REVIEW: Managing urban ecosystems for goods and services

Authors


Correspondence author. E-mail: k.j.gaston@exeter.ac.uk

Summary

  1. Concomitant with the rise in the proportion of the global human population that resides in urban areas has been growth in awareness of the importance of the provision of ecosystem goods and services to those people. Urban areas are themselves of significance in this regard because of their areal extent, and hence the quantity of services falling within their bounds, and because of the need for local provision of services to urban residents.
  2. Here, we review key challenges to the effective management of ecosystem goods and services within urban areas.
  3. These challenges include the structure of green space, its temporal dynamics, spatial constraint on ecosystem service flows, occurrence of novel forms of flows, large numbers of land managers, conflicting management goals, possible differences between perceptions of urban dwellers and the reality of the distribution and flow of ecosystem services, and the ‘wicked’ nature of the problem of ecosystem service management.
  4. Synthesis and applications. Urban areas present very particular combinations of challenges and opportunities for the management of ecosystem goods and services. The spatial and temporal heterogeneity of green spaces greatly complicates the maintenance and improvement in service provision as well as dramatically inflating costs. Spatial constraints on ecosystem service flows mean that these can be highly dependent on the maintenance of particular areas of connectivity, but also that provision of additional key points of connectivity may be disproportionately beneficial to those flows. The existence of novel forms of flows of ecosystem services in urban areas offers means of overcoming spatial constraints on more natural flows, but will require the development of new kinds of ecosystem process models to inform their design and management. The large numbers of land managers, conflicts between the best approaches for managing for different goods and services, and frequent differences between the perceptions of urban dwellers and the reality of urban landscapes create a complex management context. The management of ecosystem goods and services is closely allied to the challenges of conventional urban planning. However, applied ecology has a broad range of tools available to assist in determining solutions, including the use of high-resolution remote sensing techniques, landscape ecology principles and theory (e.g. patch and matrix frameworks, meta-population models), and systematic conservation planning approaches.

Introduction

The individuals of a species are seldom randomly distributed in space (Gaston 2003). Rather, they tend to be highly aggregated, with most individuals occurring in the close proximity of many others. This is true of humans, with the present centres of aggregation being the towns, cities and conurbations in which the majority of us live and work. Indeed, in much of the world, an increasing proportion of people occur in such urban foci (United Nations 2008).

Unlike the aggregated populations of the majority of other species (with obvious exceptions such as seabird and bat colonies), human urban populations typically obtain most of their ecosystem resources from sources that are distributed over a substantially larger area – although there is much variation, their ecological footprint is often one to two orders of magnitude greater than the area occupied by the population itself (Rees 1992, 1999; Wackernagel et al. 2006). Acknowledging that urban areas can make more efficient use of some resources than more dispersed populations (Bettencourt et al. 2007), a vital issue in limiting human impacts on the environment at large is how those footprints can most effectively be reduced (particularly as demand outstrips supply; e.g. McDonald et al. 2011). Potential solutions include reducing the overall demand for resources (e.g. controlling population growth, promoting more sustainable resource use), increasing their supply (e.g. increasing the flow of food and energy from within parts of the existing footprint) and increasing the intensity (decreasing the area requirement) of this supply (e.g. more intensive agriculture) (e.g. Newman & Jennings 2008; Nelson et al. 2010; Phalan et al. 2011; Sulston et al. 2012). However, a key challenge to the second two approaches is that, to date, increases in yield and intensity in modern agriculture have largely been achieved by unsustainable means; there is a global need to reduce the environmental impact of the agricultural system (Godfray et al. 2010; Sulston et al. 2012). Whilst approaches to increasing the supply and the intensity of supply of resources focus on nonurban areas, approaches to reducing the overall demand for resources focus on the urban areas themselves. This has tended to foster a belief that urban areas have little role to play in the direct provision of ecosystem goods and services, and such possibilities are commonly ignored both in discussions of the distribution of those goods and services and in global and regional accounting procedures for them (e.g. Haines-Young 2009; Harrison et al. 2010). Instead, attention is focussed on the influence of urban areas on the destruction of potential regional provision and alteration of natural regional patterns of flow.

Conversely, a growing body of research is demonstrating that urban areas can themselves be vitally important for the provision of ecosystem goods and services. This occurs in two ways. First, as urban areas increase in extent, the ecosystem goods and services provided within their bounds will inevitably constitute a growing proportion of their regional and global provision. This is particularly so given that no such areas are entirely covered by impermeable surfaces, indeed in many cities and towns green spaces contribute a significant proportion of total urban land cover (e.g. see Churkina, Brown & Keoleian 2010; Davies et al. 2011b). The global coverage of urban areas remains relatively small, with a figure of 2–3% of land (excluding permanent ice cover) commonly quoted (e.g. Millennium Ecosystem Assessment 2005). However, regional coverage may be substantially larger; figures for 165 countries vary from close to zero to 32% (World Resources Institute 2007). At least for more temperate zones, this means that urban areas can make substantial contributions to ecosystem service stocks and flows, particularly where these have been heavily depleted from nonurban areas and where urban areas tend to be developed in zones that are rich in these resources (Nowak & Crane 2002; Gaston 2005; Pataki et al. 2006; Pouyat, Yesilonis & Nowak 2006; Davies et al. 2009, 2011a,a,b; O'Neill & Abson 2009; Hutyra, Yoon & Alberti 2011). At the very least, this may often be true to the point where it is necessary to ensure the inclusion of urban areas when determining regional baselines and conducting regional accounting for ecosystem goods and services.

Second, and arguably much more significantly, urban areas can themselves be key in the local provision of ecosystem goods and services to their occupants. Here, the contribution to global stocks and flows of goods and services is of far less importance than is their spatial and temporal coincidence with these centres of human population. Examples of such local provision arise amongst each of the major groups of ecosystem goods and services: supporting (e.g. soil formation and nutrient cycling), provisioning (e.g. urban food production), regulating (e.g. local climate and flood regulation) and cultural (e.g. aesthetic, sense of place and health benefits of green space and wildlife; Davies et al. 2011a). In most cases, the actual level and pattern of provision remain to be well documented, at least beyond a few exemplar case studies. However, it is clear that existing local provision, and the improvements that can be attained, makes urban areas both practically much more functional and more pleasant places to live for their occupants. Indeed, a growing number of papers have highlighted the significance of the levels of provision and of the inequalities in their availability and/or how they are accessed by different socioeconomic groups (e.g. Jo & McPherson 2001; Hope et al. 2003; Kinzig et al. 2005; Grove et al. 2006; Barbosa et al. 2007; Tratalos et al. 2007; Alexandri & Jones 2008; Davies et al. 2011a,b, 2012; Fuller et al. 2012). The consequences of compromising the provision of particular ecosystem services in urban areas can in some cases be extreme, as evidenced for example by the rise in human mortality rates associated with urban heat island effects (e.g. Goggins et al. 2012).

Accepting that the provision of ecosystem goods and services within urban areas is an important issue, then so is their effective management; the term ‘management’ has been used in a variety of ways in the context of urban ecosystems (Jansson & Lindgren 2012), but here we employ it broadly to refer to any change that positively influences the availability and/or provision of an ecosystem good or service (and ideally multiple goods and services). Indeed, substantial public and private sums are already spent annually in many urban areas on environmental management actions, at a variety of spatial scales, that intentionally or unintentionally have impacts on ecosystem goods and services. These include the creation of green spaces (e.g. new public spaces, landscaping developments), hard and soft landscaping, managing vegetation in existing or newly created green spaces (e.g. grazing, mowing, tree planting and surgery, coppicing, growing fruit and vegetables), managing green waste (e.g. composting, wood chipping), installation of green roofs and walls, and ‘wildlife gardening’ (e.g. provision of nectar-rich plants, bird feeders, nest boxes, bat boxes, ponds) (Snep & Opdam 2010; Douglas & Ravetz 2011; Rowe 2011; Sadler et al. 2011; Hale & Sadler 2012).

There are several overarching challenges to the management of ecosystem goods and services in urban areas, including the structure of green space, its temporal dynamics, the spatial constraint on service flows, the occurrence of novel forms of flows, the large numbers of land managers, conflicting management goals, the possible differences between the perceptions of urban dwellers and the reality of urban landscapes, and the ‘wicked’ nature of the management problems. In this paper, we examine each of these challenges in turn. We draw largely on examples from the UK. In keeping with much of Europe, many UK cities have a long history of human settlement and a compact urban form, in contrast to the shorter history and sprawling urban form that typifies many cities in, for example, the USA and Australia (Gaston 2010). However, most of the key points generalise widely.

Green space structure

Although often not of itself sufficient, key to the provision of many, and perhaps the majority, of ecosystem goods and services within urban areas is the provision of green space (here used broadly to mean any area of land not covered by impermeable surface, including remnant vegetation, public parks, public and private landscaping, domestic gardens, sports and playing fields, allotments, cemeteries, nature reserves, and derelict lands). Such space can vary greatly in its overall extent within any given town, city or conurbation (Fuller & Gaston 2009; Gaston 2010). However, almost invariably this overall area is spatially distributed in complex ways. Of particular relevance to the provision of ecosystem goods and services are that the majority of green spaces tend to be small (Fig. 1) and that habitat heterogeneity is commonly high amongst spaces. In some cases, the level of fragmentation is such that cumulatively the smaller patches (albeit not the smallest) comprise a large proportion of green space (Fig. 2; Gaston et al. 2005; Fuller et al. 2010). The spatial distribution of patches of different sizes tends to be highly variable amongst different urban areas; however, large green spaces are often more peripheral, whether these urban areas are best described as monocentric or polycentric in their built structure.

Figure 1.

Patch size frequency histogram, representing the number of green spaces with a given areal extent (in km2), within Newcastle-upon-Tyne, UK, urban boundary: (a) histogram including all green space patches; (b) histogram of green spaces patch size excluding those with an area smaller than 0·001 km2; (c) histogram of all green spaces with patch sizes exceeding 0·01 km2; and (d) histogram of those green spaces exceeding 0·1 km2. Data derived from MasterMap.

Figure 2.

Cumulative contribution of different-sized green space patches to the overall areal extent of green space within Newcastle-upon-Tyne, UK. The patches of green space are ranked in order of size (smallest to largest), and the cumulative size (km2) is plotted against the rank. Data derived from MasterMap.

The habitat heterogeneity of green spaces is influenced foremost by the different uses to which they are put (see above). However, within these categories, there can remain substantial additional heterogeneity, with fine-scale patchworks of habitat types typifying some land uses, such as public parks and domestic gardens (e.g. Smith et al. 2005; Loram, Warren & Gaston 2008). This heterogeneity is such that it may only become apparent when urban green spaces are mapped at spatial resolutions that are much finer than commonly employed when studying rural landscapes, and necessitating high-resolution remote sensing data (Fuller et al. 2010). A comparison between two different areas of the city of Leicester, UK, the first displaying nondomestically owned green space dominated by a public park (Fig. 3a) and the second displaying domestically owned or rented land within the city's suburbs (Fig. 3b), illustrates the heterogeneity of green space within urban areas. For example, the cover of trees within the area of the city dominated by park was 40% of total nondomestic green space, dropping to only 13% in the domestically owned or rented green space within the city's suburbs. However, the most stark contrast between these two areas of the city was the patch size of the different vegetation land cover classes. In total, the domestic green space within the suburban area was comprised of over 6200 patches (Fig. 3b), dropping to c. 2200 in the nondomestic green space (Fig. 3a). Mean patch size ranged from 93 m2 for domestic green space to 204 m2 for nondomestic, with median values of 64 m2 and 81 m2, respectively.

Figure 3.

A comparison of the spatial heterogeneity of green space in the city of Leicester between; (a) nondomestically owned green space in an area of the city dominated by a public park and (b) domestically owned or rented green space within a suburban area of the city. Herbaceous vegetation (light green); shrubs, tall shrubs and trees (dark green); open water (blue); artificial surfaces (grey); buildings (black). Within (a) white areas are domestically owned or rented green space, and (b) white areas are nondomestically owned green space. Data derived from LandBase and MasterMap.

This size and habitat structure of green spaces present a major challenge to the management of the ecosystem goods and services that they provide. First, they make determining the levels of those goods and services difficult. Carefully structured survey programmes need to be employed to ensure adequate coverage of the full breadth of types and sizes of green spaces (e.g. Nowak et al. 2008a,b; Davies et al. 2011b). Complex techniques may often also be required to extrapolate local observations more widely. For example, whilst estimating the potential impacts of individual trees on building energy use is relatively straightforward, mapping these effects across an entire city is not simple (Fahmy & Sharples 2009). Second, the structure of green space is such that the provision of many ecosystem goods and services scales, typically positively, with patch size (e.g. Fuller et al. 2010; Su et al. 2012). This makes an understanding of these scaling relationships, many of which are markedly nonlinear, and thus the potential consequences of increasing or decreasing green space patch sizes, key to the management of those goods and services. Third, the heterogeneity of urban green spaces, with complex mosaics of habitats (Davies et al. 2011b), means that management practices are likely to have to be varied both within and between them, which can render these practices costly, with economies of scale readily being lost. Fourth, the manner by which ecosystem service provision is best maintained may vary with green space size. For example, heavy management may be most cost-effective for smaller (but probably not the smallest) patches, but it may be better to rely more fully on more natural processes in larger patches both because these processes are themselves likely to be more functional and because the scale of more interventionist approaches may be impractical. Fifth, the cumulative extent of smaller green spaces is significant because strategic urban planning, and indeed much green space planning per se, tends to focus almost exclusively on the larger patches and typically ignores the smaller ones. However, it is important to note that management to maintain existing urban green spaces should be used prudently as certain techniques may alter the ecosystem service balance in these areas; for example, the use of fossil fuel-powered machinery to prune urban trees will result in a net reduction of carbon sequestration (Nowak et al. 2002; Davies et al. 2011b).

The consequences of differentially managing green spaces of different sizes can be profound. In Table 1, in a purely illustrative exercise which could be extended to other ecosystem goods and services, we estimate the consequences for above-ground carbon storage of managing the green space stock of Newcastle-Upon-Tyne in the following ways – Model 1: all patches bigger than 0·1 km2 are converted into woodland, any smaller patch is managed as a domestic garden; Model 2: all patches bigger than 0·01 km2 are converted into woodland, any smaller patch is managed as a domestic garden; Model 3: medium size patches (>0·01 km2 and <0·1 km2) are converted into woodland, smaller patches are managed as domestic gardens, and larger patches are devoted to agriculture; and Model 4: all patches managed similarly, with 50% woodland cover and 50% herbaceous cover. Carbon storage potential can double from model 1 to model 3 and triple from model 1 to model 2, whereas an intermediate increase in storage could be achieved with model 4.

Table 1. Potential carbon stored above-ground within the urban green space (87 km2) of Newcastle-upon-Tyne urban boundary on different management scenarios, based on average above-ground carbon quantifications by Davies et al. (2011a,b) (0·76 kg C m−2 stored in domestic gardens, 0·14 kg C m−2 stored in land covered by herbaceous vegetation, and 28·86 kg C m−2 associated with tree covered land): (i) potential amount of carbon stored within the Newcastle-upon-Tyne urban boundary if all the green spaces were covered by trees; (ii) if all the green spaces were managed as domestic gardens; (iii) if all the green spaces were covered by herbaceous vegetation; (iv) Model 1: if all the large green spaces(area >0·1 km2) were tree covered and all the smaller patches were managed as domestic gardens; (v) Model 2: if all large and medium size green spaces (area >0·01 km2) are tree covered and all the smaller patches were managed as domestic gardens; (vi) Model 3: if all medium size green spaces (<0·1 km2 to >0·01 km2 area) are tree covered, smaller patches are managed as domestic gardens, and all large green spaces are devoted to agriculture (with herbaceous cover); (vii) Model 4: each greenspace is 50% tree cover and 50% herbaceous cover. Distribution of green spaces derived from MasterMap
ScenarioPotential C storage (tonnes)
All green spaces tree covered2 508 301
All green spaces managed as domestic gardens66 054
All green spaces covered by herbaceous species12 168
Model 1524 508
Model 21 477 882
Model 31 035 098
Model 41 260 234

Temporal dynamics

Urban areas are typically very dynamic and responsive to history, policy and other drivers, meaning that through time a proportion of green spaces will turn over and change in size and shape (e.g. Pauleit, Ennos & Golding 2005; Uy & Nakagoshi 2007; Kattwinkel, Biedermann & Kleyer 2011; Zhou et al. 2011; Gillespie et al. 2012). For example, Fuller et al. (2010) show how the proportion of green space in 250 × 250 m grid cells across Sheffield, UK, declines with the time (from 1860) since each cell became predominantly urbanised; once urbanised, green space in a cell has become progressively further eroded. On a shorter time horizon, Dallimer et al. (2011) document a net increase in the extent of green space in all but one of 13 UK cities between 1991 and 2006. However, this gain in the main occurred before 2001, since when green space declined in nine of the cities, following policy reforms towards greater housing densification in 2000. Likewise, across the early years of the present century, tree cover decreased significantly in 17 of 20 US cities (Nowak & Greenfield 2012).

Particularly when combined with urban expansion, such dynamics typically lead to mosaics of green space of different ages, potentially including remnants of original vegetation, heavily contaminated lands, well-established stands of new vegetation, through to areas of fresh bare ground and pioneer communities. The ages of green spaces, or of the buildings with which they are associated, can be correlated with their levels of vegetation cover (e.g. Smith et al. 2005; Kendal, Williams & Williams 2012), species richness and composition (e.g. Smith et al. 2006; Luck & Smallbone 2010), and organic carbon storage (e.g. Golubiewski 2006; Smetak, Johnson-Maynard & Lloyd 2007). This dynamism of green spaces suggests that the provision of ecosystem goods and services in urban areas may need to be viewed in something akin to one of the ‘patch’ frameworks of population ecology. By analogy, one can envisage a range of patterns of provision of goods and services, spanning the equivalents of single population, classical metapopulation, mainland-island metapopulation, patchy population and non-equilibrium metapopulation spatial structures (Harrison 1994). This will be challenging, particularly because management frameworks often do not match the spatial and temporal scale of ecosystem processes (Borgström et al. 2006). Much of the implementation of management actions will need to be resourced and conducted on the understanding that its consequences will not be realised potentially for many years or even decades.

Spatial constraint on flows

The spatial flows of ecosystem goods and services are highly constrained (canalised) across urban landscapes. That is, the built infrastructure (Fig. 4a and b) imposes major limitations on how goods and services can pass across these landscapes, whether those flows deliver the benefits locally, regionally (e.g. to rural communities or other urban ones) or beyond. It does so both at a gross level in terms of the pattern of buildings and transport networks (Fig. 4a and c) and at a finer resolution in terms of the position of, for example, individual roads and waterways (Fig. 4c and d). This constraint has perhaps been best characterised for organismal movement (Andrieu et al. 2009; Shanahan et al. 2011; Tremblay & St. Clair 2011; Hale et al. 2012; Vergnes, Le Viol & Clergeau 2012) and hence the services which wildlife provide or to which they contribute. However, it generalises much more widely. Indeed, it seems likely that the majority of ecosystem goods and services show much more spatially constrained flows than in many other environments.

Figure 4.

Spatial structure of green spaces in Newcastle-upon-Tyne, UK, and spatial structure of identified barriers to and channels for the flow of ecosystem services. (a) Green spaces exceeding 0·1 km2 (in green) and built infrastructure (in grey); (b) all green spaces within the urban area (in green) and their physical boundaries (in grey); (c) transport infrastructure (in grey: minor roads; in red: major roads; in black: railway lines); (d) surface water (rivers, streams, open canals, ponds and lakes) (in blue). Data derived from MasterMap.

From a management perspective, such spatial constraint means that, on the one hand, flows of ecosystem goods and services across urban landscapes may be especially vulnerable to the severance of particular corridors of flow. On the other hand, there may be considerable opportunities for disproportionately improving flows by the creation of new corridors. Such concerns have been widely recognised in plans to protect and develop the blue and green infrastructures of many urban areas, with riverine corridors having been a heavy focus of attention (e.g. Gledhill, James & Davies 2008; Kazmierczak & Carter 2010; Kelly, Luke & Lima 2011). A persistent issue, however, has been that such planning tends also to concentrate, sometimes exclusively, on the role of large unified ‘green/blue corridors’ through urban landscapes, and to ignore the potential roles of the huge numbers of small patches of vegetation (and perhaps water). This matrix within which the larger corridors are set will often in practice be of great (and perhaps greater) significance to the flows of ecosystem goods and services, which may exhibit patterns of constraint across this matrix that need critically to be identified and appropriately managed.

Key to planning for the management of urban ecosystem services will be the use of spatial planning tools that prioritise areas in terms of their realised or potential importance for spatial flows and that can be used to explore the consequences of different planning decisions. Although the general principles have long been promoted (McHarg 1992), lessons need to be learnt from those tools that have been developed for biodiversity conservation, which faces many closely related issues, and which can in some cases be adapted to this end (e.g. Margules & Pressey 2000; Moilanen et al. 2012; Moilanen, Wilson & Possingham 2009; for application to ecosystem services see Moilanen et al. 2011); the use of these tools for species conservation in urban areas has already been highlighted (e.g. Gordon et al. 2009).

Novel flows

As well as often being strongly spatially constrained, ecosystem goods and services can also flow across urban landscapes in ways that do not occur, or are typically much less significant, in other landscape types. In particular, humans directly move material around urban landscapes on a massive scale. This includes soil (e.g. for landscaping), vegetation (e.g. through waste collection systems and urban horticulture) and water (e.g. through drainage systems). For example, in a survey of 575 residents in the city of Leicester, UK (Living in Leicester; Lomas et al. 2010), respondents reported annually removing in total the equivalent of nearly 6000 bin bags of green waste from their gardens. Nearly 30% of this waste was hedge and shrub clippings and tree prunings; this can be converted to c. 630 kg organic carbon removed from the urban green space (estimated using analysis of organic carbon concentration and measurement of the dry weight of full bin bags of seven common urban tree species and 11 common urban shrub species). To contextualise the scale of movement of waste across the city, the total garden area of survey respondents was c. 8 ha, which was <1% of total garden area in Leicester. Removal from gardens was greatly in excess of the flow of organic matter into gardens, with respondents reporting annually adding the equivalent of an estimated 1200 bin bags of material to their gardens (including manure, commercial and own produced compost, bark and tree chippings, straw and topsoil). However, this represents an addition of organic carbon to urban gardens above the natural inputs from vegetation.

Such novel flows often remain rather poorly characterised and thus constitute major unknowns in documenting the dynamics of ecosystem goods and services in urban areas and formalising these in ecosystem models. Nonetheless, they do offer potential avenues for managing those goods and services that may be easier to alter than might be the case with more natural patterns of flow.

Multiple managers

Urban ecosystems are unlike almost any others in the exceedingly large numbers of land managers present. Whilst substantial tracts of land may be under the control of a few governmental or nongovernmental organisations (private individuals, companies or charities), typically much is divided amongst a vastly greater number of individual homeowners or tenants, each of whom may be responsible for an area of the order of tens to hundreds of square metres. For example, in Leicester, green space covers 56% of the urban area (73 km2), 80% of which is privately managed. Green space associated with c. 123 000 households, and therefore individual managers, within the city constitutes 40% of the total, with a further 40% of private land managed nondomestically (Fig. 5). A smaller proportion of the city's green space is owned and managed by the local authority (20%); however, much of this land is held in larger patches such as urban parks.

Figure 5.

The distribution of residential and nonresidential spaces in Leicester, UK; nonresidential green space (light green), nonresidential artificial surface (pale grey), residential artificial surface (dark grey), residential green space (dark green), buildings (white) and inland water (blue). Distribution derived from the MasterMap and LandBase GIS data sets.

Most obviously, the large numbers of land owners and tenants in urban areas make coordination of management activities extremely difficult. Some level of control can be exerted through legislation; however, this provides a rather blunt instrument. The converse argument is that because of the large numbers of land managers, only a relatively small proportion may need to carry out a particular management action for a wide impact to result. For example, the establishment of new ponds in just 10% of the domestic gardens in the urban area of Sheffield would result in the addition of 17 500 such habitat patches (albeit typically small ones), at a density of c. 120 per km2 (Gaston et al. 2005). In Leicester, the urban green space managed by 575 survey respondents (Living in Leicester survey; Lomas et al. 2010) was c. 8 ha in extent (<1% citywide garden area), of which nearly 50% was covered by herbaceous vegetation. If 10% of this herbaceous area was converted to land dedicated to own-grown food, potential yield could exceed 10 tonnes per annum (based on UK agricultural potato yields and unpublished data from the Royal Horticultural Society allotment yield trials in 1974; Tompkins 2006; Supit et al. 2010). A further 20% of the respondents' gardens were capped by artificial surfaces (e.g. driveways, patios and footpaths; all common features of UK gardens; Loram, Warren & Gaston 2008). Excavation of surface soil is routine during the construction of artificial surface; on average, the top 15 cm of soil is lost in domestic gardens and consequently 6·7 kg m-2 organic carbon (Edmondson et al. 2012). If 10% of the artificial surface in these domestic gardens was removed and converted to lawn, soil organic carbon storage, to 1 m depth, could increase by 13 tonnes, from 26 to 39 tonnes (data from Edmondson et al. 2012). Only 20% of the same respondents used a compost heap, with a smaller number adding household fruit and vegetable waste, and any increase in the number of people using this method could significantly reduce the resource demand associated with waste management in our urban areas (Gaston et al. 2005).

Other challenges that result from the large numbers of managers include that (i) lots of managers are not managing for ecosystem service provision but for alternative, and often conflicting, goals; (ii) different groups of managers may have different perceptions as to what changes are most desirable (e.g. Hofmann et al. 2012); (iii) there is a loss of benefits of scale of management costs, both financial and environmental (e.g. need for physical tools for management, and the gas emissions that those tools give rise to); and (iv) universal management actions to target-specific environmental problems across whole urban ecosystems, such as maximisation of green space flood mitigation or organic carbon storage potential, are difficult to achieve.

Conflicting management goals

Inevitably, different ecosystem goods and services in urban areas require different management approaches, and in some cases, these will conflict. Thus, for example, (i) increasing carbon sequestration and reducing summer temperatures will often involve retaining or planting trees, but this may result in an increase in emissions of biogenic volatile organic compounds which are hazardous to human health (Leung et al. 2010); and (ii) increasing urban food production will reduce land available for growing trees, and may involve use of chemicals that impact on water quality. Given the large number and small size of many green spaces, such conflicts are probably best handled by managing for different goods and services in different patches, rather than in different parts of the same patch. Unfortunately, this may sometimes be at odds with the goals of the owners or tenants of individual green space patches. For example, in the UK, urban gardens are often managed so as to maintain high levels of habitat heterogeneity, such that although the number of habitat features increases with garden size, many features are present in a high proportion of gardens with their extent scaling strongly with garden size (Smith et al. 2005; Loram, Warren & Gaston 2008).

Perception and reality

Throughout, we have focussed on the challenges for the management of ecosystem goods and services in urban areas posed by their actual distribution in space and time. Under some circumstances, there may also be important challenges that result from differences between their actual distributions and flows, and those that urban dwellers perceive to prevail. For example, Dallimer et al. (2012) have shown that whilst there are no consistent relationships between the psychological well-being of urban green space visitors and the species richness of various groups of organisms, there are positive relationships with the richness that those users perceive to be present. This is potentially highly problematic, as management for actual levels of biodiversity may well conflict with management that would enhance perceived levels of biodiversity.

Appreciation of the value of an ecosystem service may also vary with the role and perception of the stakeholder (i.e. beneficiary, manager, policymaker) and the ecosystem service considered. For example, beneficiaries may place high value on local resources (e.g. parks, allotments) that in broader planning terms are insignificant. Furthermore, the value of the ecosystem service can be expressed in different ways (e.g. economic, ecological or social value), and these valuations are often not comparable (de Groot et al. 2010).

Wicked problems

From a more over-arching perspective, The Royal Commission on Environmental Pollution (2007) argues that urban environmental management presents a classic case of a ‘wicked problem’; ‘wicked’ in the sense of nasty or vicious, rather than an ethical judgement. The maintenance of and improvement in the provision of ecosystem goods and services is increasingly seen as a substantive component of this management, and the same conclusion might justifiably be reached in this regard. The notion of wicked problems derives from a treatise by Rittel & Webber (1973), who observed that the kinds of societal problems that planners deal with are ill-defined and cannot be definitively solved and thus are intrinsically different from archetypal problems in science. Their characterisation of such problems, which derive in large part from ‘the interdependencies and complexities of living together without a shared set of values and views’ (Roberts 2000), is rephrased here in the context of the management of ecosystem goods and services in urban systems:

  1. There is no definitive formulation of an ecosystem service management problem – The process of describing the problem, say the need to improve urban food production, and of solving it are essentially the same. This can serve to fuel disagreement as to what the ‘problem’ actually is and lead to a framing of the problem in a manner that more readily connects it with the solution preferred by a particular stakeholder (Roberts 2000).
  2. Ecosystem management problems have no stopping rule – Because there is no definitive formulation of the problem, there is no point at which the solution has been found. There is, for example, no point at which an improvement in urban climate regulation would definitively be sufficient.
  3. Solutions to ecosystem service management problems are not true-or-false, but good-or-bad – The nature of a particular solution is likely to depend on who provides it, with, for example, local residents and regional government likely to manage a given green space in a different way (the former tending to focus on their own needs, the latter on standardising practices across a regional portfolio of green spaces).
  4. There is no immediate and no ultimate test of a solution to an ecosystem service management problem – A given solution is likely to have many consequences, some unintended and unexpected, and these may play out over long periods. This highlights a particular need for multiple studies of the outcomes of ecosystem service management actions in urban areas.
  5. Every solution to an ecosystem service management problem is a ‘one-shot operation’; because there is no opportunity to learn by trial-and-error, every attempt counts significantly – This is because management actions are seldom entirely reversible but have many consequences. It is impossible, for example, accurately to predict all the ecosystem service consequences of establishing a new green space in an urban area, and once established it would be impossible simply to reverse some of those consequences.
  6. Ecosystem management problems do not have an enumerable (or an exhaustively describable) set of potential solutions, nor is there a well-described set of permissible operations that may be incorporated into the solutions – The set of potential solutions and the extent to which they are permissible will depend on who provides them. Some have argued that in a complex world, what is required are ‘clumsy solutions’, which combine alternative (and sometimes conflicting) ways of perceiving and organising answers (Verweij et al. 2006).
  7. Every ecosystem service management problem is essentially unique – There are always particularities to a problem that may override the commonalities with other problems. Thus, for example, the detail of approaches to improving the human–wildlife interactions that benefit people's well-being will vary amongst cities, amongst communities within a city, and at different times for any given community.
  8. Each ecosystem service management problem can be considered a symptom of another problem – For example, poor local climate regulation might follow from a lack of carbon storage and sequestration, as a consequence of poor management of vegetation cover.
  9. The existence of a discrepancy representing an ecosystem service management problem can be explained in numerous ways. The choice of explanation determines the nature of the problem's resolution.
  10. The ecosystem service manager has no right to be wrong – The objective is to improve a situation, and there are plenty of tools available to assist a manager to that end.

Roberts (2000) distinguishes three generic strategies for identifying wicked problems and their solutions. If power is concentrated amongst a small number of stakeholders, or it is placed in their hands by consent, then authoritative strategies can be employed. This can greatly reduce the complexity of the process, but can also result in problems being too narrowly or incorrectly characterised, and in other parties being ill-informed and unengaged. If power is dispersed and contested, then competitive strategies can be employed, encouraging different sets of stakeholders to garner sufficient power to define the wicked problem and its solution. This has the advantage of encouraging innovation, but can result in protracted, distracting and costly battles for that power, which in the extreme can prevent any practical progress being achieved. Finally, if power is dispersed but not contested, then collaborative strategies can be used to define the problem and solution. This can be efficient, in spreading costs, and increasing ‘weight of numbers’ and the breadth of ‘solution space’ that can be explored. However, effective collaboration can be challenging to achieve, and costly in the effort required. Across the breadth of wicked problems posed by ecosystem service management in urban areas, no one of these three strategies will be adequate in itself. For some issues, power is highly concentrated (e.g. water flows), for others it is highly dispersed (e.g. vegetation management).

Conclusion

The local provision of ecosystem goods and services in urban areas is essential to the populations that benefit from them and will help reduce the regional and global footprint of cities and towns. The management of these goods and services poses a number of substantive challenges, including the structure of green space, its temporal dynamics, the spatial constraint on ecosystem service flows, the occurrence of novel forms of those flows, the large numbers of land managers, conflicting management goals, the possible differences between the perceptions of urban dwellers as to the distribution and flow of ecosystem services and the reality of that distribution and flow, and the ‘wicked’ nature of ecosystem service management in urban landscapes. However, there is also clearly a broad range of tools available from applied ecology to assist in their resolution. These include the use of high-resolution remote sensing techniques, landscape ecology principles and theory (e.g. patch and matrix frameworks, meta-population models) and systematic conservation planning approaches. These will need to be employed within a broader transdisciplinary framework (Ervin et al. 2012) to address the interactions between natural and human systems that are arguably at their most complex in urban ecosystems.

Acknowledgements

This work was supported by EPSRC grant EP/I002154/1 SECURE: Self conserving urban environments (a consortium of Loughborough Univ., Newcastle Univ., Univ. Exeter and Univ. Sheffield); EPSRC grant EP/F007604/1 4M: Measurement, Modelling, Mapping and Management (a consortium of Loughborough Univ., De Montfort Univ., Newcastle Univ., Univ. Sheffield and Univ. Exeter); and NERC grant NE/J015237/1 Fragments, functions and flows (a consortium of British Trust for Ornithology, Cranfield Univ., Univ. Exeter and Univ. Sheffield). Infoterra provided access to LandBase; MasterMap data were supplied by Ordnance Survey. We are grateful to S. Gaston, Z. Grabowski, J. Jones and an anonymous reviewer for comments.

Ancillary