Ecological Redline Policy using ecosystem services assessments for land use planning
NATURE COMMUNICATIONS |
Developing China's Ecological Redline Policy using ecosystem services assessments for land use planning
Christina P. Wong
Alice C. Hughes
Ecosystems services (ES) assessment is a significant scientific topic recognized for its potential to address sustainability issues. However, there is an absence of science-policy frameworks in land use planning that lead to the ES science being used in policy. China's Ecological Redline Policy (ERP) is one of the first national policies utilizing multiple ES, but there is no standardized approach for working across the science-policy interface. We propose a transdisciplinary framework to determine ecological redline areas (ERAs) in Shanghai using: ES, biodiversity and ecologically fragile hotspots, landscape structure, and stakeholder opinions. We determine the five criteria to identify ERAs for Shanghai using multi-temporal, high resolution images (0.5 m) and biophysical models. We examine ERP effectiveness by comparing land use scenarios for 2040. Compared to alternative land uses, ES increase significantly under the ERP. The inclusion of ES in spatial planning led stakeholders to increase terrestrial habitat protection by 174% in Shanghai. Our analysis suggests that strategic planning for ES could reduce tradeoffs between environmental quality and development.
Eticing integrated approaches to address the serious policy
cosystem services (ES) assessments offer a means of
pracchallenge of incorporating environmental issues into
development decisions1,2. To promote wise decision-making,
policymakers need information from scientists about how
different land use decisions may affect the condition of ecosystems
and the flow of ES3. A major challenge is developing capacities on
ES assessments for spatial planning, which is difficult because ES
requires new interdisciplinary methodologies to assess multiple
environmental and social concerns4.
Despite increasing political interest in ES, the use of ES
information remains quite limited5–8. National ES assessments
have been conducted in China9, United Kingdom10, and
Mexico11. Scientists have conducted regional ES assessments for
spatial planning in Pampas (Argentina)12, Hawaii (USA)13,
Boredeux (France)14, Vancouver Island (Canada)15, Tampere
(Finland)16, and coastal areas in Belize17. A number of these
studies use the Integrated Valuation of Ecosystem Services and
Tradeoffs models (InVEST) to assess ES, but few demonstrate
an application by decision-makers for determiningland use
planning targets8,18. Albert et al.6 state the research priority is
developing transdisciplinary case studies of application to
advance ES tools and methods for real-world planning.
One institutional obstacle is the lack of ES standards (i.e.,
assessment protocols and targets)19. Many researchers suggest
scientists focus their work on municipalities to develop the
required case studies for ES standards since cities are critical
administrative units for land use planning6,20,21. However
Hansen et al.20 show ES assessments are only being used as a
supporting concept in urban planning. There are no published
examples of a municipality implementing spatial plans based on
ES assessment. However, a unique opportunity is emerging in
China where municipalities want ES information to meet
sustainability goals. China is the first major economy to formulate a
national policy, mandating governments to establish ES
assessments in land use planning known as the Ecological Redline
China is experiencing an environmental crisis, and President
Xi realizes China must transform its development approach from
“grow first, clean up later” to the “ecological civilization” where
development respects ecological carrying capacities24,25. ERP
seeks to sustain critical ES for social welfare using coordinated
planning at a national scale. The policy instrument is establishing
key ecological function zones (EFZs) using ecological redline
areas (ERAs)26. EFZs were selected to sustain five national ES:
biodiversity conservation, water resources conservation, flood
mitigation, soil conservation, and sandstorm prevention27. ERAs
represent an attempt at establishing ES assessment standards in
land use planning, defined as the “minimum ecological area
needed to guarantee and maintain ecological safety and
functionality, and biological diversity for national security”23,24. All
municipalities and provinces must create ERAs where ES
information should inform selection. The problem is we lack
standardized methods, which is impacting the consistency, credibility,
and usability of ES assessments22.
Conceptual frameworks1,2,5,28–33 and spatial mapping34–36
have enabled progress on ES science; however, there is a
fundamental lack of science–policy frameworks, explaining
methodological standards for application in policy7,37. We create a
science–policy framework (Fig. 1) that builds upon core elements
of other ES frameworks (i.e., MA1, IPBES29, Ecosystem Services
Cascade38, and Natural Capital Project39) but details both the
specific institutional and ecological components (e.g., types of
information, indicators, and methodological steps) for public
policy. Here we present the transdisciplinary framework and
methodology for the ERP, proposing five indicators to
standardize ERA designation processes: ES hotspots; biodiversity
hotspots; ecologically fragile hotspots (vulnerable to stressors);
landscape structure (composition and configuration); stakeholder
opinions (Fig. 2). We employ our approach to inform China’s
governance process on the selection of ERAs for Shanghai
Municipality. We select Shanghai as our case study because it is a
priority ERP region, possessing global significance as one of the
world’s most urbanized cities. The three objectives of the study
are to: (
) present a science–policy framework for ES assessments
applicable to ERP; (
) explain methods and ES results for
determining ERAs to illustrate possible ES assessment standards;
) evaluate the effectiveness of ERP for reducing tradeoffs on
We implement our framework to determine ERAs for
policymakers in Shanghai. We first use policy goals and public
preferences to select the ecosystem services. We create land use and
land cover (LULC) maps to estimate ES and biodiversity, and
expert opinion to determine highly vulnerable ecosystem areas to
major stressors in Shanghai (known as ecologically fragile areas).
We use these three indicators to determine optimal ERAs to
inform a stakeholder negotiation process to determine an
implementable ERA for Shanghai. To evaluate the effectiveness of
the selected ERAs we compare the ERP to alternative LULC
scenarios. The objective is to help policymakers select a land use
plan for 2040 to inform Shanghai’s Urban Plan (2016–2040),
which outlines Shanghai’s land use decisions. We compare ES
outcomes under four scenarios codeveloped by policymakers and
) baseline, current ERAs for 2014; (
for 2040 (no environmental constraints); (
) future ERP for 2040
(expansion of ERAs by 46%); and (
) planning for 2040 (existing
ecological protection measures, except ERP). We use Markov and
CLUE_S models to estimate changes in LULCs under different
urbanization policies. Lastly, we compare ES synergies and
tradeoffs among scenarios at local (i.e., district level) and regional
(i.e., Shanghai Municipality) scales to assess whether ERP can
improve ES outcomes.
From our analysis, we determine a current ecological redline
target of 1098 km2 at the municipal scale. ERAs cover 16% of
Shanghai’s total land area, representing a 174% increase in
terrestrial protected area. ES criteria expand ecosystem protection by
142% (681 km2). Furthermore, we find ERP significantly increases
ES flows compared to other land use scenarios. If properly
implemented ERP could potentially reduce the tradeoff between
urbanization and ecosystem protection in Shanghai.
Ecological redline for Shanghai. As the world’s third most
populous city, Shanghai embodies the sustainability challenges
confronting cities worldwide; Shanghai’s experience on
implementing comprehensive planning is critical to enhancing
international knowledge on urban sustainability. The novel
component of our approach is illustrating how scientists can use
institutional information to guide the biophysical and land use
modeling. Scientifically this requires understanding policies and
stakeholder preferences to determine the desired human benefits
to select appropriate indicators. First we select the desired ES for
Shanghai by working with policymakers to determine policy
goals, and conduct stakeholder surveys to assess public opinion.
The government and local stakeholders want to invest in
ecosystem protection/restoration to generate improvements in water
resources, water quality, coastal erosion control, and climate
change mitigation. The policy goals as described in Shanghai’s
Urban Plan (2016–2040)40 are: (
) continuous water supply from
freshwater sources (e.g., Qingcaosha Reservoir and Huangpu
) improve water quality of major rivers for drinking
(Salinization, soil erosion, and
(China’s key ecological function zones)
Stakeholder needs and expertise
and ecological condition)
Desired societal benefits
(Climate change mitigation, water supply,
water quality, erosion control, and biodiversity)
(Carbon sequestration, water conservation,
water purification, and soil retention)
(Shanghai ecological redline targets)
(Composition and configuration)
(Land use and land cover change)
water and recreation—47% of Shanghai’s freshwater ecosystems
are unsafe for any use (i.e., worse than Grade V)41; (
control of coastal erosion for land stabilization; and (
carbon sequestration to offset carbon emissions (reduce peak
carbon emissions by 15%) for climate change mitigation.
Stakeholders select five ES: carbon sequestration, water resources
conservation, water purification, soil retention, and biodiversity.
Next, we use aerial images (0.5 m) to quantify LULC for 2005
and 2014 in Shanghai Municipality to model land use patterns
and ES (Supplementary Fig. 1; Supplementary Table 1). We
categorize LULC into Shanghai’s five main LULC categories: (
) constructed land; (
) agriculture; (
) open water; and
) beach. Second, we use the InVEST 3.2.0 models13,39,42 to
estimate ES and biodiversity (Table 1). Ecological areas are
ranked where those providing the top 10% of each ES are ES
hotspots, and areas with greatest suitable habitat for biodiversity
are biodiversity hotspots. Third, we gather expert opinion to rank
ecologically fragile areas for key stressors: soil erosion, coastal
erosion, and salinization. Ecologically fragile hotspots are the
most vulnerable ecological areas to the main drivers of ecological
change in Shanghai selected by local experts (see Methods). We
sum hotspot areas to determine optimal ERAs to formulate a
planning map (resolution 50 m). Subsequently local governments
facilitate a stakeholder negotiation process using scientific maps
on optimal ERAs to guide discussions to build consensus on an
We determine 929 km2 of the land area in Shanghai is ES
hotspots, consisting of Shanghai’s key drinking water sources
(e.g., Huangpu River, Qingcaosha, Chenhang, and
Dongfengxisha) and major forest patches (e.g., Dongping Forest Park,
Haiwan Forest Park, and Gongqing Forst Park). Next, we
estimate 171 km2 of Shanghai’s land area is ecologically fragile
hotspots, representing the most threatened habitats in Shanghai,
mainly located on Chongming Island due to sea-level rise (e.g.,
Dongtan Wetland Park and Xisha Wetland Park). Furthermore,
we determine 643 km2 is biodiversity hotspots, which encompass
major wetland reserves, coastal marshes, bird sanctuaries, etc.
Subsequently we evaluate whether the 10% threshold for hotspots
represent the majority of ES production, vulnerable areas, and
biodiversity areas. We calculate ES hotspots represent 35% of
estimated carbon sequestration; 49% of estimated water
conservation; 33% of estimated water purification; and 42% of
estimated soil retention. Ecologically fragile hotspots represent
8% of expert selected vulnerable areas, and biodiversity hotspots
represent 15% of estimated suitable biodiversity habitat
(Supplementary Fig. 2). For spatial planning, we have to balance space
constraints with the greatest potential to improve environmental
quality for human welfare given the policy goals. Hotspot
1. Link national goals on ecological function zones and ecosystem services to local
conditions (e.g., ecology, threats and socioeconomic context)
Work with local stakeholders to select priority ecosystem services that reflect national and local
Work with local experts to identify: (
) ecosystem types, and (
) main drivers of ecological
change impacting provision of services and biodiversity.
2. Determine ecological redline target using stakeholder opinions and hotspots for critical
ecosystem services, ecologically fragile areas, and biodiversity.
Create spatially explicit results by measuring ecosystem services, ecologically fragile areas,
and suitable habitat for biodiversity conservation using field data, models, and land cover data.
Identify hotspots for these socioecological criteria (top 10% of areas) assess whether hotspots
cover essential areas for each criteria, adjust hotspot thresholds appropriately given local
conditions. Generate maps of potential ecological redline areas (sum of hotspots, considering
overlapping areas). Evaluate credibility using local knowledge.
Present potential ecological redline areas to stakeholders to guide negotiation process where
concerns and needs are evaluated to find agreement on ecological redline target.
3. Generate land use scenarios to evaluate synergies and tradeoffs for land use zoning.
Work with policymakers to determine future land use scenarios for the region, determine
longterm ecological redline target. Use land cover models to generate land cover maps for scenarios.
Use models to assess changes in ecosystem condition (composition and configuration) and
ecosystem services under different scenarios.
Assess potential synergies and tradeoffs for different scenarios to identify optimal land use plan.
Present results to policymakers to inform land use zoning.
Note: InVEST models were used to estimate the ecosystem services and biodiversity hotspots; delphi method used to estimate the ecologically fragile areas; survey was conducted to determine
stakeholder preferences for different ecosystem services.
indicators, in particular ES hotspots, seem to strike this balance
for expanding the consideration of multiple environmental
benefits in the spatial planning process in Shanghai. Lastly, we
combine these three spatially explicit layers (Fig. 3b–d)
subtracting 470 km2 of overlapping areas to produce the optimal ERAs,
totaling 1273 km2.
The optimal ERAs established through this analysis were
presented as a part of a stakeholder negotiation process led by the
Shanghai Municipal Government (SMG), Shanghai
Environmental Protection Bureau (SEPB), and Shanghai Municipal Planning
and Land Resources Administration (SPLRA). The stakeholder
groups are district and township governments, enterprises,
farmers, and residents that are likely to be impacted by the
ERAs. There are two types of consultations: (
scoping and (
) public comment. For the district workshops, the
aim is to obtain agreement on the location of ERAs at the district
level. The SEPB and SPLRA presented the spatial maps to district
governments and other key agencies then we discussed the
proposed ERA spatial distributions, ERA criteria, and our
methodology. Coproducing the ES information with the SEPB
was critical for legitimizing the ES analysis (see Supplementary
Methods for details on the stakeholder engagement;
Supplementary Table 2 defines the stakeholder groups).
Disagreements mainly concern the prevention of future development
activities in certain areas. District leaders refined the optimal
ERAs using local expertize to determine manageable ERAs (i.e.,
elimination of small patches). The SMG plans to financially
compensate impacted stakeholders to offset economic losses
associated with relocation and prohibition of development near
ERAs; currently the financial compensation program is
undergoing development. Preliminarily, the stakeholder negotiation
process removed 176 km2 of optimal ERAs (Fig. 3e). Scientifically,
we approve of this reduction because it removes mainly isolated
patches of forest and grassland; it represents the lowest estimated
impact on ES and biodiversity given stakeholder concerns. The
current selected ERAs encompass a total protected area of 1098
km2 (142% increase in protected areas from 2014), accounting for
16% of Shanghai’s total land area (Fig. 3f). Please note these are
not the final ERAs but our analysis is being used as part of the
process to form Shanghai’s final ERAs. The political process
will take time and several rounds of negotiation, which is beyond
the scope of our analysis . To date ERAs represent China’s
strictest environmental protection standard. According to China’s
Environmental Protection Law (2015), ERAs must incur no
ecological degradation and no reduction in acreage, the
government can only increase acreage overtime22–24. If properly
implemented ERAs for the first time would protect the majority
of Shanghai’s terrestrial ecosystems. Prior to ERP, the main
ecosystem types under protection were drinking water sources,
wetlands, and marine reserves typically determined using single
criteria (Fig. 3a). The ERP using ES criteria leads to a spatial
plan that for the first time coordinates ecological protection
across terrestrial ecosystems aimed at multiple social benefits.
From the stakeholder negotiation process, we determine
stakeholders came to agreement on ERAs using the ES assessment
because the central government is now prioritizing the
enforcement of ecosystem protection to sustain ES.
Future land use plan. ES assessment for spatial planning should
help clarify tradeoffs between development and environmental
protection to promote wise management that fosters synergies
and reduces tradeoffs1,2. We explore three alternative future
scenarios for 2040 (Fig. 4) derived from the Markov and CLUE_S
models, which were codeveloped with policymakers and urban
planners for the SMG and Shanghai Development and Reform
Commission. We compare the spatial configuration and ES
outcomes among the current ERAs (S1), future development
scenario (S2), future ERP scenario (S3), and future planning scenario
(S4) in 2040 (Table 2). Scenarios have different impacts on the
ecosystem composition and spatial distribution of the landscape
(Fig. 4; Supplementary Table 3). In S2 constructed area increases
by 15% compared to S1 mainly from the conversion of agriculture
and open water. S2 continues the conventional urbanization trend
in Shanghai. In S4 constructed area increases by 3% from S1,
however, in S3 constructed area decreases by 10% from S1. Forest
area increases under all future scenarios (S2–S4) when compared
to S1. This result is consistent with the LULC changes in recent
years in Shanghai due to increased afforestation efforts43. In S2
and S4 the increase in forest area is due to future afforestation
efforts, however, existing forest area decreases (Supplementary
Fig. 3). An increase in young secondary forests, however, may not
guarantee the same quality and/or quantity of ES as more mature
forests. Forests and agriculture retain most of their original
distributions in S3 where 92% of forests and 81% of farmlands are
unchanged. Forests in S2 and S4 retain 68% and 86% of their
original areas, while agriculture retains 71% and 75%, respectively
(Supplementary Fig. 4). The largest difference in ecosystem
composition between the future scenarios is the reduction in
beach area between S3 and S2 and S4. S3 is the only future
scenario that retains open water and beach areas at current levels
We also calculate connectivity index (CI) and fragmentation
index (FI) values to evaluate the degree of connectivity between
ecosystem types known to influence ecosystem functionality and
biodiversity; each scenario has different ecosystem configurations
compared to S1 (Supplementary Fig. 5). For S2 the FI value is
13.65 km−2 (16% decrease from S1), while the CI value is 90.16
(4% increase from S1). In comparison, FI for S3 is 11.27 km−2
(31% decrease from S1) and CI is 95.73 (11% increase from S1).
The increase in CI and decrease in FI values in S3 from S1 are due
to the establishment of ecological corridors to form riparian
buffers along Shanghai’s major rivers like Huangpu River and
Suzhou River. Our results suggest if properly enforced, the ERP is
the only urban plan that can maintain ecosystem composition by
protecting forest, open water, and beach areas while minimizing
agricultural losses (Fig. 5).
Ecosystem services. Shanghai wants to enhance ES levels
overtime, despite expected future population growth, but
performance differs among the scenarios. Overall, S3 has the
No implementation of Ecological Redline Policy (ERP), and no
new policies to constrain growth. Uncontrolled urbanization
with population size of 31 million (25% total growth rate)
and 5% GDP growth rate for 2040.
Expansion of ERAs by 501 km2 by: (
) planting vegetation
buffers along river banks and (
) transforming industrial
and agricultural areas to forests (afforestation). Condensed
urbanization with projected population size of 25 million
(4% total growth rate) and 5% GDP growth rate for 2040.
Implementation of existing ecological protection policies
outlined in Shanghai’s Urban Plans (1999–2020; 2016–2040),
excluding the ERP. Projected population size of 25 million
(4% total growth rate) and 5% GDP growth rate for 2040.
) Constructed land = 42%, (
) agriculture = 35%,
) forests = 12%, (
) open water = 9%, and
) beach = 2%
) Constructed land = +6%, (
) agriculture = −7%,
) forest = +3%, (
) open water = −1%, and
) beach = −1%
) Constructed land = −4%, (
) agriculture = −4%,
) forest = +9%, (
) open water = 0%, and
) beach = 0%
) Constructed land = +1%, (
) agriculture = −4%,
) forest = +5%, (
) open water = 0%, and
) beach = −1%
highest estimated ES production compared to all scenarios. We
estimate the future ERP will incur fewer ES tradeoffs with
development compared to other land use policies (Fig. 6). We
find a strong positive correlation among all four ES for each
scenario (Pearson correlation; df = 234, P < 0.01) (Supplementary
Table 4). When one ES significantly increases from S1 to S4,
we find the other three ES also significantly increase across S1–S4
(Fig. 6). We also find S3 has the highest landscape connectivity
and lowest fragmentation, and has significantly greater ES
levels compared to all other land use scenarios (paired samples
t test; df = 235, P < 0.01). We find statistically significant ES
tradeoffs between S3 (i.e., future ERP) and S2 (i.e., development)
(paired samples t test; df = 235, P < 0.01). If the ERAs are
not implemented and Shanghai continues to pursue conventional
urbanization policies, we estimate Shanghai will receive:
0.33 million tonnes less carbon sequestration; 0.31 billion m3 less
water supply; 0.22 million kg less nutrient removal; and
0.02 million tonnes less coastal erosion control (Fig. 6). We also
observe a statistically significant increase in ES in S3 compared
to S1 (i.e., current ERP) since the main objective of the future
ERP is to enhance ecosystem functionality through restoration.
From the baseline scenario, we estimate S3 increases: carbon
sequestration by 0.42 million tonnes; water retention by
0.30 billion m3; nutrient removal by 0.50 million kg; soil retention
by 0.01 million tonnes.
However, synergies and tradeoffs on ES production shift when
evaluating ES production at the local scale (i.e., district level).
Locally, the most dramatic improvements and reductions in ES
occurr in S3. S3 has the most patches with greater than 1000%
improvement in carbon sequestration and nitrogen retention,
and greater than 100% improvement in water retention (Fig. 7).
However, substantial improvements in ES production in the
ERAs lead to the highest estimated (70–100%) reductions in the
same ES in southern districts. From the modeling, we observe a
tradeoff at the local scale: the expansion of ERAs by increasing
forest area causes increased development in southern districts.
Policymakers project Shanghai’s population will grow by 4% from
2014 to 2040. Under this condition, achieving more ecosytem
protection causes future development to increase in southern
districts. Therefore an important future research topic is
evaluating whether the spatial alterations in the ES production
translate to tradeoffs in the spatial distribution of ES beneficiaries.
To evalute ES outcomes and human welfare improvements will
require monitoring to: (
) find ways of matching ES supply and
ES demand (i.e., minimize the unequal distribution of benefits)
) determine ways of managing ERAs and other land uses to
meet Shanghai’s ES goals.
Our work provides new insights on integrating ES science into
urban planning by illustrating the various methodological steps of
the science–policy process from design to application in policy.
Currently, we lack science–policy examples of ES application in
policy8,44. A core objective of the ES approach is to integrate
ecosystem management into development decisions to promote
sustainability5, however, the majority of academic
recommendations on the advantages of ES assessment are theoretical with
minimal evidence of ES assessment leading to more
comprehensive planning7,21. A key reason for the science–policy
disconnect is that the majority of ES studies and assessments focus
on biophysical and/or monetary accounting with minimal
integration into a social (decision-making) process6. In our study we
implement a framework that details how to incorporate the needs
of stakeholders in particular policymakers into the development
of the ES science, and illustrate how policymakers can use the
science in the policy process. Our findings suggest that including
ES criteria into the urban planning process doubled ecosystem
protection in Shanghai. Stakeholders approved of the expansion
of ecosystem protection because ES information clarified how
Shanghai’s ecosystems support various societal goals. We found
the key is clearly relating ES information (ecosystem components
and human benefits) to the given institutional context by having
local policies guide the ES assessment framework. Scientists need
to relate ES categories and indicators to specific policy goals
(targets). For spatial planning, managers want credible and
legitimate spatial targets. Therefore we need to craft an
interdisciplinary science that relates stakeholder needs (wants) to core
ecological measurements across space and time.
Since 1997 the number of scientific publications addressing ES
has increased by near 30-fold, yet interdisciplinary assessments
make up less than 9% of ES studies45. A major challenge is
unifying fragmented disciplines to address real-world problems.
In our experience, we are finding a critical first step is formulating
practical interdisciplinary frameworks to organize indicators and
methodologies. The problem is the majority of existing ES
frameworks are general because they lack connections to actual
decision-making contexts46,47. Not engaging with
decisionmakers has led to a proliferation of frameworks and methods
that lack cohesion, which is limiting scientific progress47. Our
work addresses this specific gap since we utilize the ERP to
develop an approach for China’s regulatory context in order to
> 1000% Improvement
> 50% Improvement
> 1000% Improvement
> 500% Improvement
explore methodological steps of moving the science toward ES
standards. We recognize there are limitations to our ES
measurements since we were unable to create ecological production
functions37 given China’s policymaking timeline. Despite
limitations in the ES measurements, we address the fundamental
dilemma of whether ES standards (e.g., ES assessment for ERAs)
can inform urban planning. Overtime, our scientific capabilities
on ES measurements will improve, but better data alone will not
guarantee relevance for policy. Our work shows scientists must
explicitly consider the institutional context to advance the science
and practice of ES assessment in urban planning.
The ES approach can help organize core policies into societal
needs/wants to clarify how ecological changes associated with
different actions may impact desired goals. We use China’s
national EFZs and the Ministry of Environmental Protection’s
general guidelines on ecological redlines (Supplementary
Table 5). China currently has a unique decision-making context
since the government wants ES information to select ecological
redlines. However, the lack of ES frameworks for policy has led to
confusion and inconsistency across China. Having a
decisionmaking context where institutions need ES information should
accelerate the use of ES information, however, we are learning
that the organization of information and the type of information
remain critical for effective application. Many ES assessments
have been conducted in China, but few have led to local
policy changes, despite the central government’s mandate on
implementing the ERP. On February 14, 2017 the SMG officially
announced it will implement the ERAs, placing over two-fifths
of its land area and 1000 km2 of surrounding waters into
“no development zones” to preserve ecological functionality48.
Our assessment is not the final ERAs, but the work presented in
this study represents a stage of the policy process to establish
final ERAs. Shanghai officially released its preliminary ERAs on
June 28th, 2018. Also urban planners are using the ERP scenario
information to inform the creation of ecological corridors to
meet the 2040 ERA target. The objective is to increase ecosystem
connectivity by reducing structural fragmentation since current
ERAs are small patches. Full implementation of the ERAs will
take time since Shanghai has to find innovative political solutions
to reconcile serious tradeoffs. Doubling the amount of ecosystem
protection will require major land-use shifts.
From our experience, we identify four key lessons on the
science–policy process for integrating ES information into urban
planning. First governments likely are unable to implement
individualized ES standards for a diversity of ES. In the case of
China, the Chinese Government has been trying for over a decade
to implement EFZs for different ES. However, managers need
clear spatial targets to manage large landscapes. Thus, we believe
a plausible pathway on ES standards is establishing ES assessment
protocols to evaluate the provision of multiple ES from an
ecosystem area (spatial configuration) target. We evaluate the
ecosystem area target in terms of ES criteria where the evaluation
process has been standardized on evaluating ES. Second, scientists
need to develop frameworks outlining core socioecological
criteria for determining ecosystem area targets. We need clear
illustrations of the main methodological steps for ES assessment
to communicate the importance of different socioecological
components to stakeholders. In our assessment we select five
) ES hotspots, (
) ecologically fragile areas, (
biodiversity hotspots, (
) landscape structure, and (
opinions. Third stakeholder engagement and negotiation are
critical for implementing spatial targets. The SMG oversaw the
stakeholder negotiation process, but we worked closely with
policymakers to develop and refine ES maps to inform ongoing
negotiations. Furthermore, we worked closely with urban
planners to develop feasible scenarios where we utilize official
urbanization projections and spatial planning regulations to assess ES
tradeoffs. Lastly, for general planning we are finding basic ES
measurements are useful, but empirical measurements linking
ecosystem structure and functions to human benefits are needed.
Moving from planning to practice requires monitoring; the
challenge is measuring ES using ecological production functions
to link intermediate and final ecosystem services37. The next step
is to develop empirical datasets to link ecosystem characteristics
and human welfare benefits to evaluate the socioecological
outcomes from the ERAs to refine actual actions on the ground.
In conclusion, our study supports the use of ES information in
urban planning for developing more comprehensive plans on
ecosystem protection. Currently planning policies and actions are
informed by data with low temporal and spatial resolutions21.
Our study supports the hypothesis that ES assessment can help
policymakers generate more comprehensive spatial plans.
Moreover, we develop a new framework for the ERP, which we hope
can advance domestic efforts in China as well as similar efforts
internationally. Ultimately, our study helps provide evidence that
ES assessment can enhance urban planning when scientists
integrate institutional and ecological components together.
Selection of ecosystem services. An ES assessment was conducted to determine
an ecological redline for Shanghai Municipality, and assess the potential
effectiveness of ERP for enhancing ES. We worked with government officials and
surveyed the public to determine priority ES for the assessment. First, we worked
with the SEPB to link local and national policy goals under the ERP. Second, we
surveyed Shanghai residents to determine public preferences and expectations from
increased green space via the ERP. The questionnaire featured demographic
questions and ES questions in Mandarin. We conducted the questionnaire from
July 1 to August 1, 2014 (N = 849). We asked participants to select the ES they felt
most important to them and Shanghai. The questionnaire consisted of 21 questions
divided into four sections, including single choice, multiple choice, and open-ended
questions. We designed the first section to determine public perceptions of urban
green space in Shanghai. For example, we asked residents how often they visit local
parks, and their opinions (satisfaction and problems) of current green space in
Shanghai. The second section contained questions on public expectations of ES.
The third section asked for sociodemographic information: gender, age,
occupation, and income. We crafted the fourth section to determine the
distribution of beneficiaries of urban green space. We used random selection to
determine ten urban parks in Shanghai to conduct in-person interviews. The
total number of respondents was 849. For each respondent we collected
demographic information, such as gender, age, family size, education, occupation
class, location of residence, and income (see Supplementary Methods for details
on the social survey).
We used the questionnaire to determine the priority ES to the public in
Shanghai. We asked surveyors: “In your opinion, which of the following ecosystem
services is the most important for Shanghai and you (single choice)?” We listed the
following ES choices: (
) carbon sequestration, (
) water quality regulation, (
water resources conservation, (
) soil conservation, (
) biodiversity conservation,
) leisure and recreation, (
) flood regulation, and (
) other. Before introducing
the question, we gave a brief explanation of each ES and how they relate to people’s
daily lives. The top three ES were: water purification, water resources conservation,
and soil conservation. Lastly, we related the top ES of policymakers with the top ES
of surveyors to strike a balance between policy goals and public demand. The final
selected ES are: (
) water resources conservation; (
) water purification; (
) soil conservation; and (
) biodiversity conservation. For the
ecological redline analysis we categorize the first four as the ES criteria, and
biodiversity conservation as the biodiversity criteria.
LULC analysis. Multitemporal aerial images (spatial resolution = 0.5 m) were
taken by helicopters to determine LULC information for 2005 and 2014. We
obtained 41 images from January to March, 2005 and January to March, 2014.
The images were merged together in ERDAS Imagine 9.3, and manual visual
interpretation via ArcGIS 10.0 was used to delineate polygons for the five LULC
) forest; (
) agriculture; (
) constructed land; (
) open water; and (
beach. The overall classification accuracy is 94% for the 2014 LULC map
(Supplementary Methods for more details; Supplementary Table 6). The LULC
layers were converted to a grid format with a spatial resolution of 50 m to conduct
the ecological redline and scenario analyses.
Ecological redline analysis. First we performed the scientific analysis to identify
an optimal ecological redline target for Shanghai using three indicators: (
) ecologically fragile hotspots, and (
) biodiversity hotspots. Hotspots
have been widely used for identifying priority areas for biodiversity conservation,
and scientists are extending the concept to identify priority areas for ES and
vulnerable areas. Hotspots are areas of critical management importance since they
are critically threatened, high in ES production, and high in biodiversity49. The
Supplementary Methods contains a detailed explanation on our selection of the
indicators. For each of the three hotspots, we selected areas representing the top
10% of each indicator49,50; we verified this threshold relating each hotspot to the
total estimated ES production, total estimated vulnerable areas, and total estimated
suitable habitat for biodiversity (Supplementary Fig. 2). The total optimal ERA was
the sum of the hotspot areas, considering overlapping areas. Second, we engaged
in a stakeholder negotiation process among local governments, scientists, and
residents (Fig. 3e). Stakeholders first reviewed the optimal ERAs and identified
manageable areas (i.e., removal of small patches) then removed areas of serious
disagreement. The spatial planning process for ERAs was an iterative,
codevelopment process. Local governments reported to the SEPB who reported
to scientists where we assessed the impact of stakeholder concerns on the ES,
ecologically fragile areas, and biodiversity. We reported to the SEPB at least five
times per year; the SPLRA and SEPB used the scientific information to guide
the stakeholder visioning and negotiation on ERAs.
Ecosystem services hotspots were estimated using the InVEST 3.2.0 models,
which is a GIS-based method for estimating ES across a landscape, given different
LULC scenarios13,28,42. We calculated carbon storage using the InVEST carbon
storage and sequestration model to estimate aboveground biomass, belowground
biomass, soil, and dead organic matter per LULC type. Carbon sequestration was
evaluated as net primary production based on photosynthesis process. We
parameterized the model using biomass values (Supplementary Table 7) from
studies in Shanghai, and the Intergovernmental Panel on Climate Change
Guidelines for National Greenhouse Gas Inventories51. We compared our carbon
sequestration values with the empirical carbon sequestration values for Shanghai in
the literature52,53. We found no significant difference between our estimates for
carbon sequestration and the literature values (t test; t = −0.96, df = 8, P = 0.36).
Water resource conservation is defined as the ability of ecosystems to intercept
or store water resources from precipitation, which is calculated by precipitation
minus evapotranspiration and runoff. First, we estimated precipitation minus
evapotranspiration by using the water yield model in InVEST. Water conservation
was then evaluated as water yield minus runoff. The InVEST model estimates the
relative contributions of water from different parts of the landscape to evaluate how
possible changes in land use patterns could impact the annual surface water yield.
The model does not differentiate between surface, subsurface and baseflow, but
assumes water yield from a pixel reaches the point of interest via one of these
pathways. We derived input values using local data on rainfall, runoff54 and ET
coefficients39 (Supplementary Table 8). Hamel and Guswa55 found the InVEST
water yield model was able to represent differences in land uses and the spatial
distributions of water provisioning. The most important model parameters for
reducing model errors are climate variables in particular annual precipitation.
We obtained annual precipitation from 11 monitoring stations in Shanghai54. Our
modeled water yield value was 3.02 billion m3 similar to the observed value of
2.28 billion m3 in 201354.
Water purification was calculated using the InVEST nutrient retention model to
estimate the amount of nitrogen retained in the landscape based on runoff, digital
elevation model, soil characteristics, and the pollution export, and filtration
coefficients56 linked to different LULC types (Supplementary Table 9). InVEST
estimates the contribution of vegetation and soil to water purification through the
removal of nutrient pollutants from runoff. The model assumes that non-point
sources of water pollution result from export that can be mitigated by terrestrial
vegetation. We used discharge of dissolved nitrogen as our proxy for pollution. The
final model output is the total nitrogen retention (kg y−1) for each grid cell. We
compared our model results to four studies that measured nitrogen retention for
Taihu River Basin where Shanghai is located. Measured nutrient retention values
ranged from 1.56 to 7.27 ton km−2 (
); our modeled value was 2.02 ton km−2,
which falls within the range of reported observed values.
Soil retention was calculated using the InVEST sediment delivery ratio model
as the average annual amount of soil loss from each parcel of land. InVEST uses
the Universal Soil Loss Equation to identify the land parcel’s potential soil yield
and capacity to retain sediment60. Input data is DEM, management practices,
sediment retention value, vegetation cover, and management factor per LULC
type (Supplementary Table 10). Management factor reflects the impact of soil
and water conservation measures (e.g., terraced fields and cement protective walls)
on soil retention. Hamel et al.60 found the InVEST soil loss model is useful for
first order assessments of sediment dynamics. Shanghai is a highly urbanized
watershed, thus we expect low soil loss values due to the high percent impervious
surfaces. We compared our model values to measured soil retention in the Taihu
River Basin. Soil retention values from the literature were 1.1158 and 0.79 thousand
tonnes km−2 yr−161 similar to our value of 0.70 thousand tonnes km−2 y−1.
We calculated ecologically fragile hotspots by first estimating the sensitivity of
Shanghai’s ecosystems to three key external disturbances (i.e., main drivers of
ecological change): soil erosion, desertification, and salinization. We created a
scoring matrix for each disturbance using expert consultation based on the Delphi
method then classified each indicator into five levels using average scores
(Supplementary Tables 11 and 12). We defined areas with the highest sensitivity
as ecologically fragile hotspots (top 10% of most sensitive areas to the selected
We calculated biodiversity hotspots using the InVEST model for habitat quality,
which estimates the extent of suitable habitat for organisms by combining
information on LULC suitability and threats to biodiversity. This approach
generates information on the relative extent and degradation of different habitat
types in a region. The model is based on the hypothesis that areas with higher
habitat quality support higher richness of native species, and decreases in habitat
extent and quality lead to reductions in species persistence. Habitat quality is
estimated as a function of four factors: (
) relative impact of each threat; (
sensitivity of each habitat type to each threat (Supplementary Table 13); (
distance between habitats and sources of threats (Supplementary Table 14); and
) degree of legal protection. Terrado et al.62 found InVEST estimates for
biodiversity richness are useful surrogates for biodiversity in terrestrial and
aquatic ecosystems; modeled habitat quality correlated with biodiversity at the river
The stakeholder negotiation process was led by the SEPB and SPLRA who
oversaw an iterative process of scientific analysis, stakeholder visioning, review/
negotiation, and ERP development. The SPLRA first invited relevant stakeholder
groups to participate in the scoping discussions (i.e., district governments and
scientists). SPLRA established an interdisciplinary team of scientists, consisting of
specialists from the Shanghai Academy of Environmental Sciences, Shanghai
Urban Planning and Design Research Institute, Shanghai Municipal Institute of
Surveying and Mapping, and the Shanghai Ocean Planning and Design Research
Institute, etc. We worked collaboratively with SEPB and SPLRA to conduct the
ecological redline analysis to determine the optimal ERAs. There are two types
of consultations: (
) district level and (
) public comment. At the district level
workshops SEPB and SPLRA presented the spatial maps to district governments
and scientists. Together we discussed our selection of the ERA criteria, our
methodology, and the results (Supplementary Table 2 and Supporting Materials
describe the stakeholder groups in detail). Local governments reported stakeholder
visioning, expertise, and concerns to the SEPB and SPLRA who reported
stakeholder input to us. We utilized the stakeholder information to refine the
scientific analysis. Subsequently the scientific information was reported to the SEPB
and SPLRA who negotiated with stakeholders to find a balance between their
concerns and securing ES, ecologically fragile areas, and biodiversity.
Land use scenarios. We worked with policymakers and urban planners to
determine development scenarios for the SMG and Shanghai Development and
Reform Commission. The SMG was developing Shanghai’s Urban Plan
(2016–2040), thus policymakers wanted information, comparing baseline ERAs to
different land use scenarios to determine the best approach for improving the
environment. First the SMG predicted future economic and population growth
rates working with scientific and policy research teams (e.g., Fudan University,
Shanghai Academy of Social Sciences, Shanghai Tongji Urban Planning and Design
Institute) using past population trends to generate future population projections40.
They generated linear regression models considering population growth factors
(e.g., historical birth and death rate per age group, degree of education,
immigration and migration rates) and GDP growth rate. Next, we refined the
scenarios using different spatial planning policies for regulating urban spatial form
defined by policymakers. We created one baseline scenario and three alternative
future scenarios (Fig. 5; Table 2):
Scenario 1 (S1) is ERP baseline using current LULC with targeted ERAs for 2014.
Scenario 2 (S2) is development scenario with no ERP implementation for 2040
and no policy constraints on development. This scenario is characterized by
uncontrolled urbanization with projected population size of 30.69 million (25%
total growth rate) and 5% GDP growth rate in 2040.
Scenario 3 (S3) is future ERP scenario for 2040 is where we worked with
policymakers to increase ERAs by 501 km2, representing optimal areas for
enhancing ecosystem connectivity among natural and seminatural habitats. This
scenario is characterized by condensed and slower urbanization in which
population size is limited to 25 million (4% total growth rate) and 5% GDP growth
rate in 2040. The main LULC difference between S3 and S1 is the expansion of
) vegetation buffers along river banks (main action); (
industrial areas to forests (afforestation); and (
) transforming agricultural areas to
Scenario 4 (S4) is planning scenario for 2040 considers existing ecological
protection measures in Shanghai’s Urban Plans (1999–2020 and 2016–2040),
excluding the ERP. This scenario is characterized by condensed and slower
urbanization in which population size is limited to 25 million (4% total growth
rate) and 5% GDP growth rate in 2040. Also, S4 and S3 both implement the
permanent farmland policy and urban boundary policy, which are aimed at
reducing cropland losses.
We forecasted different LULCs for alternative scenarios using the: (
model to estimate areas of different land-uses in 2040 and (
) CLUE_S model to
estimate spatial patterns in 2040 (Supplementary Fig. 6). For the Markov model, we
generated a transfer area matrix and transfer probability matrix for 2005–2014
(Supplementary Table 15). The Kappa coefficient was used to validate the
simulation results where the Kappa coefficient was 0.88, indicating suitable
simulation results. We used the Markov model to predict LULC areas under S2,
and Shanghai’s Urban Plan (2016–2040) to estimate LULC areas under S3 and S4.
Next, we created logistic regression models to determine the relationship
between various LULC types (response variables) and drivers of LULC change
(explanatory variables) to forecast the spatial distribution of LULCs. We selected
seven known drivers: (
) GDP, (
) population density, (
) distance to main rivers,
) distance to roads, (
) distance to railways, (
) distance to ports, and (
to airports. Population and GDP were collected from the Shanghai Statistical
Yearbook and Shanghai’s Urban Master Plan (2016–2040); the other factors were
calculated in ArcGIS 10.0 using the spatial analysis module. We only set one LULC
rule for S3 where the established ERAs in S1 are not changed in S3; no other rules
were set. LULC types can transform freely from one type to another. We generated
the logistic regression models to determine the probability of occurrence of each
LULC type in a particular grid (Supplementary Table 16). We used the Relative
Operating Characteristic value (ROC) to test the significance of the logistic
regression models. ROC values were greater than 0.7 (Supplementary Table 18),
indicating model suitability for predicting and simulating the spatial pattern of
Lastly, we used the CLUE_S model to generate alternative future scenario maps.
CLUE_S model is a dynamic, spatially explicit LULC model developed for small
regions (e.g., watershed or province). CLUE_S has been widely used throughout the
world to develop future LULC scenarios for cities. The model is subdivided into
two distinct modules: (
) nonspatial demand, which calculates the area change for
all LULC types at an aggregate level and (
) spatially explicit allocation procedure
to translate LULC changes to different locations within a study region using a
raster-based system. Model conditions must be defined by the user for four
categories, representing each scenario: (
) spatial policies and restrictions; (
use type specific conversion settings; (
) land use requirements (demand); and
) location characteristics. We ran CLUE_S using the following inputs: (
land use map for 2014 as the reference map; (
) predicted LULC areas for
alternative scenarios from the Markov model; and (
) logistic regression model
results. We used the Kappa coefficient to evaluate the accuracy of CLUE_S model
results, and obtained a Kappa coefficient of 0.92, indicating suitable spatial
Evaluating effectiveness of ERP. We compared four different development
scenarios using three criteria: (
) ecosystem composition, (
) ecosystem configuration,
) ES. ES were estimated using the above methods for the land use scenarios.
To assess ecosystem composition we used ArcGIS 10.0 to measure the percent area
and the total area of each land cover type. We analyzed the configuration using
FRAGSTATS to calculate: (
) landscape CI and (
) landscape FI. We conducted a
paired t-test comparing mean ES levels at the sub-district level (N = 236
subdistricts) to determine statistically significant ES tradeoffs among scenarios. We
also conducted a correlation analysis on ES for each scenario to evaluate the
strength of the synergies among ES.
Data availability. All relevant data are available upon request from the authors.
Y.B., C.P.W, B.J., and M.W designed research; Y.B. and Q.W. performed research; Y.B.
and B.J. analyzed data; Y.B., C.P.W., B.J. and A.C.H. wrote the paper.
Supplementary Information accompanies this paper at
Competing interests: The authors declare no competing interests.
Reprints and permission information is available online at http://npg.nature.com/
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons
Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
1. Millennium Ecosystem Assessment (MA) Ecosystems and Human Well-being: Synthesis . (Island Press: Washington, DC, 2005 ).
2. Guerry , A. D. et al. Natural capital and ecosystem services informing decisions: from promise to practice . Proc. Natl Acad. Sci. USA 112 , 7348 - 7355 ( 2015 ).
3. United States Environmental Protection Agency (US EPA) Valuing the Protection of Ecological Systems and Services: A Report of the EPA Science Advisory Board. (US EPA: Washington , DC, 2009 ).
4. De Groot. , R. S. et al. Challenges in integrating the concept of ecosystem services and values in landscape planning, management and decision-making . Ecol. Complex . 7 , 260 - 272 ( 2010 ).
5. Carpenter , S. R. et al. Science for managing ecosystem services: beyond the Millennium Ecosystem Assessment . Proc. Natl Acad. Sci. USA 106 , 1305 - 1312 ( 2009 ).
6. Albert , C. et al. What ecosystem services information do users want? Investigating interests and requirements among landscape and regional planners in Germany . Landsc. Ecol. 29 , 1301 - 1313 ( 2014 ).
7. Förster , J. et al. Assessing ecosystem services for informing land use decisions: a problem-oriented approach . Ecol. Soc . 20 , 1 - 19 ( 2015 ).
8. Martinez-Harms , M. et al. Making decisions for managing ecosystem services . Biol. Conserv . 184 , 229 - 238 ( 2015 ).
9. Ouyang , Z. et al. Improvements in ecosystem services from investments in natural capital . Science 352 , 1455 - 1459 ( 2016 ).
10. Bateman , I. J. et al. Bringing ecosystem services into economic decisionmaking: land use in the United Kingdom . Science 341 , 45 - 50 ( 2013 ).
11. Sarukhán , J. et al. Natural Capital of Mexico . Synopsis: Current Knowledge, Evaluation, and Prospects for Sustainability. (National Commission for the Knowledge and Use of Biodiversity, Mexico City , Mexico, 2010 ).
12. Mária Paula , B. & Néstor Oscar , M. Land use planning based on ecosystem service assessment: a case study in the Southeast Pampas of Argentina . Agric. Ecosyst. Environ . 154 , 34 - 43 ( 2012 ).
13. Goldstein , J. H. et al. Integrating ecosystem service tradeoffs into land use decisions . Proc. Natl Acad. Sci. USA 109 , 7565 - 7570 ( 2012 ).
14. Cabral , P. et al. Assessing the impact of land cover changes on ecosystem services: A first step toward integrative planning in Bordeaux , France. Ecosyst. Serv. 22 , 318 - 327 ( 2016 ).
15. McKenzie , E. et al. Understanding the use of ecosystem service knowledge in decision-making: lessons from international experiences of spatial planning . Environ. Plann. C Gov. Policy 32 , 320 - 340 ( 2014 ).
16. Tammi , I. , Mustajärvi , K. & Rasinmäki , J. Integrating spatial valuation of ecosystem services into regional planning and development . Ecosyst. Serv . 26 , 329 - 344 ( 2017 ).
17. Arkema , K. K. et al. Embedding ecosystem services in coastal planning leads to better outcomes for people and nature . Proc. Natl Acad. Sci. USA 112 , 7390 - 7395 ( 2015 ).
18. Ruckelshaus , M. et al. Notes from the field: lessons learned from using ecosystem service approaches to inform real-world decisions . Ecol. Econ . 115 , 11 - 21 ( 2015 ).
19. Polasky , S. , Tallis , H. & Reyers , B. Setting the bar: standards for ecosystem services . Proc. Natl Acad. Sci. USA 112 , 7356 - 7361 ( 2015 ).
20. Hansen , R. et al. The uptake of the ecosystem services concept in planning discourses of European and American cities . Ecosyst. Serv . 12 , 228 - 246 ( 2015 ).
21. BenDor , T. K. et al. A research agenda for ecosystem services in American environmental and land use planning . Cities 60 , 260 - 271 ( 2017 ).
22. People 's Republic of China Ministry of Environmental Protection (MEP) . Technical guide for ecological red line . No. 2015 . 5. (Chinese Ministry of Environmental Protection , Beijing, China, 2015 ).
23. Bai , Y. et al. New ecological redline policy (ERP) to secure ecosystem services in China . Land Use Pol . 55 , 348 - 351 ( 2016 ).
24. People 's Republic of China State Council. Integrated Reform Plan for Promoting Ecological Progress (State Council , Beijing, China, 2015 ).
25. United Nations Environment Programme (UNEP). Green is Gold: The Strategy and Actions of China's Ecological Civilization. (UNEP, Nairobi , Kenya, 2016 ).
26. Fan , J. et al. Major function oriented zone: new method of spatial regulation for reshaping regional development pattern in China . Chin. Geogr. Sci . 22 , 196 - 209 ( 2012 ).
27. State Council of the People's Republic of China (State Council PRC) . National Key Functional Zoning Plan . ( State Council PRC , Beijing, China, 2010 ) [in Chinese].
28. The Economics of Ecosystems and Biodiversity (TEEB) . TEEB for National and International Policy Makers. (Earthscan , Washington, DC, 2009 ).
29. Díaz , S. et al. The IPBES Conceptual Framework - connecting nature and people . Curr. Opin. Environ. Sustain 14 , 1 - 16 ( 2015 ).
30. United States Environmental Protection Agency (US EPA ) . National Ecosystem Services Classification System: Framework Design and Policy Application . (US EPA , Washington, DC, 2015 ).
31. Fu , B. et al. Linking ecosystem processes and ecosystem services . Curr. Opin. Environ. Sustain 5 , 4 - 10 ( 2013 ).
32. Chan , K. M. A . et al. Where are cultural and social in ecosystem services? A framework for constructive engagement . Bioscience 62 , 744 - 756 ( 2012 ).
33. Rosenthal , A. et al. Process matters: a framework for conducting decisionrelevant assessments of ecosystem services . Int. J. Biodivers. Sci. Ecosyst . Serv. Manag. 11 , 190 - 204 ( 2015 ).
34. Crossman , N. D. et al. A blueprint for mapping and modelling ecosystem services . Ecosyst. Serv. 4 , 4 - 14 ( 2013 ).
35. Luck , G. W. et al. Identifying spatial priorities for protection ecosystem services . F1000Research 1 , 1 - 16 ( 2012 ).
36. Neugarten , R. A. et al. Rapid assessment of ecosystem service co-benefits of biodiversity priority areas in Madagascar . PLoS ONE . 11 , e0168575 ( 2016 ).
37. Wong , C. P. et al. Linking ecosystem characteristics to final ecosystem services for public policy . Ecol. Lett . 18 , 108 - 118 ( 2015 ).
38. Haines-Young R. H . et al. Ecosystem Ecology: A New Synthesis . (eds Raffaelli, D. & C. Frid). 110 - 139 (Cambridge University Press, 2010 ).
39. Sharp , R. et al. InVEST +VERSION+ User's Guide. The Natural Capital Project , Stanford University, University of Minnesota, The Nature Conservancy, and World Wildlife Fund ( 2016 ).
40. Shanghai's Urban Plan ( 2016 - 2040 ). Outline of Shanghai Mater Plan ( 2016 - 2040 ). Leading Group Office of Shanghai Master Plan, Shanghai, China ( 2016 ); Retrieved from: http://2040.shgtj. net (July 28 , 2017 ) [in Chinese].
41. Shanghai Municipal Government. Half of Shanghai's rivers and lakes heavily polluted . Retrieved from: http://www.shanghai.gov.cn/shanghai/node27118/ node27818/u22ai73035.html ( August 9 , 2013 ).
42. Nelson , E. et al. Modeling multiple ecosystem services, biodiversity conservation, commodity production, and tradeoffs at landscape scale . Front. Ecol. Environ. 7 , 4 - 11 ( 2009 ).
43. Zhao , S. et al. Ecological consequences of rapid urban expansion: Shanghai, China . Front. Ecol. Environ . 4 , 341 - 346 ( 2006 ).
44. Daily , G. C. et al. Ecosystem services in decision-making: time to deliver . Front. Ecol. Environ . 7 , 21 - 28 ( 2009 ).
45. Abson , D. J. & Hanspach , J . Response to Turnhout et al.' s rethinking biodiversity: from goods and services to “living with . ” Conserv. Lett. 7 , 334 - 335 ( 2014 ).
46. Scarlett , L. & Boyd , J. Ecosystem services and resource management: institutional issues, challenges, and opportunities in the public sector . Ecol. Econ . 115 , 3 - 10 .
47. Polasky , S. , Tallis , H. & Reyers , B. Setting the bar: standards for ecosystem services . Proc. Natl Acad. Sci. USA 112 , 7356 - 7361 ( 2015 ).
48. Huang , M. Shanghai puts two-fifths of landmass in ecological protection zone . Xinhua, Retrieved from: http://news.xinhuanet.com/english/2017-02/14/ c_136056269. htm ( February 14 , 2017 ).
49. Egoh , B. et al. Mapping ecosystem services for planning and management . Agric. Ecosyst. Environ . 127 , 135 - 140 ( 2008 ).
50. Eigenbrod , F. et al. The impact of proxy-based methods on mapping the distribution of ecosystem services . J. Appl. Ecol . 47 , 377 - 385 ( 2010 ).
51. Intergovernmental Panel on Climate Change (IPCC). 2006 IPCC Guidelines for National Greenhouse Gas Inventories. (Institute for Global Environmental Strategies , Japan, 2006 ).
52. Wang , Y. P. Net primary production of urban vegetation ecosystem in Shanghai. Dissertation for master's , East China Normal University ( 2009 ) [in Chinese].
53. Han , B. et al. Soil carbon sequestration and its potential by cropland ecosystems in China . Acta Ecol. Sin . 28 , 612 - 619 ( 2008 ).
54. Shanghai Water Authority. Shanghai Bulletin of Water Resources. (Shanghai, 2013 ).
55. Hamel , P. & Guswa , A. J. Uncertainty analysis of a spatially explicit annual water-balance model: case study of the Cape Fear basin , North Carolina. Hydrol. Earth Syst. Sci . 19 , 839 - 853 ( 2015 ).
56. Wang , X. Y. , Guo , B. & Jiang , L. Progress on soil erosion in watershed scale, based on USLE, GIS and RS . Subtrop. Soil Water Conserv . 24 , 42 - 48 ( 2012 ).
57. Jia , J. M. et al. Valuation of changes of ecosystem services of Tai Lake in recent 10 years . Acta Ecol. Sin . 35 , 2255 - 2264 ( 2015 ).
58. Dong , C. Y. & Gao , J. F. Assessment for polder terrestrial ecosystem functions of maintaining and regulating in the western Taihu basin . J. Nat. Resour . 29 , 420 - 430 ( 2014 ).
59. Yang , L. Y. et al. Water ecological function of green space in Anji city, Taihu lake basin . J. Nat. Resour . 26 , 599 - 608 ( 2011 ).
60. Hamel , P. et al. A new approach to modeling the sediment retention service (InVEST 3.0): Case study of the Cape Fear catchment , North Carolina, USA. Sci. Total Environ. 524-525 , 166 - 177 .
61. Zhang , B. et al. The benefit assessment on reduction of soil erosion and nutrient loss from the forests in Anji county of Taihu basin . Res. Soil Water Conserv . 18 , 111 - 118 ( 2011 ).
62. Terrado , M. et al. Model development for the assessment of terrestrial and aquatic habitat quality in conservation planning . Sci. Total Environ . 540 , 63 - 70 ( 2016 ).