Increasing urbanisation of the growing global population forces new and increasingly complex challenges into the urban system. The United Nations predict world population to soar from the current 7.6 billion to 9.8 billion by 2050 and 11.2 billion by 2100 [1
]. Simultaneously, reports abound the rapid influx of countryfolk to urban centres, increasing from 50% globally to 68% by 2050 [2
], bringing with them demands for marketed goods and services such as materials and energy, and demands for non-marketed services related to ecosystem functions. However, the demand for marketed goods and services is not without consequences and making them available (extracting raw materials, processing them, transporting, using the products and discarding them) inevitably has impacts on the environment. In particular, this can degrade the geobiophysical components and processes of an ecosystem that underpin the biological cycles, and the services they provide.
Ecosystem services (ES) are the ecosystem-supplied goods and services that benefit humans in the forms of provisioning services (e.g., food yield), regulating services (e.g., climate, water and nutrient cycles) and cultural services (e.g., aesthetic and healthful nature interactions) [3
]. Thus, it becomes necessary to understand how urbanisation can be channelled to consolidate and mobilise behaviours and decisions at the urban level to reduce humanity’s effect on ecosystems’ capacities to supply ES at global and local levels. However, there is still little knowledge on the magnitude, drivers and effects of interregional ES flows, especially for regulating and cultural ES [4
In this research, we focus on the increasing urbanisation, so it follows that we pay attention to the ES supplied to cities, and impacted by urban activities [5
]. Accordingly, we draw the reader’s attention to the concept of urban ecosystem services (UES)—the subset of ES that supply benefits within the urban system [6
]. This includes both locally specific ES (e.g., flood protection) and global ES that are nonetheless relevant to urban liveability, such as global climate regulation.
In anticipation of the rising social and ecological demands concentrated in cities, it is necessary to quantify the cause–effect relationships between elements of the urban system, such as material and energy flows, and their impact on ecosystem functionality. This will enable society to make more socially and ecologically conscientious decisions today for the needs of tomorrow by providing pathways that allow cities to tackle not only urban, but also global sustainability development goals [7
]. To this end, the implementation and use of tools capable of modelling and assessing the relationships between elements within the urban system will be key to support the development of policies oriented to preserve the supply of UES.
Several assessment methods and coupled, integrated, hybrid methods have been proposed to measure and trace the social and environmental impacts associated with the flow of materials and energy. For example, these include life cycle assessment (LCA) [8
], mass balance [9
], emergy analysis [10
], cost-benefit analysis [11
] and multi-criteria decision assessment [12
]. Some of these can be adapted to suit the urban level [13
]. Moreover, many of these assessment methods sit under the umbrella metaphor of urban metabolism (UM). UM is used to describe the material, energy, social and economic flows (‘metabolic flows’) through the urban system [14
]. The term was coined by Karl Marx [15
] and later brought to popularity by Wolman [16
] and has since opened up a large research domain linking the disciplines of engineering, political ecology and political economy, industrial ecology, social ecology and ecological economics [17
]. We define UM as the metaphor for the stock and flow of socio-ecological resources circulating in and through the urban system. As Wolman (1965) suggested, this helps understanding the complex dynamics and interdependencies between urban environments and their surroundings (i.e., ecosystems). Therefore, UM may be a suitable assessment framework that can be expanded to elucidate the valorisation of ES.
UM models usually assess the flow of physical quantities of water [21
], materials [22
], food [23
], chemical elements, solar emergy joules [24
], and various energy units [25
]. Some of these metabolic flows can also be used to assess and model ES [26
] thus creating an opportunity to connect the relatively newer concept of UES to UM assessment methods [17
]. Furthermore, there have been compelling appeals for UM research to expand its scope towards the holistic assessment of various new levels, including ES [17
]. In fact, urban planning of phenomena/structures capable of generating UES may bring opportunities to enhance the resilience and ecological functioning of urban systems [36
]. This leads us to question how the ES concept can be integrated into an UM-based method to allow a better assessment of UES.
Several ES indicators are implicitly assessed by UM methods. Urban Material/Energy Flow Analysis (MEFA) traces the flow of products that do not use ES nomenclature, but are nonetheless equivalent to many provisioning ES. In the UM community, yields of provisioning ES correspond to raw materials/resources. For example, the stock and flow of energy, water, food and material provisions are traditional metabolic flows while also being the physical exploits of ES. In general, we can use MEFA, mass balance, or emergy analysis for the quantification of provisioning ES. However, it may be more challenging to use these UM methodological bases for the quantification of regulating and cultural ES. Burkhard and Maes [37
] suggest that for these latter groups of ES, simple measurement of stocks and flows are not enough, but a more complex modelling approach is needed. Subsequently, we pose the following research questions to be explored in this paper:
What are the steps needed to include UES assessments within an UM modelling framework, and what are the relevant key methodological issues that such a UM-UES paradigm may enhance or resolve?
We believe these questions can be addressed by exploring the viability of creating an integrated UM-UES modelling approach. This approach may add to the completeness and representativeness of these two currently distinct concepts and foster the development of an integrated urban sustainability analysis method. To this end, the first objective of this paper is to critically review UM literature that encompasses the ES concept and identify the approaches that may allow or improve the integration of UES assessment. This can provide a relevant scientific background, methodological knowledge and research evidence to address the second objective of this study, which is to define pathways towards an integrated UM method enabling a quantitative assessment of UES.
The main message we could retrieve from the above analysis of key investigation themes is that a critical gap exists between state-of-the-art UM assessment methods and their application to assessing UES. In fact, the reviewed literature did not even suggest the linkage between UM and UES. We have seen that some modelling frameworks are favoured for different types of information: spatial, temporal, different levels and scales, and the dynamics between them. All the cases we reviewed do not capture the full complexity of the UM stocks and flows, but between these various approaches some strengths can be drawn.
In the next sections, we inspect and discuss the main causes underlying the omission of UES assessments in the reviewed UM case studies and underlying models, emphasising on the potential assets for further improvement of the UM-UES framework. Based on these strengths and weaknesses of those cases we go on to propose pathways towards an advanced integrated UM-UES modelling framework, including which methodological aspects should be incorporated and which have lower priority.
4.1. Modelling Complex Information
Spatio-temporal details are a major part of the complexity of urban systems, and the methods for integration of these details in UM approaches remain an open research question [32
]. Urban activities, and therefore the metabolic flows, occur in heterogeneous spatial patterns [105
]. While we found studies that addressed either spatial heterogeneity or considered temporal evolution, only one study simulated future spatio-temporal patterns by implementing the dynamic land use change model LEAM [68
Accounting for spatio-temporal heterogeneity requires additional data complexity. Using static spatial information only provides snapshots of the spatial patterns. In contrast, dynamic land use models are useful for predicting some ES supplies, but not all UES are so directly linked to land use [107
]. For example, supply of climate regulation is global regardless of the land use and origin of greenhouse gas emissions [108
]. However, understanding how these patterns change in relation to material-energy stocks and flows and their impact on the supply of ES requires a detailed database of historical land use maps from which those spatio-temporal dynamics may be revealed, calibrated and validated [47
]. Land use and land cover maps (e.g., Urban Atlas; Montero, et al. [110
]) can be used to estimate physical qualities (e.g., as surface type and imperviousness, tree cover), and as proxies to spatially disaggregate material-energy stocks and flows (e.g., construction materials in urban fabric) [110
]. In this case, data should be disaggregated to a spatial unit of measurement to capture the spatial patterns. For example, the space within the urban boundary may be subdivided into spatial units each described by a matrix of material and energy flow data specific to that spatial unit. These data can provide the information for measuring UES indicators for that spatial unit [104
]. This is illustrated in Figure 2
, which represents the multi-level system in terms of a foreground urban level nested within a background global ecosystem. The background level serves as source and sink of resources and emissions, while the urban level houses the socio-economic demands for those resource flows. In doing so, the urban system causes impacts to the supply of ES, both within the urban level (the UES), but also the ES at the non-urban level. Those UES are linked to spatial units. Black stocks and flows represent the aspects of the UM metaphor which are already exercised in state-of-the-art methods. That is, the material-energy inputs and outputs, and the disaggregation within the subsystem by ENA or spatial information. The grey lines and icons represent exchanges between human and natural systems, which are currently missing in state-of-the-art UM methods, and necessary for the advancement to a holistic UM-UES integrated model.
Among the strengths of using such an approach is the opportunity to represent spatially relevant patterns at high resolution and eventually quantify and map UES and their values according to well-established modelling tools for the ES community [96
]. The individual models case studies found in this literature review are apparently not sufficient to cope with these aims, however, so an appropriate integrated modelling framework is needed to support the data complexity.
4.2. Methodological Bases
We saw in Section 3.6
that material-energy flow analysis, emergy analysis, network analysis, system dynamics, and life cycle approaches were used to illuminate and reveal different aspects of the urban complexity. Some of these model types may be especially relevant for advancing a UM-UES modelling framework thereof.
To understand the link between urban activities and associated socio-ecological impacts the data should capture both the direct and indirect impacts (embodied along the life cycle of a material) [86
]. This requires taking a life cycle thinking approach when defining the system boundary of material and energy flows whose life cycle impacts transcended the urban level [33
]. Data that only deal with impacts at the urban level can give limited results by missing the impacts on ES supplies up and downstream in the life cycle of material-energy flows. This was the purpose of Goldstein, et al. [113
] developing the UM-LCA, from which it was concluded that embodied impacts associated with urban metabolic flows in fact are not trivial in the calculation of urban environmental footprints.
While this model focused on the environmental impacts typically considered in LCA, it could also be useful to inform the socio-ecological aspects of an UM. Embodied impacts can be measured with, for example, environmentally extended multi-regional input-output tables [114
]. However, the disaggregation of economic sectors into the necessary granularity of the urban scales (e.g., to specific material flows) involves assumptions and increases uncertainty. Many of the ENA models described in Section 3
used input–output data that are mostly available at the city level in China. European cities rarely have city level input–output data tables available. Some studies have been done on the factorisation of city-specific data from national level input-output (IO) tables. A hybrid method for constructing regional (e.g., city-level) IO tables is described by Miller and Blair [115
]. This may provide the best solution for using IO data to capture those embodied impacts. Alternatively, other life-cycle inventories based on bottom-up data collection are available for specific regions, especially in Europe. However, in the case where city-specific life cycle inventory data are not available there would be associated with applying one city’s data to fill the gaps of another. In this case, IO tables seem to offer the more attractive option due to their methodological consistency and replicability across cities and nations. IO and multi-regional IO tables can also inform the socio-economic flows entering and leaving the urban system. However, as the relationships between socio-economic elements in the urban system should not be assumed as linear, additional information is needed to link these flows to UES [116
]. These data should be collected on a city-specific basis and validated by (and calibrated to) observed historical time-series data to understand their potentially emergent properties [116
4.3. Linking Elements of the Urban System
CHANS is an important characteristic of urban systems [44
]. Understanding the interactions between elements in the urban system, especially the CHANS, is a recurrent theme in urban systems modelling [20
]. While it is not yet well-adopted in most UM studies, strong cases have been made for including related aspects such political and social ecologies in the UM framework [17
One of the models assess in the literature review—MuSIASEM—was designed to model urban systems considering CHANS [119
]. However, MuSIASEM is not a dynamic forecasting model meaning it does not reveal the self-organising and emergent relationships borne of the complex internal urban dynamics [116
]. Modelling the urban system without considering those causal relationships thus limits the representativeness of information pertaining to spatio-temporal dynamics [120
]. In contrast, SD modelling takes causal relationships into consideration and equates them by a system of difference equations. For example, the MIMES (Multiscale Integrated Models of Ecosystem Services) is a modelling approach that models ES linked to CHANS across scales, but it is not based on UM stocks and flows [123
]. However, based on the findings in this review the tailored UM-UES framework should also be based on SD. This is supported by the conclusions of Beloin-Saint-Pierre, et al. [32
]. SD modelling frameworks can represent UES in spatio-temporally specific detail across multiple levels and acknowledges the role of the relationships between system elements [124
]. Existing attempts based on the MIMES suggest that new models may be specifically tailored for the network UM framework to track physical flows as influencers of ES supply to include aspects of urban material-energy flows [123
]. Combining these attributes—spatially explicit, multi-level, life cycle information and dynamic cause–effects integrated across social, economic and ecological—can provide a detailed holistic representation of the complex urban dynamics [133
Network analysis is another fundamental piece of knowledge capable of linking elements of the UM system [32
] that was used in many studies reviewed in this paper [53
]. Network nodes usually represent economic sectors while materials, water, energy, and waste flows are transferred between nodes, represented by network edges. This method is a useful way to measure the distribution of material-energy flows to specific nodes or sector and therefore the competition between different sectors for those resources [134
]. Network analysis can also be linked to spatial information (e.g., land use maps) as demonstrated by Leduc and Van Kann [66
], and can be used with CHANS to model relationships between ecosystems and political economy [135
], to address for instance the sociological issues raised by Pincetl [17
]. Network analysis has been applied to trophic webs [136
] and increasingly to other aspects of ecology [99
]. How to quantify UES according to network models for UM thus opens another room for further research and development. Incorporating network analysis in modelling ecosystems can further facilitate the linkage from UM flows to CHANS [125
4.4. Pathways Towards UM-UES Assessments
Our results showed that the UM metaphor, subject to some reorientation and increased modelling complexity, has the potential to assess provisioning and regulating UES, thus strengthening the socio-ecological capabilities of UM. However, in our review we saw no convergence towards measuring and mapping cultural UES. The UM assessments we reviewed considered various geobiophysical flows and derived socio-economic indicators, but the educational, intrinsic, spiritual or recreational phenomena were not assessed. These types of cultural ES are not so obviously connected to resource flows, and so it is not surprising that the UM community, which has historically dealt with water, energy, material and economic indicators, does not yet intersect with this dimension of ES research. Even in the ES research field, assessment of cultural ES is an open question [137
]. This may be because cultural UES are generally more difficult to conceive of in physical quantities [137
]. Proposals to measure and map cultural ES tend to focus around variables such as accessibility to and visitation of natural structures [95
]. As we have seen, these evaluations share little in common with state-of-the-art UM methods. However, we did encounter UM case studies that used survey and participatory methods [62
]. This type of UM assessment may hold the key to incorporating cultural services in a comprehensive integrated UM-UES model. Studies focusing on the advancement of cultural UES assessments emphasise the need to take a systematic CHANS approach using social science tools such as participatory mapping, structured interviews, and linking these to spatial information [138
Most importantly, our review identified links between the types of information deemed relevant to assessing and predicting UES, and the modelling tools being used in UM research. That is, to model and predict spatio-temporal changes in UES that relate to UM stocks and flows, models must respect the cause–effect relations of CHANS, embodied life cycle impacts of material and energy stocks and flows, and spatio-temporal information. Therefore, the integration of SD, ENA, and LCA modelling approaches must be considered in the evolution of the UM-UES models. The proposed set of pathways (Figure 3
) shows our results on investigation themes relate to assessing provisioning, regulating and cultural UES using the UM metaphor. Figure 3
is composed of three layers: ES (what we want to model), investigation themes (information the models should incorporate), and modelling complexity (the types of models that may be needed). Lines representing dependencies link these layers.
The top and middle layers, representing ES and investigation themes, are connected by solid lines or dashed lines. Solid lines represent an investigation theme as necessary for the assessment of the connected ES group. Dashed lines stemming from cultural ES are connections that were not identified in the literature albeit may have potential to be developed. This is consistent with current research on cultural ES, which generally points to a lack of cultural ES assessments due to the need for spatially specific information, interregional (i.e., multi-level) flows and integrated modelling across social and ecological scales [4
The layers representing investigation themes (middle) and modelling complexity (bottom) are connected by solid lines or dotted lines. Solid lines represent prioritisation of the modelling approach to cope with the data detail associated with that investigation theme based on the literature we reviewed. Dotted lines represent links that are possible but that are not a priority (i.e., the modelling approach has the capacity to deal with that data type, but the investigation theme is out of the scope of the model).
These pathways emphasise the importance of integrated analysis based on SD, life cycle thinking and ENA in order to assess provisioning, regulating and cultural UES in a comprehensive, integrated methodological framework. Each of these modelling frameworks resolve the current gaps in UM capability to assess UES, while increasing the demands on information complexity. Linking these modelling approaches such that the benefits of each remain useful and coherent may be challenging. However, outside the UM and ES communities, efforts are already underway to link aspects of these approaches [99
]. For example, Onat, et al. [143
] have already forged a way to weave the non-linearity of system elements into a life-cycle sustainability approach.
This paper intends to mark the way towards a novel modelling approach to incorporate the assessment of urban ecosystem services (UES) in the framework of UM models. Our results showed that the UM metaphor, subject to some reorientation and increased modelling complexity, has the potential to assess provisioning and regulating UES, thus strengthening the socio-ecological capabilities of UM.
The number of both urban metabolism (UM) and urban ecosystem services (UES)-related studies have increased rapidly in recent years. However, there are still few inter- and trans-disciplinary examples in the literature exploring how they may be integrated as an UES modelling tool. Moreover, the reviewed UM literature has invariably limited scope of cross-scale integration. These state-of-the-art UM studies largely focus on ‘opening’ the UM modelling box by coupling one investigative theme at a time among temporal, spatial and multi-level factors. In this regard, no study demonstrated a holistic UM-UES approach, and while many studies assessed metabolic flows that could easily be linked to UES, the scope of those studies were not comprehensive in modelling coupled human-nature systems (CHANS) in an integrated manner. Overpassing the interactions between elements in the urban system limit the model’s ability to acknowledge and reveal complex and internal system dynamics. If too few elements are considered, the results may not be robust and this can generate burden shifting, whereby results may show positive trends in one aspect while the negative trends in another aspect may go unaccounted. Avoiding this requires a holistic, integrated and systematic modelling approach.
Considering this identified gap, we propose new pathways to address the UM paradigm with potential to support an UES assessment approach. These pathways address the key investigative themes such as spatio-temporal details and multi-level and cross-scale integrations. This model borrows the concept of CHANS as a basis from which to develop a more advanced UM method. System dynamics (SD) modelling is well suited to deal with CHANS. Additionally, our results showed that SD enabled for predictive forecasting at multiple geographic levels making use of spatially explicit information and landscape metrics. This may allow for the simultaneous incorporation of multi-level and spatio-temporal, and life-cycle information, as well as a systemic picture about the dynamic cause–effect relationships among the model variables that link the UM to its potential consequences in the supply of UES.
These enrichments to the contemporary UM framework are anticipated to better inform urban planners on the long run consequences of deploying sustainability interventions with the goal of meeting urban and global sustainability challenges. In this way a SD modelling approach could capture the spatio-temporal dynamics that so many UM reviews have highlighted as current methodological shortcomings. Incorporating life cycle thinking can also capture the metabolic flows occurring outside the urban system, such as resource extraction and end of life processes. These life cycle stages may add valuable information to the assessment of non-urban ES supplies that are, nonetheless, causally tied to the socio-economic activities within the UM system. In this way, an integrated UM-UES framework with SD and life cycle thinking may have the potential to assess and predict future UES and ES supplies linked to spatial information at multiple geographic levels.
Future research also opens new and interesting questions about the application of integrated modelling for urban planning decision support. Among others, we foresee this model quantitatively assessing the value of nature-based solutions to support sustainable urban planning. More specifically, we intend the identification of these methodological pathways to serve as a benchmark for future development of tools capable to address sustainable development goals such as sustainable cities, water, climate, human equity, and biodiversity. We expect this to aid in the design of sustainable cities and more comprehensive evaluation of urban systems quantifying the gains and losses of multiple UES supplies associated with metabolic flows.