A historic overview of the development of the Ecosystem Services (ES) concept in a few pages is almost impossible and unavoidably biased and, for this chapter, we focused on the main events and publications. Some key publications are listed at the end of this chapter as suggestions for further reading.

Most authors agree that the term "ecosystem services" was coined in 1981. It was pushed to the background in the 1980s by the sustainable development debate but came back strongly in the 1990s with the mainstreaming of ES in professional literature and with an increased attention to their economic value.

Over time, the definitions of the concept have evolved with a focus on either the ecological basis as ES being the conditions and processes through which natural ecosystems and their species sustain and fulfil human life or at the level of economic importance, where ES are the benefits humans derive, directly or indirectly, from ecosystem functions. As a compromise, the TEEB (The Economics of Ecosystems and Biodiversity) study (2008-2010) defined ES as the direct and indirect contributions of ecosystems to human well-being. Despite these differences, all definitions stress the link between (natural) ecosystems and human wellbeing (see Figure 1) and the services are the 'bridge' between the human world and the natural world, with only humans being virtually separated from that natural world.

Figure 1.

Dependence of Human Wellbeing on Natural, Social, Built and Human capital. Source: Costanza et al. 2014.

The term ecosystem function was originally used by ecologists to refer to the set of ecosystem processes operating within an ecological system. In the late 1960s and early 1970s, some authors started using the term "functions of nature" to describe the 'work' done by ecological processes, the space provided and goods delivered to human societies.

When describing the flow of ES from nature to society, the need to distinguish 'functions' from the fundamental ecological structures and processes was emphasised to highlight that ecosystem functions are the basis for the delivery of a service. Services are actually conceptualisations ('labels') of the "useful things" ecosystems "do" for people that provide direct or indirect benefits.

In the late 1960s and early 1970s, a wave of publications was produced which addressed the notion of the usefulness of nature for society, other than being an object to conserve based on ethical concerns. Terms such as functions of nature, amenity and spiritual value were used in addition to, but not replacing, intrinsic values of nature, emphasising the importance to cultural identity, livelihood and other non-material benefits.

This expanding field, recognising the dependence of people on nature, finally led to the coining of the term "ecosystem services" in the early 1980s.

The ways nature provides benefits to humans are discussed throughout economic history from the classical economics period to the consolidation of neo-classical economics and economic sub-disciplines specialised in environmental issues. Some of the classical economists explicitly recognised the contribution of nature rendered by 'natural agents' or 'natural forces'. However, although they recognised their value in use, they generally denied nature's services role in exchange value, because they were considered as free, non-appropriable gifts of nature. The physiocrat's belief that land was the primary source of value was followed by the classical economist's view of labour as the major force behind the production of wealth.

Marx considered value to emerge from the combination of labour and nature: "Labour is not the source of all wealth. Nature is just as much the source of use values (and it is surely of such that material wealth consists!) as labour, which itself is only the manifestation of a force of nature".

In the 19th century, industrial growth, technological development and capital accumulation led to changes in economic thinking that caused nature to lose importance in economic analysis. By the second half of the 20^th century, land or more generally environmental resources, completely disappeared from the production function and the shift from land and other natural inputs to capital and labour alone and from physical to monetary and more aggregated measures of capital, was completed. In the second half of the 20^th century, environmental problems became a topic of interest to some economists who founded the Association for Environmental and Resource Economists in 1979. The undervaluation in public and business decision-making of the contributions by ecosystems to welfare was partly explained by the fact that they were not adequately quantified in terms comparable with economic services and manufactured capital.

From the perspective of environmental economics, non-marketed ecosystem services are viewed as positive externalities that, if valued in monetary terms, can be more explicitly incorporated in economic decision-making. In 1989, the Society for Ecological Economics was founded which conceptualises the economic system as an open sub-system of the ecosphere exchanging energy, materials and waste flows with the social and ecological systems with which it co-evolves. The focus of neo-classical economists on market-driven efficiency is expanded with issues of equity and scale in relation to biophysical limits and to the physical and social costs involved in economic performance using monetary along with biophysical accounts and other non-monetary valuation languages.

Neo-classical and ecological economists differ markedly regarding their approach to the sustainability concept. The so-called "weak sustainability" approach, which assumes the ability to substitute between natural and manufactured capital, is typical for neo-classical environmental economists. Ecological economists generally embrace the so-called "strong sustainability" approach, which maintains that natural capital and manufactured capital are in a relation of complementarity rather than of one of substitutability. They also differ with respect to approaches to ES valuation. Monetary valuation, costs versus benefits, of marketed goods and services have been primary in neo-classical approaches, while ecological economists tend to show more interest in inclusion of non-monetary and non-market goods and services approaches.

In the 1970s and 1980s, ecological concerns were framed in economic terms to stress societal dependence on natural ecosystems and raise public interest for biodiversity conservation. Already in the 1970s, the concept of 'natural capital' was used and shortly thereafter several authors started referring to "ecosystem (or ecological, or environmental, or natural) services". The rationale behind the ecosystem service concept was to demonstrate how the disappearance of biodiversity directly affects ecosystem functions that underpin critical services for human well-being. The 1997 calculation of the total value of the global natural capital and ES was a milestone in the mainstreaming of ES. The Millennium Ecosystem Assessment (2005) constitutes another milestone that firmly placed the ES concept on the policy agenda.

The TEEB study (2010), building on this initiative, has added a clear economic connotation. The interest of policy makers has turned to the design of market-based instruments to create economic incentives for conservation (see Chapter 4.3), e.g.

Although one has to be careful that the concept is not misused, the benefits of greater awareness of the full spectrum of values of nature outweigh the risk and with the adoption of the Aichi-targets (see below) at the CBD convention and the creation of the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES in 2012) as described below the ES-concept has been firmly placed on the political agenda. Especially CBD-Aichi Biodiversity Targets 1 and 2 are relevant: Target 1, "by 2020, at the latest, people are aware of the values of biodiversity and the steps they can take to conserve and use it sustainably" and Target 2, "by 2020, at the latest, biodiversity values have been integrated into national and local development and poverty reduction strategies and planning processes and are being incorporated into national accounting, as appropriate, and reporting systems". The efforts to achieve these targets, in Europe coordinated by the Mapping and Assessment of Ecosystems and their Services (MAES) contribute much to greater awareness of the many benefits of nature and help to give them more weight in everyday decision-making (see Chapter 7.1). Recently, the business-world is also waking up to the 'ecosystem services-movement' and created the Natural Capital Coalition to better account for ES and biodiversity conservation in their business models.

Although much has been achieved, even more remains to be done to further develop the ES 'science' and embed the concept in everyday policy and practice to enhance nature conservation and sustainable use of ES which is the main objective of the Ecosystem Services Partnership (ESP), founded in 2008.

Braat LC, de Groot RS (2012) The ecosystem services agenda: bridging the worlds of natural science and economics, conservation and development and public and private policy'. Ecosystem Services 1: 4-15.
Costanza R, d'Arge R, de Groot RS, Farber S, Grasso M, Hannon B, Limburg K, Naeem S, O'Neill R, Paruelo J, Raskin RG, Sutton P, van den Belt M (1997) The Value of the World's Ecosystem Services and Natural Capital. Nature 387: 253-260.
Costanza R, de Groot RS, Sutton P, van der Ploeg S, Anderson SJ, Kubiszewski I, Farber S, Turner RK (2014) Changes in the global value of ecosystem services. Global Environmental Change 26: 152-158.
Daily G (Ed.) (1997) Nature's Services. Societal Dependence on Natural Ecosystems. Island Press, Washington, D.C., 412 pp.
Gomez-Baggethun E, de Groot R, Lomas PL, Montes C (2010) The history of ecosystem services in economic theory and practice: from early notions to markets and payment schemes. Ecological Economics 69: 1209-1218.
Potschin M, Haynes-Young R, Fish R, Turner RK (Eds.) (2016) Routledge Handbook of Ecosystem Services. Routledge, T&F Group, 640 pp.

Formally, the natural base for ecosystem services (ES) arises from the performance of the living and non-living components of an ecosystem and the interrelations between them. The respective ecosystems can be characterised as a result of their structural features, their functional attributes and their organisational properties. While the latter items demonstrate the overall schemes of ecological interactions, the self-organising processes and the whole system's dynamics, the functional viewpoint highlights the flows and pools of energy, water, matter and information.

The structural aspect of ecosystems is related to the spatio-temporal characteristics of the biotic and abiotic elements. The focal features of this viewpoint are the components of biodiversity, which play a significant role for the support of ES. The 2020 targets of the Biodiversity Strategy are focussing on two perspectives: the 'intrinsic value' of biodiversity and the 'life insurance value' essential for ES supply (see Chapter 5.1). In the following pages, the second perspective will be discussed by examining the cross-correlations between biodiversity, ecological integrity, ecosystem functions and ES.

Ecosystems and society are closely connected within a Social-Ecological-System (SES) (Chapter 2.3). The flow from the ecosystem towards society is generated through the supply of ES. The flow back into the system is society's influence on the ecosystem generated by drivers and governance. Each step within the system is related to biodiversity, which is the total stock or the living part of our natural capital. It determines the self-regulating capacity of the system and the attitudes of biodiversity dynamics, such as resilience or adaptability.

Within the system, specific ecological functions are essential to support and supply a specific ES: for example, primary production and pollination for food production, water infiltration capacity for water provision and organic decomposition for soil fertility. These specific functions depend upon a specific part of biodiversity and often, increasing biodiversity will optimise these functions.

Based on supply and demand, the final ES is generated, e.g. as a yield of food or wood, or a direct use of green infrastructure. Based on the benefits of a service, people will eventually value the components of biodiversity. This can be an ethical or 'intrinsic value', but also a cultural or instrumental value.

To complete the circle, the societal impact and the governance flow can be adjusted, which is based upon a biodiversity strategy. Here targets are formulated and adjusted on different scales. In line with these objectives, management plans will be developed and implemented and indicators will be chosen to measure the trend and to control the distance to the target.

Biodiversity as a whole is the 'living' part of the natural capital. It is our main capacity to generate ES and to ensure adaptation to environmental changes. Figure 1 shows the essential components of the natural capital and the connection with ES and nature conservation. To characterise biodiversity aspects, each of the four organisation levels (gene, species, ecosystem, landscape) should be represented. All levels can be studied from different perspectives: the first perspective is 'composition' or the presence or absence of a specific property, such as a specific genetic allele, a rare species or a historical landscape. Also for cultural ES, such as ecotourism, the presence or absence of specific or charismatic species or landscapes is crucial. The second perspective is 'diversity of functions'. This part focuses on indicators for specific ecosystem functions such as predation, photosynthesis, carbon flows, or nutrient cycling. This part of biodiversity is important for the supply of many regulating ES and for the adaptive capacity to environmental changes and perturbations. The third perspective is 'structural diversity': how fragmented is the landscape, how many vegetation layers has a lake or a forest? The landscape patterns or vegetation structures are part of the way people perceive nature and this is closely related to cultural ES such as the maintenance of historical landscapes or unobstructed views. The degree of fragmentation and connectivity in a landscape are also crucial for the migration capacity of species and their adaptive capacity to climate change. The fourth and last perspective is 'stock', a prerequisite to harvest a provisioning ES, but also to most other ES.

Figure 1.

Four complementary perspectives of biodiversity, applicable to four organisation levels (gene, species, ecosystem & landscape).

To observe the dynamics of these biodiversity components, several indicator approaches are utilised. In most regions there is a dominance of 'composition' indicators linked with the nature conservation strategy while indicators for diversity of functions, connectivity or vegetation structure are rarely developed.

Understanding how key ecosystem functions determine ES supply, how it depends on biodiversity and understanding the effects of shortcutting these functions by technological variants is crucial in the search for nature-based solutions. The basic interrelations between these components are sketched in Figure 2. In the lower box, basic ecosystem elements and relations are depicted. In this work, biodiversity structures are perceived as biotic processors which perform active life processes and which can be distinguished, e.g. due to their roles in food webs. On the other hand, the abiotic processors, such as features of soil, geomorphology or climate, are creating and degrading concentration gradients and determining the living conditions of the biota. Both are linked by ecosystemic process bundles that are the dynamics or pools and flows of energy, carbon, water and nutrients. All of these elements are operating in complex, self-organised interaction schemes.

Figure 2.

Diagram sketching the relations between ecological structures and processes (self-organised ecosystem interactions), exemplary ecosystem functions and ecosystem services. The interrelations are also described in the following Chapter 2.3.

Their characteristics can be aggregated into different groups of functional outcomes. To assess the overall state of these complex schemes, aggregated indicators such as ecosystem integrity or ecosystem health are developed. For instance, the indication of ecosystem integrity is based on an accessible number of structural items of biodiversity and ecosystem heterogeneity, combined with the functional items representative for the energy balance, the water balance and the matter balance of ecosystems.

The aggregation of functional units can also be made to represent specific ES. For example, photosynthesis leads to the fixation of CO₂ which is influenced by the static abiotic site conditions, the dynamics of solar radiation, rainfall, evapo-transpiration or air temperature, but also by the nutrient and water provision and the state of competition with other plants. The result is an increase in phytomass and, on a longer time scale, an input of litter into the soil subsystem, where the carbon can be transferred and sequestered into long-term stable humic compounds.

These process sequences are interpreted as a functional subsystem, e.g. as carbon sequestration. These subsystems are illustrated by the middle box in Figure 2. They connect the system with a potential ES supply (see Chapter 5.1). Normatively it is only recognised as a service delivery if there is a human benefit related to its performance (see Chapter 2.3). In our example, the ability of ecosystems to fix carbon from the atmosphere becomes a service because this process can be helpful in mitigating elevated CO₂ concentrations in the atmosphere which are responsible for global temperature rises. Therefore the described process sequence is a basic component of the regulating ES global climate regulation. The production of this service emerges from a complex sequence of interrelated processes, which in turn is influenced by all self-organised ecosystem interactions illustrated in Figure 2.

Such connections are also responsible for most provisioning ES, because the primary and secondary production functions are strongly linked to the sequestration sequence. Also the regulation of nutrient budgets depends on the cycling and accumulating activities of the biotic system components, as well as the potential of the abiotic sphere to physically or chemically retain nutrients within the soil matrix. As a result of these process sequences, the seepage water is filtered and can be used for human purposes, e.g. as drinking water. Finally, cultural ES also depend on ecological interactions, because resulting ecosystem functions provide the basic preconditions to create and maintain certain structural conditions which human beings perceive as attractive phenomena.

As a result, we can observe very complex interrelations between ecosystem functions and ES. Some key functions and structures for 16 ES are listed in Table 1. A steering variable is the direct driver of a service, e.g. primary production for wood production. A supporting variable creates important boundary conditions, e.g. pollination and pest control for crop production. Most ecosystem functions serve various ES.

Table 1.Download as CSV 

Representation of ecosystem functions and structures steering (?) or supporting (?) an ecosystem service or a biodiversity target linked with intrinsic valuation. White fields demonstrate indirect effects.

Essential functions or structures for the supply of a service

Food

Wood production

Production energy crops

Venison

Water production

Pollination

Pest control

Preserving soil fertility

Flood control

Coastal Protection

Global climate regulation

Nutient regulation

Water regulation

Regulation air quality

Noise remidiation

Control erosion risk

Green space outdoor activities

Natura 2000

Green infrastructure

Provisioning ES

Cultural ES

Nature conservation

Primary production

Animal production

Soil formation

Nutrient availability / -cycling

Decomposition of organic material

Carbon storage

Conservation carbon stock

Storage rain water (infiltration capacity)

Ground water retention

Storage river water

River Drainage

Combating soil loss

Pollination

Pest control

Prevent disease

Air purification capacity

Scattering and absorption sound

Buffering coastal storms

Regulate population dynamics

Regulating ecosystem dynamics, succession

Stability ecosystem processes

Ecosystem resilience

Development of complex ecological networks

Develop ecosystem diversity / habitat quality

But what is the role of biodiversity for each of those functions? Many experimental studies demonstrate that an increase in the variety of genes or species contributes to the optimisation of one of the functions. Sowing a grassland ecosystem with more species will, for example, generate a higher biomass. For wood biomass usually a positive diversity-production relationship is found, as a result of synergies between species and a better utilisation of resources, although some combinations create a negative effect due to competition. The fact that many functions are optimised by a higher biodiversity also means that a loss of diversity will generate a suboptimal function, often compensated by human inputs of energy, materials or technology (Chapter 5.1). It is a reality that technical compensation can lead to a disintegration of ES potentials and biodiversity in land use. For example, the correlation between species numbers and productivities is broken by the additional inputs of energy, manpower, fertilisers or pesticides. Thus, today, modern agriculture produces the highest biomass under conditions of (optimally) single-species monocultures.

Each ES can be delivered in a gradient from naturally to technologically based solutions. Nature-based solutions depend more on biodiversity, generate a lower impact on surrounding ecosystems and guarantee a lower impact on other ES and a more sustainable use of the service itself. The use of a service is always a balance between supply and demand. In highly populated areas, for most ES the current demand is much higher than the supply. The excessive demand, together with a high drive for more human control, has affected and transformed most natural ecosystems towards the technological side of the gradient, in order to maximise a single service. The supporting and regulating role of biodiversity is systematically replaced by technological inputs, energy inputs, chemical inputs and management. This is true for nearly all provisioning ES, but also for most regulating and cultural ES. The challenge is to optimise the total supply of a bundle of ES, ensuring ES delivery and maintaining ecosystem functioning in the long term. Relying on more nature-based solutions will increase positive and decrease negative interactions.

All relationships in social-ecological-systems are driven by different aspects of biodiversity. All these interactions should be analysed in order to set up biodiversity strategies.

The creation of ES is founded on very complex schemes of ecological interactions with very high mutual interdependencies.

Understanding how key functions determine ES supply and how they depend on biodiversity and understanding the effect of short-cutting these functions by technological variants, is crucial in the search for nature-based solutions.

Moving towards more nature-based solutions of ES supply, generates positive effects for both biodiversity and the sustainable supply of ES bundles.

Cardinale BJ et al. (2012) Biodiversity Loss and Its Impact on Humanity. Nature 486(7401): 59-67.
Haines-Young R, Potschin MP (2010) The links between biodiversity, ecosystem services and human well-being. In: Raffaelli D, Frid C (Eds.) : Ecosystem Ecology: A New Synthesis. BES Ecological Reviews Series, CUP, Cambridge, 110-139.
Kandziora M, Burkhard B, Muller F (2013) Interactions of Ecosystem Properties, Ecosystem Integrity and Ecosystem Service Indicators - A Theoretical Matrix Exercise. Ecological Indicators 28 (SI): 54-78.
Mace GM, Norris K, Fitter AH (2012) Biodiversity and Ecosystem Services: A Multilayered Relationship. Trends in Ecology & Evolution 27(1): 19-26.
Morin X, Fahse L, Scherer-Lorenzen L, Bugmann H (2011) Tree Species Richness Promotes Productivity in Temperate Forests through Strong Complementarity between Species. Ecology Letters 14(12): 1211-19.
Noss R F (1990) Indicators for monitoring biodiversity - a hierarchical approach. Conservation Biology 4: 355-364.
Schneiders A, Van Landuyt W, Van Reeth W, Van Daele T (2012) Biodiversity and Ecosystem Services: Complementary Approaches for Ecosystem Management? Ecological Indicators 21: 123-33.

Although people have always depended on nature, in modern societies it is easy to lose sight of the fact that we still do. Indeed, many have argued that our failure to recognise the value of nature and especially the contribution that biodiversity makes to our well-being, explains much of our damaging behaviour towards the environment. It is against this background that the concept of ecosystem services (ES) is so important as it highlights the ways in which people and nature are connected.

The links between people and nature are, however, complex and so it is hardly surprising that people have defined ES in different ways. Some think of ES as the benefits that nature provides to people, like security and the basic material we need for a good life. Others view ES as the contributions that the ecosystem makes to such things. These differences in definition are explored in more detail in Chapter 2.4. For the moment it is sufficient to note that despite differences in the way ES are defined, most commentators agree that there is some kind of 'pathway' that goes from ecological structures and processes at one end through to the well-being of people at the other (Figure 1). This idea can be represented in terms of what we call the 'cascade model'. It is a way of expanding thinking about ecosystems to include people and, as such, it might be described as a 'socio-ecological system'. Finding out how these socio-ecological systems work and how we can act to sustain them are core issues in the field of ecosystem services. The task not only involves the study of ecology, but also such things as social practices, governance and institutional structures, technology and, most importantly, the things people value.

Figure 1.

The cascade model. Credit: Haines-Young and Potschin.

Note: 'CICES' in Figure 1 is the Common International Classification of Ecosystem Services, it is described in more detail in Chapter 2.4; it is a way of categorising and describing the final services that sit at the interface of nature and society.

To understand how socio-ecological systems work, it is useful to 'unpack' the cascade model to see the inter-relationships between the elements. Ecosystem services are at the centre of the cascade model which seeks to show how the biophysical elements of the socio-ecological system are connected to the socio-economic ones; ES are at the interface between people and nature.

The 'ecosystem' is represented by the ecological structures and processes to the far left of the diagram. Often we simply use some label for a habitat type, such as woodland or grassland (Chapter 3.5), as a catch-all to denote this box, but there is no reason why we cannot also refer to ecological processes, such as 'primary productivity' as something that can also occupy this part of the diagram (Chapter 2.2). In either case, given the complexity of most ecosystems, when we want to start to understand how they benefit people, then it is helpful to start by identifying those properties and characteristics of the system that are potentially useful to people. This is where the idea of a 'function' enters into the discussion. In terms of the cascade model, these are taken to be the 'subset' characteristics or behaviours that an ecosystem has that determines or 'underpins' its capacity to deliver an ecosystem service. Some people call these underpinning elements 'supporting' and 'intermediate' services, depending on how closely connected they are to the final service outputs; we believe, however, this terminology deflects attention away from the important characteristics and behaviours of an ecosystem that generate different services. Thus using our terminology for one of the examples in Figure 1, the primary productivity of a woodland (i.e. an ecological structure) generates a standing crop of biomass (i.e. a functional characteristic of the woodland), parts of which can be harvested (as a 'provisioning' service).

In the cascade, it is envisaged that services contribute to human well-being through the benefits that they support; for example by improving the health and safety of people or by securing their livelihoods. Services are therefore the various ecosystem stocks and flows (Chapter 5.1) that directly contribute to some kind of benefit through human agency. The difference between a service and a benefit in the cascade model is that benefits are the things that people assign value to; they are therefore synonymous with 'goods' and 'products'. The cascade model suggests that it is on the basis of changes in the values of the benefits that people make judgements about the kinds of intervention they might make to protect or enhance the supply of ES; this is indicated by the feedback arrow at the base of the diagram. The importance of 'values' is that they can be expressed in many ways; for example, alongside monetary values, people can express the importance they attach to the benefits using moral, aesthetic and spiritual criteria (Chapter 4).

Despite the simplicity of the cascade model, it is useful in highlighting a defining characteristic of an ecosystem service, namely that they are, in some sense, final outputs from an ecosystem. They are 'final', in that they are still connected to the ecological structures and processes that gave rise to them and final in the sense that these links are broken or transformed through some human interaction necessary to realise a benefit. Often this intervention can take the form of some physical action such as harvesting the useful parts of a crop. The interaction might also be non-material and more passive involving, for example, the benefit obtained from the reduction or regulation of some kind of risk (flood risk is the example shown in Figure 1), or the intellectual or spiritual significance of nature in a particular cultural context. Thus services are at the point where the 'production boundary' is crossed between the biophysical and the socio-economic parts of the socio-ecological system.

Socio-ecological systems are, of course, more complex than Figure 1 suggests. However, this simple diagram does help us to understand that all the different elements of the cascade need to be considered if we want to appreciate what an ecosystem service really is and how it connects people and nature (see for example: http://www.biodiversity.fi/ecosystemservices/cascade/). We need to map and measure indicators across the entire pathway to build up a complete picture. The left hand side of the cascade captures the important elements that determine the capacity of an ecosystem to supply services, while the right hand side identifies the aspects of the demand for them. And understanding the balance between them is at the heart of the contemporary sustainability debate and key to our understanding of the way people and nature are linked.

Potschin M, Haines-Young R (2011) Ecosystem Services: Exploring a geographical perspective. Progress in Physical Geography 35(5): 575-594.
Potschin M, Haines-Young R (2016) Defining and measuring ecosystem services. In: Potschin M, Haines-Young R, Fish R, Turner RK (Eds.) Routledge Handbook of Ecosystem Services. Routledge, London and New York, 25-44.

Categorising and describing ecosystem services (ES) is the basis of any attempt to measure, map or value them. It is the basis of being transparent in what we do, so that we can communicate our findings to others, or test what they conclude. So fundamental is the need to be clear about how we classify ES that it might seem that it is an issue that must already be well and truly resolved. The aim of this chapter is to suggest that this might not, in fact, be the case entirely and that the way we categorise ES is something that still represents a challenge.

A number of different typologies, or ways of classifying ES are available, including those used in the Millennium Ecosystem Assessment (MA) and The Economics of Ecosystems and Biodiversity (TEEB) and a number of national assessments, such as those in the UK, Germany and Spain. The problem with them is that they all approach the classification problem in different ways, involving different scale perspectives and different definitions resulting in the fact that they are not always easy to compare. In order to try to partly overcome this 'translation problem', the Common International Classification of Ecosystem Services (CICES) was proposed in 2009 and revised in 2013. A typology translator is available via the OpenNESS-HUGIN website".

We do not argue that it is better than any other system, but it illustrates the difficulty of designing a classification system that is simple and transparent to use. We will argue that the problem of classification is still worth working on - and it is certainly not something that can be taken for granted. We would encourage everyone to think about it when they embark on any kind of analysis involving ES.

The conclusion that we would like to advance is that the ES community probably needs to develop a number of different classifications or typologies that can be used to name and describe all the elements in the cascade that we described in Chapter 2.3, namely: the ecosystem or habitat units that give rise to the ES of interest, the ecological functions that are associated with them, as well as the benefits and beneficiaries whose well-being is dependent on the output of services and, of course, the values that people assign to these benefits. Services can also be classified according to such criteria as whether they give rise to private or public benefits, whether people can be prevented from accessing the service ('excludable' vs 'non-excludable'), or whether the use of a service by one individual or group affects the use by others ('rival' vs 'non-rival').

CICES was originally developed as part of the work on the System of integrated Environmental and Economic Accounting (SEEA) led by the United Nations Statistical Division (UNSD), but it has been used by the wider ecosystem services community to help define indicators of ES, or map them. In designing it, the intention was to provide a way of characterising 'final services', namely those that interface between ecosystems and society. In this sense, it follows the definition used in TEEB, namely that these final services are the things from which goods and benefits are derived. However, it did try to use as much of the terminology that was already widely employed and so used the categorisation of 'provisioning', 'regulating' and 'cultural' services that were made familiar by the MA.

Material and energetic outputs from ecosystems from which goods and products are derived are contained in CICES provisioning services. Regulating services categories refer to all the ways that ecosystems can mediate the environment in which people live or depend on in some way and therefore benefit from them in terms of health or security, for example. Finally, the cultural category identified all the non-material characteristics of ecosystems that contribute to, or are important for people's mental or intellectual well-being. CICES is hierarchical in structure, splitting these major 'sections' successively into 'divisions', 'groups' and 'classes'. Figure 1 illustrates how this works using the example of 'cereals'.

The full version of CICES is available online.

Figure 1.

The hierarchical structure of CICES illustrated with reference to a provisioning service (cultivated crops - cereals). Credit: Haines-Young and Potschin.

The first challenge that working on CICES showed was how difficult it is to categorise 'final ecosystem services'. These, according to Boyd and Banzhaf, are the 'end-products of nature' who argue that it is important to define them clearly to avoid the problem of 'double counting' when we calculate their value; i.e. assessing the importance of a component of nature more than once generally because it is embedded in, or underpins, a range of different service outputs. More formally these authors suggest final services 'are components of nature, directly enjoyed, consumed, or used to yield human well-being'. The problem is that, what constitutes a final service, generally depends on the context in which the assessment or mapping exercise is being made; thus CICES lists potential final services.

A second challenge was whether abiotic ecosystem outputs like wind or hydropower, or minerals like salt, should be categorised as 'ecosystem services'. In the end, the argument that the category 'ecosystem services' should be restricted to those ecosystem outputs that were dependent on living processes won the day, because it strengthened arguments about the importance of 'biodiversity' to people; an accompanying provisional classification of abiotic services that follows the CICES logic has, however, been developed and is available.

It is worth mentioning that the final challenge which we encountered in designing CICES, was the difficulty that people have in distinguishing services and benefits. The distinction is a difficult one to make because it involves deciding where the 'end-product of nature' is transformed into a good, product or benefit, product or benefit as a result of human action of some kind. The distinction we use in CICES is whether the connection with the underlying ecological processes and structures is retained; hence the standing crop of wheat in the field is a final service from an agricultural ecosystem, but the grain in the silo is the good or benefit.

The distinction between services and benefits is an important one because a single service can give rise to multiple goods and benefits that all need to be identified if services are to be valued appropriately. In the case of rice for example, in addition to the harvest of the grain, rice straw and husks can be used for animal feed or as raw material for energy.

In this chapter we have used CICES to explore some of the challenges that we need to face when developing systems for categorising ES. These systems are complex and experience suggests that they will need to be developed in an iterative way, using experience to find out what works where and how naming conventions and definitions can be improved. While we have used CICES to illustrate some of these issues, it is important not to overlook the fact that it is a system that, despite limitations, has been used effectively.

For example, CICES forms part of the mapping framework designed to support the EU's Biodiversity Strategy to 2020 (the second report of the Mapping and Assessment of Ecosystem Services (MAES) uses CICES classes to identify a range of indicators that can be used for mapping and assessment purposes (see: http://biodiversity.europa.eu/maes/#ESTAB, accessed 30/01/2016; see also Chapter 7.1). A number of papers have appeared in peer-reviewed scientific literature that have either used CICES or commented upon it as part of their methodological discussion.

CICES has, for example, been used as the basis of the German TEEB study as well as the scoping work for a German National Ecosystem Assessment, NEA-DE. The TEEB report on Agriculture also recommends the use of CICES. Elsewhere, CICES has been refined at the most detailed class level to meet the requirements of ecosystem assessment in Belgium. Research in Finland used CICES to develop an indicator framework at the national scale. These kinds of applications suggest that the detailed class level in CICES can be useful as building block from broader reporting categories, the advantage being that these broader categories are themselves defined in a transparent way. These types of use illustrate the kinds of application that any good classification system must be able to support. Many more applications can be found - several are listed in the further reading material.

While the applications of CICES suggest that the current framework is appropriate for many uses, it is also clear that we need to think carefully about how such systems can be developed. For example, researchers have suggest that it may need to be adapted to ensure that it is suitable for the assessment of marine and coastal ecosystems, or integrated more closely with typologies for describing underlying ecosystem functions. It is probable that marine interests were under-represented in the consultations that led to the current CICES version.

Thus while the current version of CICES clearly works for many purposes, given the importance of categorising ES in clear and transparent ways, the development of this and other systems needs to be reviewed constantly as our needs and concepts evolve. They are essential tools for our mapping and assessment work. It has been suggested, for example, that a classification, such as CICES, might form part of a more general systematic approach or 'blue print' for mapping and modelling ecosystem services. Other authors have emphasised that it is important to develop classification systems, such as CICES, that are 'geographically and hierarchically consistent' so that we can make comparisons between regions and integrate detailed local studies into a broader geographical understanding.

Our concluding point is that, whether CICES has a role to play or not, these kinds of systems will not build themselves. We need to be aware of the challenges that the categorisation of ES still poses and the fact that we have only just started to address them.

Note: At the time of writing, version 4.3 is to be used. This version is currently under revision and version 5 is under development. All details are available on the CICES webpage.

Boyd J, Banzhaf S (2007) What are ecosystem services? The need for standardized environmental accounting units. Ecological Economics 63: 616-626.
Haines-Young R, Potschin M (2013) Common International Classification of Ecosystem Services (CICES), version 4.3. Report to the European Environment Agency EEA/BSS/07/007 (download: www.cices.eu).
Potschin M, Haines-Young R (2016a) Defining and measuring ecosystem services. In: Potschin M, Haines-Young R, Fish R, Turner RK (Eds.) Routledge Handbook of Ecosystem Services. Routledge, London and New York, 25-44.
Potschin M, Haines-Young R (2016b) Report on Workshop on "Customising CICES across member states". Milestone 19 of ESMERALDA (download at: http://www.esmeralda-project.eu/documents/).

Cartography (from Greek ?????? khartes, "map"; and ??????? graphein, "write") is the art and science of representing geographic data by geographical means. Maps are the main products of cartographic work and are graphic representations of features of an area of the Earth or of any other celestial body drawn to scale. Regardless of the map type or the mapping technique applied (Chapter 3.2), every map has a coordinate system, a projection, a scale and includes specific map elements. These attributes usually depend on the size and shape of the mapped geographical area and the graphical design of the map representation that needs to be informative and understandable for the map-user (Chapters 5.4 and 6.4).

Geographic Information Systems (GIS) are powerful tools for data Input, Management, Analysis and Presentation (IMAP principle) providing multiple possibilities for a better understanding of the structures and patterns of human and natural activities and phenomena (Chapter 3.4). Nevertheless, much of its easy-to-apply default-functionality can be misleading for an inexperienced map-maker.

In the present chapter, we discuss the main characteristics of maps such as coordinate system, geodetic datum, projection, scale and map elements; how to choose them accordingly and what their role is for proper use of a map. The use of GIS has significantly simplified mapping and provides a good environment for the visualisation of Ecosystem Services (ES).

The coordinate system of a dataset is used to define the positions of the mapped phenomena in space. It furthermore acts as a key to combine and integrate different datasets based on their location. This enables the performance of various integrated analytical operations, such as overlaying or merging data layers from different sources. Coordinate systems can be geographic, projected or vertical systems.

Geographic coordinate systems

A Geographic Coordinate System (GCS) uses a three-dimensional spherical surface to define locations on the Earth, i.e. the Earth is represented as a sphere or a spheroid. A point on that sphere is referenced by its longitude and latitude values. Longitude and latitude are angles measured in degrees from the Earth's centre to a point on its surface. The Prime meridian and Equator act as reference for longitude and latitude respectively (Figure 1).

Figure 1.

The world as a globe with longitude and latitude values.

Vertical coordinate systems

A vertical coordinate system defines the vertical position of the dataset from a reference vertical position - usually its elevation (height) or depth from the sea level (Figure 2).

Figure 2.

Two vertical coordinate systems: mean sea level and mean low water.

While the definition of a geographic or projected coordinate system is obligatory for all datasets, vertical coordinate systems are only needed if the vertical height of data is of relevance. Lack of, or wrongly defined, coordinate system information leads to problems of spatial data integration. (Figure 3).

Figure 3.

Integration of datasets for the same area (inspired by Buckley 1997).

Therefore it is very important when using digital mapping tools that the used datasets are defined in an eligible coordinate system.

The geodetic datum defines a) the size and shape of the Earth and b) the orientation and origin of the used coordinate system through a set of constants. The geodetic datum can be based on flat, spherical or ellipsoidal Earth models:

• Flat Earth models are used over short distances so that the actual Earth curvature is insignificant (< 10 km);
• Spherical models represent the figure of the Earth as a sphere with a specified radius, leading to deformations in the model which are largest at the poles; used for short range navigation and global distance approximations; and
• Ellipsoidal models are the most accurate models of Earth; used for calculations over long distances; the reference ellipsoid is defined by semi-major (equatorial radius) and flattening (the relationship between equatorial and polar radii).

The ellipsoidal model can represent the topographical surface of the Earth (actual surface of the land and sea at some moment in time), the sea level (average level of the oceans), the gravity surface of the Earth (gravity model) or the Geoid. The Geoid is the equi-potential surface that the Earth's oceans would take due to the Earth's gravitation and rotation, neglecting all other influences such as winds, currents and tides.

The World Geodetic System 1984 (WGS-84) datum defines geoid heights for the entire Earth in a ten by ten degree grid. The Global Positioning System (GPS) is based on the WGS-84.

The geodetic datums can be horizontal (latitude and longitude), vertical (height) and complete. The transformation between datums requires the application of strict mathematical rules and sets of parameters, depending on the required transformation. Most GIS and mapping platforms support automated transformation between datums and coordinate systems.

Map projections are mathematical representations of the Earth's spherical body on a plain surface through mathematical transformations from spherical (latitude, longitude) to Cartesian (x, y) coordinates. Map projections usually depend for the transformation on a form which can be developed or flattened - a plane, a cone, or a cylinder - which is attached to the sphere at one point or at one or two standard lines. The respective map projections are referred to as planar, conic and cylindrical (Figure 4).

Figure 4.

Developable (flattenable) surfaces (in Monmonier 1996).

The transformation of a spherical surface into a plane leads to different distortions in the lengths, angles, shapes and areas of the mapped surface. The distortions are usually smallest along the standard lines and close to the attachment point. Depending on the shape and size of the mapped area, appropriate projection and standard lines should be selected. Distortions are inevitable and it is impossible to create the "perfectly" projected map that fulfils all map projection properties. The four properties of the map and their respective projection types are:

• Local shapes of the features on the map are the same as on the Earth's surface. This conformal projection maintains all angles.
• The areas of the features on the Earth are in the same proportions as on the map. Other properties - shape, angle, and distance - are distorted in equal-area projections.
• The scaled distances along the standard lines, or from the attachment point, to all other points on the map are maintained in equidistant projections. This is not valid along all lines or between any two points on a map.
• The directions on the map are correct in the true-direction (azimuthal) projection. It gives the directions (or azimuths) of all points on the map correctly with respect to the centre. Some true-direction projections are also conformal, equal-area, or equidistant.

For every map, only one or two of those properties can be fulfilled and the cartographer has to make a choice, depending on the purpose and needs of the map (see Chapter 5.4).

The scale represents the ratio of the distance between two points on the map to the corresponding distance on the ground. Thus large scale maps (with a large reciprocal value of the scale, such as 1:5,000) cover small areas with great detail and accuracy, while small scale maps (e.g. 1:1,000,000) cover larger areas in less detail (Figure 5). The map scale also influences generalisation (Chapter 3.4) and symbolisation (Chapter 3.3) of the map.

Figure 5.

Interaction between map content and scale selection.

When choosing the map scale, the cartographer should consider:

• Purpose of the map - the mapped phenomena need to be well-represented in the selected scale;
• Map size - the scale need to be adapted to the size of the mapped area and the desired final size (format) of the map;
• Detail - the scale need to be adapted to the detail in which the phenomena are mapped.

Map scales can be expressed as a ratio, a verbal statement or as a graphic (bar) scale (Figure 6). On non-analogous (digital) maps, it is essential to use a graphic scale bar (linear bar). A scale bar adjusts to the resolution of the respective display, a parameter which cannot be controlled by the map maker. The variability of map size by using a projector is an example of this problem.

Figure 6.

Examples of ratio, verbal and graphic scales.

Elements of a map are crucial for providing the map-user with critical information about the map content. Making a thematic map is to a large extent a creative act and the choice of map elements depends on the context, audience and the preferences of the map-maker. Nevertheless, there are three levels for representation of the elements of a map, presented here by by their level of relevance (Figure 7):

Figure 7.

Example map and its elements: actual map, scale, north arrow, legend, title, coordinate system and projection, cartographer's name and institution, date of production, data source and neatline.

• Elements that make the proper reading of the map possible and it is recommended to add them to all maps:
  • Scale information;
  • Map direction - a symbol, usually an arrow, that indicates the true north (the direction to the North Pole); if a coordinate grid (graticule) is added to the map or on small-scale (e.g. continental) maps, a north arrow is not required;
  • Legend - the legend lists all symbols, their sizes, patterns and colours used in the map and the features they depict (see Chapter 3.3); they should appear in the legend exactly as they are found in the body of the map;
• Elements that provide context:
  • Title - should provide a short and clear statement about the map content, usually stating the name of the mapped area and the map theme (in ES maps - the mapped ES) along with the depicted year in thematic maps; it should be considered that this information can be included in the map legend title also;
  • Projection - provides information about the projection and possible distortions in the area, distance, direction and shape of the mapped features;
  • Cartographer's name and/or the authority responsible for the composition of the map;
  • Date of production;
  • Data sources used to create the map.
• Elements used selectively to assist effective communication (optional):
  • Neatlines (clipping lines) - used to frame the map and indicate the exact area of the map;
  • Locator maps - to place the body of the map within a larger geographical context;
  • Inset map - a "zoomed in" map of small areas from the map with high relevance, where information is too clustered for the scale of the map body;
  • Index maps - when labels or other information cannot be placed effectively in the body of the map, they can be input separately to increase readability.

Cartography is based on a long tradition and comprehensive knowledge of map-creation and map-use. ES map-makers still need to be aware of the general principles, techniques (Chapter 3.2) and logics (Chapter 3.3) of cartography, although with today's software programmes, it seems all too easy to create lots of maps rather quickly. Digital maps are the main means of map representation nowadays and the main tool for geographic data interpretation, visualisation and communication. They provide multiple opportunities but also 'traps' for the map-maker. Therefore, instead of producing large quantities of badly-compiled and misleading maps, ES map producers should harness the available knowledge and techniques in order to support the proper application of ES and ES mapping in science, decision making and society (Chapter 7).

Bugayevskiy LM, Snyder JP (1995) Map Projections - A Reference Manual. Taylor & Francis, Great Britain.
Fenna D (2007) Cartographic Science: A Compendium of Map Projections, with Derivations. CRC Press, Boca Raton, Florida.
International Hydrographic Bureau (2003) User's Handbook on Datum Transformations Involving WGS 84. 3rd Edition (Last correction August 2008). Special Publication No. 60. Monaco.
Maling DH (1992) Coordinate Systems and Map Projections, 2nd Ed. Pergamon Press. Oxford.
Monmonier M (1996) How to lie with maps. 2nd ed. The University of Chicago Press.
Pearson F (1990) Map Projection: Theory and Applications. CRC Press, Boca Raton, Florida.
Snyder JP (1987) Map Projections - A Working Manual. U.S. Geological Survey Professional Paper 1395. U.S. Government Printing Office. Washington, D.C.
Snyder JP (1993) Flattening the Earth: Two Thousand Years of Map Projections. University of Chicago Press. Chicago, Illinois.

ArcGIS (ESRI Desktop Help): http://resources.arcgis.com/en/help/

Buckley DJ (1997) The GIS Primer. Pacific Meridian Resources Inc.: http://planet.botany.uwc.ac.za/nisl/GIS/GIS_primer/index.htm

Further:

http://geokov.com/education/map-projection.aspx
http://www.progonos.com/furuti
http://www.colorado.edu/geography/gcraft/notes/mapproj/mapproj_f.html
http://www.colorado.edu/geography/gcraft/notes/cartocom/cartocom_ftoc.html
http://www.colorado.edu/geography/gcraft/notes/datum/datum_f.html
http://www.librry.arizona.edu/help/how/find/maps/scale
http://awsm-tools.com/geo/convert-datum
http://gitta.info/LayoutDesign/en/html/index.html

Mapping is about the graphical representation of spatio-temporal phenomena. Illustrating our complex environment by symbols and graphics requires important decisions: Does the chosen map type properly reflect the Ecosystem Service(s) (ES) to be portrayed? Are more intuitive design choices available to visualise and explain a particular dataset? What happens if the map type does not fit the data? This chapter aims to investigate popular map types like dot maps, choropleth maps, proportional symbol maps, isarithmic maps and marker maps. We relate those types to inherent spatial and statistical characteristics of certain ES phenomena and give advice on advantages and possible pitfalls related to their usage.

Every ES map, whether paper or digital, is a graphical representation of ES in their geographic context. In most cases, such maps are built to facilitate understanding of ES in their spatial (Chapter 5.2) and/or temporal (Chapter 5.3) dimension. What kind of ES data should be presented to whom (e.g. general public, scientific community, ES-practitioners) greatly determine the mapping process: a process of abstraction from geographic reality to the final map. Scientific cartography developed an extensive body of theory and derived practical guidelines to accomplish this process. A major goal thereof is the provision of maps that can be intuitively read and correctly understood and used by the intended end user (Chapter 6.4).

Data are the result of measurements (Chapter 4.1), modelling (Chapter 4.4) or other quantifications (Chapter 4) of geographic phenomena. Air temperature data, for example, is typically gathered by taking measurements at several point locations. Data on tree diameters might look similar, since it uses the same geometry (points) and is measured on a metric level. However, the represented phenomenon (trees) is entirely different in nature, since trees only exist at discrete locations in space, while atmospheric conditions are continuously distributed and can be measured everywhere.

Different data models can be used to store, analyse and present spatial data, for example in Geographic Information Systems (GIS):

Vector data models represent discrete or continuous spatial phenomena by using points, lines and polygons. Vector data have high accuracy for displaying features with distinct boundaries; vector map data files usually use less memory capacity.

Raster data represent the world in a regular grid of cells (pixels). Raster models are often used for continuously varying phenomena or they are the result of remote sensing.

It is possible to convert vector to raster data and vice versa. However, based on the different data model concepts, such conversions normally lead to loss of information and/or data accuracy.

When defining maps as graphic representations with the aim of facilitating the understanding of spatial phenomena, mapping techniques that properly reflect their main spatial characteristics should be chosen. But what does properly reflect mean? According to the congruence principle from cognitive design, the structure and content of visualisations should correspond to the desired structure and content of mental representations. The basic mapping concept of scaling geographic space is appropriate in this respect, since distances and directions between entities are adequately represented by the scaled distances and directions of their corresponding map symbols (except when mapping on continental scale and projection distortion is apparent). Thus it facilitates the development of mental models on the respective spatial configuration. However, it makes a difference whether a spatially continuous geographic phenomenon like the air is represented as a set of discrete dots or by alternative graphic means corresponding better to its spatial continuity.

Spatial phenomena can be categorised based on spatial continuity and spatial (in)dependence. For each possible combination, Figure 1 suggests a specific mapping technique, as discussed in the following section.

Figure 1.

Models of geographic phenomena and suggested symbolisation methods. Simplified after MacEachren (1992).

While such a scheme can assist in selecting an appropriate thematic mapping technique for quantitative data, there are further corresponding considerations:

What is the intended usage of the ES map (Chapter 5.4)? Does it merely act as an interface with the ES relevant entities, should it provide an overview on general spatial patterns or is it intended to allow for local comparisons?

Is the data related to individual locations or is it aggregated to enumeration units?

Is the data standardised (e.g. rates) or not (raw counts)?

The following section describes important thematic mapping techniques while addressing such considerations.

Common thematic mapping techniques include dot (density) maps, marker maps, choropleth maps, proportional symbol maps and isarithmic maps.

In their simplest form of one-to-one feature correspondence, dot maps (also known as dot distribution maps) follow a very easy concept: at each location of the mapped entity, there is a corresponding small symbol in the map. Although this one-dot-per-feature approach is increasingly popular even in small scales and with very large numbers of features (http://demographics.coopercenter.org/DotMap/), dots quickly coalesce to a shading of variable intensity, which might be unfavourable for certain applications. In that case, a one-to-many approach is favourable, were each dot represents a fixed number of entities (e.g.: 1 dot = 100 people). The choice of the number of entities per dot is related to the chosen dot size, the scale and the density of feature locations. As a rule of thumb, points should start to coalesce in the map areas of maximum density.

Dot maps are especially suited to focus on the distribution patterns of entities or on differences in local densities. When using the dot density approach for polygonal aggregated data (e.g. number of people per district), the according number of points is placed within each polygon. To determine the position of each point within its polygon, several options apply:

• Random point distribution is straightforward and often used, although it might be misleading in cases with a very uneven distribution (e.g. randomly distributing points representing the population of Egypt on the country area).
• Adjust the point positioning within a polygon by using information on densities in neighbouring polygons.
• Use of ancillary information (e.g. settlement information from remote sensing data) for more precise point allocation.

Dot density maps which are based on aggregated data require absolute counts as a basis (e.g. number of persons per county). In addition, the use of an area-preserving map projection (see Chapter 3.1) is essential, since the density impression results from the number of dots per area unit on the map.

Heat maps are frequently seen derivatives of dot maps. Instead of showing the actual dots, they use areal colouring to represent their density. Dense areas get more reddish colours (therefore "heat") while areas with sparse data are normally coloured in blue. Although heat maps are quite popular, it is somewhat difficult to derive actual point feature numbers for a certain area.

Marker maps are a special form of dot maps that emerged with the advent of web mapping applications such as Google maps. Lying on top of a topographic base map, every marker or "pushpin" symbolises a feature of interest in its geographic location. With each marker being hyperlinked, the user can obtain additional object information or trigger certain actions, like booking a hotel room. The map itself acts foremost as an interface to data which is structured by its spatial location.

Paper maps showing the location of entities often use different symbols for different object types referenced in a legend. Thus the selection of the currently relevant object is performed visually by the user. Contrary to this, a web map allows the user to query the objects of interest within a database first and then show the query result in the map. Consequently, no further graphical differentiation of markers is necessary (but still possible).

Point markers are used to depict any type of feature geometry in the map, be it points, lines or areas. The main reason refraining from clickable areal symbols is explained by interaction challenges with other objects lying within the same area. Marker maps are often used to encode qualitative information. They mainly inform the user about individual locations and the spatial distribution pattern of the entities of interest. To prevent markers from coalescing in small scales, different mechanisms for grouping and/or selection can be applied.

Choropleth maps are preferably used to map data collected for areal units, such as states, census areas or eco-regions. Their main purpose is to provide an overview of quantitative spatial patterns across the area of interest. To construct a choropleth map, the data for each unit is aggregated into one value. According to their values, the areal units are typically grouped into classes and a colour is assigned to each class. This requires the use of meaningful colour-schemes (http://colorbrewer2.org) (Chapter 3.3), representing the sequential or diverging nature of the mapped phenomenon.

Although choropleth maps are very common, several pitfalls are inherently associated with them:

Variation within units is ignored, although the mapped phenomenon might vary considerably within (especially larger) units.

The boundaries between units often do not align with discontinuities in the mapped phenomenon. Especially the historically defined boundaries of administrative units often poorly align with spatial discontinuities of current social or natural processes (Chapter 5.2). Both problems, namely the variation within units and the definition of spatial boundaries apply for many ES and belong to the so called Modifiable Areal Unit Problem (MAUP; see Chapter 6.1).

Choropleth maps are only suitable for mapping standardised ("normalised") data like rates (yield per ha per year) or densities (persons per km?). Mapping absolute values (e.g. counts of persons per unit) is wrong since size differences of individual units will greatly affect the result: large units will tend to have higher values, small units lower ones. Even for experienced map users, it is impossible to mentally disentangle the resulting relationship between unit-size and colour for correcting the wrong impression of spatial distributions (compare Figure 2). However, in most cases, standardised values can be easily derived from raw counts.

Figure 2.

Only standardised data (rates etc.) should be mapped with choropleth maps. Inspired by Slocum et al. (2009).

In summary, choropleth maps are a good choice to demonstrate standardised data aggregated to areal units, especially if there is little variation within units and the boundaries of the units are meaningful for the mapped phenomenon.

Based on our assumption that 'larger' means 'more', proportional symbol maps use variation in symbol size to depict quantities. While the size of point symbols can be used to denote quantitative attributes of point features (e.g. spring symbols scaled to water outputs), scaled point symbols are also used to represent data aggregated to areas, as discussed for choropleth maps. Contrary to the latter, not only is the colour of the areal units modified based on an attribute, but a point symbol is positioned within each area and the size of this symbol is scaled according to the desired attribute. Since comparing sizes is much easier than comparing shades, proportional symbol maps are especially effective for comparison tasks. According to the scheme in Figure 1, proportional symbol maps best connote spatially discrete entities with spatially unrelated attributes. In contrast to choropleth maps, they are capable of handling absolute data like raw object counts within differently sized areas. This is possible due to the fact that larger symbols can be related to larger areas quite intuitively (Figure 3).

Figure 3.

Symbol size relates well to the size of areal units, making proportional symbol maps capable of mapping non-standardised, absolute values (see Figure 2 for the underlying object distribution). Inspired by Slocum et al. (2009).

In their basic form, the area of a symbol is scaled proportionally to the magnitude of the mapped attribute. However, several variants apply:

• Although the subject is controversial, perceptual scaling tries to adjust the symbol size to compensate the empirically tested tendency for underestimating the area of large symbols.
• The use of 3D-symbols like spheres or cubes allows scaling proportionally to symbol volume instead of area. Although volumes are estimated even more badly, this might be useful when large spans of data values have to be accommodated in the map.
• For data of extremely large or very small value ranges, data values might be classed and classes are assigned a set of 'graduated' symbols. While symbol sizes still represent the order of classes, symbols are not proportional to the magnitude of values any more. Thus additional information (e.g. in a legend) pointing to that fact is crucial for interpretation.
• At times, data is composed of several subgroups (e.g. total population by gender or age groups). To show this further subdivision, scaled diagrams can be used instead of plain symbols. Pie charts are often chosen due to their compactness.

Often, proportional or graduated symbols will overlap. While overall downscaling might be a solution, a small amount of overlap is acceptable. Using half-transparent, simple symbols like circles is a good strategy to cope with overlap as well. Web maps sometimes use cross-breeds of markers and proportional symbols: instead of permitting marker-overlap in small scales, nearby markers are aggregated into one symbol scaled to the number of markers it contains.

Many ecosystem processes like climate regulation or air quality regulation take place in a spatially continuous manner. As a consequence, the related ES are also gradually varying over space. Isarithmic maps connect points of the same value (at certain intervals) by a line (=isoline) and are especially useful to map such smoothly changing 'continuous field' data. The most prominent examples of isolines are contour lines in topographic maps, connecting points of the same elevation. This concept can be used for all types of continuous fields. Isarithmic maps can be combined with areal colouring using continuous colour ramps. Alternatively, the areas between the isolines can be filled with a sequence of classed colours. A combination of isolines with analytical hill-shading intensifies the 'surface'-character of the mapped phenomenon.

The construction of isarithmic maps requires surface data, commonly modelled as point grid or Triangulated Irregular Network (TIN). Grounded on a base value and an interval, isolines are constructed from the field model using spatial interpolation. Using, for example, a base value of 50 and an interval of 100 to display a surface with values ranging between 54 and 320, isolines of the value 150 and 250 will be the result. Since isarithmic maps emphasise the continuous, smoothly varying character of a phenomenon, it is advisable to use them for such phenomena even though the data is being provided as discrete samples. As an example, data on ecological vulnerability based on districts could be considered: while each district might have assigned a value indicating its vulnerability, local vulnerability might smoothly change over space, independently of sharp district borders. Depending on the intended message ('objective representation of risk' versus 'hey governor, you are responsible for this highly vulnerable district, act!') it might make sense to create a continuous vulnerability surface from polygonal data and utilise an isarithmic map for its communication. When following such an approach, it is important to use only standardised (relative) values from enumeration units for surface generation. Methods like pycnophylactic interpolation or area-to-point kriging, guarantee that the overall volume remains constant while the surface is smoothed.

Apart from the basic thematic mapping concepts described so far, there are numerous other techniques: Cartograms, dasymetric maps, flow maps, animated maps or perspective views are just some examples for techniques meeting more specialised purposes.

A typical ES map consists of a topographic base map and one or more superimposed thematic layers showing the desired ES data. The base map provides the geographic reference to the ES data, informing the user on location while simultaneously providing a sense of the actual map scale. Depending on the used mapping framework mapping framework, there is often a choice between various base maps base maps. Some base maps can also be edited by the user to highlight or subdue certain object classes.

When choosing a base map, several aspects must be considered:

• Thematic support: The base map should support the thematic ES information; therefore it depends on the mapped ES topic, what kind of geographic features should be part of the base map. While some base maps focus on the street network street network, others emphasise the terrain or highlight administrative boundaries. Users should carefully think about what kind of information is required to support the mapped ES topic.
• Visual prominence: Base maps provide ancillary information, thus their place is in the visual background. In a digital context there are two common concepts to accomplish this: A dark base map with bright and saturated thematic information on top or a light and unsaturated base map overlaid by darker and more saturated thematic layers.
• Visual density: At each scale level, the base map should have approximately the same visual density (number of shown features per area). If the thematic ES layers are rather complex, a base map with a rather low visual density (e.g. only coastline and country boundaries) should be chosen.

Due to scale limitations it is not possible to show all spatial objects with all their detail in the limited map space. Generalisation aims to represent the ES-information in a level of detail appropriate for a given scale, user group and use context. It is necessary in cases where the visual density in maps is increasing rapidly, symbols overlap or topological conflicts become evident due to graphical scaling. Figure 4 shows typical operations applied in the generalisation process. Although the application of some of those operators can be automated, it is the responsibility of the map maker to decide on the relevance of specific ES information.

Figure 4.

Typical operators for generalisation. Modified after Phillipe Thibault (in: Slocum et al. 2009).

Map-makers can harness the broad knowledge base, experience and techniques available from cartography. ES-maps display highly complex human-environmental systems, consisting of discrete and continuous features. This complexity should also be respectively reflected in the maps, which need to be logical, clear, understandable and well-designed.

Brewer CA (1999) Colour Use Guidelines for Data Representation. Paper presented at the Proceedings of the Section on Statistical Graphics, Alexandria.
Krygier J, Wood D (2011) Making Maps, Second Edition: A Visual Guide to Map Design for GIS: The Guilford Press.
MacEachren AM (1992) Visualising uncertain information. Review of. Cartographic Perspectives (13):10-9.
MacEachren AM (2004) How Maps Work - Representation, Visualisation and Design. 2 ed. New York, London: The Guilford Press.
Muehlenhaus I (2013) Web cartography: map design for interactive and mobile devices: CRC Press.
Slocum TA, McMaster RB, Kessler F, Howard HH (2009) Thematic Cartography and Geovisualisation. Clarke KC (Ed.) 3rd ed, Prentice-Hall Series in Geographic Information Science. Upper Saddle River, NJ: Pearson Prentice Hall.
Tversky B (2005) Prolegomenon to scientific visualisations. In Visualisation in science education, Springer, 29-42.

Map-making and cartography combines science, arts, aesthetics and techniques that follow map-specific logic. Thus, cartography is strongly based on semiotics, the theory and study of signs and symbols. Map symbolisation is a key attribute of each map that determines the map elements (Chapter 3.1) and their applicability for communication (Chapter 6.4) and other uses (Chapter 7). Knowledge about basic semiotic principles is needed to produce proper ecosystem services (ES) maps that are fit for purpose.

Semiotics comprises semantics, syntactics and pragmatics:

• Semantics is the study of the relationships between signs and symbols and what they represent,
• Syntactics deal with the formal properties of languages and systems of symbols and
• Pragmatics analyse the relationships between signs and their users.

This chapter introduces map semantics and syntactics, which are the basis for the proper use of symbols, patterns and colours for different mapping purposes and scales. Chapter 6.4 deals with map pragmatics.

Map features can be points, lines or areas (polygons). They are positioned on a map relating to their location in reality, the map scale and map projection (Chapter 3.1). Additional information is communicated by the choice of the map symbols' shapes, sizes, colour hues, colour values, colour intensities and textures. According to the different graphical variables' semantics, they are best used to show qualitative or/and quantitative differences. Most ES maps and ES model outputs (Chapter 4.4) are choropleth maps (Chapter 3.2) displaying areas of ES supply or demand. Some ES and landscape features are displayed as point or line features. Figure 1 gives an overview of the six key graphical variables and how they can be used for mapping.

Figure 1.

Key graphic variables and their application in mapping (inspired by understandinggraphics).

The map symbols' shapes are used to represent qualitative differences in thematic maps. In many cases, the shape is a logical connection to the feature that it represents (e.g. a petrol pump representing a gas station or a bed indicating a hotel). Text or respective letters/abbreviations are also often used (e.g. 'P' for car parking or 'H' for hotel). Shapes or adaptations of shapes are most often used for spatially discrete point features (see Chapter 3.2). They are rarely used for line features but can be applied in the form of cartograms (or anamorphic maps) for area features. In anamorphic maps, mapped areas are resized based on particular indicator values (see for examples: http://books.worldmapper.org).

Another relevant graphic feature that can be related to shape is orientation, which can be, for example, used to indicate directions of ES flows, movements or directional ES connections (Chapter 5.2). Topic-specific maps (e.g. geological map or weather maps) contain highly complex and specialised symbols. These maps often use their own logic of semantics and non-specialists can have difficulties interpreting them.

Size is mainly applied in graphics to express quantitative differences, i.e. variations in amount or count (such as 'the more the larger' and, vice versa, 'the less the smaller'). Size can also be used to suppress less important features. The size of point and line features can be chosen accordingly, but following a rule of thumb to choose the difference in size according to quantitative differences in the features (e.g. double size for a double amount; see 'Classification of data' below). For graphical reasons, some smaller linear or point features (e.g. streams) are often enlarged, although the proportional size to other symbols might not represent reality (see 'generalisation' Chapter 3.2). The meaning of the different sizes (their semantics) should be explained in the map legend by providing the quantitative numbers that are behind the symbols (Chapter 3.1). Size variations of area features should refer to anamorphic maps or cartograms.

Combinations of different visual variables are possible, such as dot maps (see Chapter 3.2), illustrating distributions and densities by the symbols' shapes and quantities by their sizes. Depending on the scale of the map and the complexity of the landscape to display, shape and size may not offer sufficient detail or visibility for small symbols.

Colour hue is arguably the most powerful of the graphical variables. It can be applied to point, linear and area map elements. Different colour hues can relatively easily illustrate qualitative differences such as different land cover types in area maps. Variations in colour value or intensity are commonly used to portray quantitative differences in both dot and choropleth maps (see Chapter 3.2).

When using colour in maps, the map-maker needs to be aware that the different colours have specific meanings for many people and cause different psychological effects when viewed. Figure 2 shows some examples, noting that there are many different interpretations on colours based on the subject area and culture. In many cultures green stands for positive developments whereas red is often related to negative things such as intense heat or danger.

Figure 2.

Possible psychological associations of different colours for viewers (based on: http://guity-novin.blogspot.de/2014/07/chapter-70-history-of-color-color-wheel.html).

Due to the omnipresent nature of mapping products in all kinds of media, the map-maker needs to be aware that many colours are connected with particular geographic phenomena (e.g. green for forests, blue for water bodies, red or black for urban areas; see for example the European CORINE land cover data set). Applying such commonly used colour schemes is essential for an easy and correct communication of the map content (see Chapter 6.4).

Texture can efficiently illustrate qualitative differences, for example different soil types, land uses or hydrological units. Similar to colour hue, texture can be applied for point, linear and area map features (see Figure 1). In combination with varying colour hues, different thematic layers (topics) can be shown in one map. Quantitative differences can be portrayed in choropleth maps by applying increasing or decreasing texture densities (Chapter 3.2). However, mixing too many different texture types or changing direction of linear patterns can result in an over-complicated map design and should be avoided.

The normal map-user has a limited capacity to differentiate between a large number of colour (or grey) values or intensities. Therefore it is often necessary to classify (group) quantitative data that are to be portrayed in the thematic map. A small amount of graphic variations then appears in the map based on the reduced number of pre-defined data classes. Aggregating of map features into appropriately-defined classes increases the readability and the usefulness of the map. Distribution patterns in the landscape can be identified easier. The choice of the appropriate data classification method and the number of classes has a significant bearing on the quality of the final map. Data classification should be carried out carefully and with consideration to the data distribution and the purpose of the map. The data distribution can be checked by using histograms (Figure 3).

Figure 3.

Example of ArcGIS™ data classification interface with histogram.

The most common data classification methods are:

• Equal intervals,
• Quantiles,
• Natural breaks (Jenks),
• Geometric intervals and
• Standard deviations.

GIS or cartography software programmes (see Chapter 3.4) normally offer algorithms and standardised procedures for classification of data (Figure 3).

Natural breaks (Jenks)

The natural breaks classification method is applied by checking the data distribution (for example in a histogram or in a graph) and placing class breaks around data clusters. This avoids large value variations within one class and highlights differences between different classes. As with quantile-based classified maps, the map-creator and reader must be aware of the effects of this (sometimes subjective) data classification on the resulting map.

Figure 4 shows examples of the different classification methods applied to the same dataset. In the example, quantiles produce the most heterogeneous map but there are only minor differences for equal intervals and jenks. The classification method must be carefully selected based on the data set and the desired map product.

Figure 4.

Effects of different data classification methods on resulting maps.

An inappropriate selection and application of graphic variables or the wrong data classification method can lead to map misinterpretation (Chapter 6.4), confusion or production of poor map products. Bad choice of colours, the most powerful graphic variable, can render a map useless.

A common mistake, which has been heavily stimulated by the seemingly easy map-creation with various GIS, cartography and presentation software programmes, is the choice of too vibrant colour ramps that combine contrasting hues and go across for example red-blue-orange-green-yellow-brown colour schemes ignoring effects that different colours have on the map-user. Most map users may not be able to differentiate more than six or eight different colours within one map, depending on the map's complexity and the size of the depicted symbols. This number may be lower for people with limited colour vision. Another important consideration when creating colour maps is that colours are not recognisable if the map is reproduced in black and white or greyscale. Even if printed in colour, mismatches can occur between the printed version and the computer screen if different colour models are used.

Using bad map symbols that do not follow the logics of map semantics, syntactics and pragmatics often leads to noises in the map- maker/map-user communication (map coding; Chapter 6.4). Different cultural, societal or educational backgrounds may lead to different interpretations of symbols. A cartographer or a trained map-user will interpret a map differently from a novice map user.

Another common mapping mistake relates to the Modifiable Areal Unit Problem (MAUP; see Chapters 3.2 and 6.1). MAUP becomes especially relevant when using different colour values or intensities in choropleth maps. Additionally, the use and combination of too many different colours, patterns and symbols hamper easy and appropriate map comprehension by giving the map a 'nervous' look. Maps that are overcharged with information might run the risk of being ignored.

The map-maker needs to be aware of the finite capacity of the map-user to differentiate between the various graphic variables, especially in complex maps covering large spatial scales (Chapter 5.7). The appropriate classification of quantitative data is therefore a very important step in thematic map compilation.

When choosing from the different graphic variables shape, size, colour hue, value, intensity and texture presented above, the map-maker needs to be aware of the semantics and syntactics relevant to the choices. The semantic and psychological effects of different colours should therefore be carefully considered and maps should not be overloaded with too many different colours. When illustrating quantitative differences, the colour ramp should only have one or two colour hues (Figure 5) and the colour intensities should be adapted according to the quantitative data.

Figure 5.

Examples for single hue colour ramps.

It is also advisable to check the visibility of the selected colours after printing in black and white. Texture may be a better choice than colour hue or intensity to portray different classes in black and white maps.

Regional peculiarities need to be taken into account when compiling ES maps because of the trans-disciplinary and complex nature of ES. Involving stakeholders and harnessing their local or subject-specific knowledge can help to avoid cultural traps or misinterpretations. A stepwise process that seeks feedback from stakeholders and map-users can help to improve maps and reduce the number of map misinterpretations.

A proper map legend showing all symbols, symbol sizes, colour hues, values, intensities and orientations that appear in the map is mandatory in each map in order to read the map accordingly (see Chapter 3.1). Text in maps, where useful and appropriate, can support the information given by the map symbols. It must be legible, easy to comprehend and in the language of the map-user.

ES science involves several scientific disciplines and links multiple topics and quantification methods (Chapter 4) at various spatial and temporal scales (Chapter 5.7). Therefore ES mapping includes several challenges (see Chapter 3.7), that can be related to map semantics and syntactics.

The six key graphical variables described above are also applied in ES maps, depending on the ES to be displayed, what has to be mapped (Chapter 5.1), where (Chapter 5.2), when (Chapter 5.3) and why (Chapter 5.4). Many regulating ES (Chapter 5.5.1) can, for example, be related to natural phenomena which are often indicated by the choice of texture or orientation (e.g. for flows). Different intensities are depicted with appropriate colour hues. Provisioning ES maps (Chapter 5.5.2) often display service providing areas (Chapter 5.2), which can be point units (graphical shape) or area units (mostly displayed in choropleth maps; see Chapter 3.2). Quantities of ES supply can be portrayed by size variations (point sources, linear flows) and gradational colour values, intensities or textures. Cultural ES (Chapter 5.5.3) can be related to spatially discrete point features (e.g. iconic landmarks or religious sites displayed by map symbol shape variations) or more continuous area features (aesthetic experience based on viewsheds or landscape setting displayed by area features).

The map-makers have to take responsibility for their products as it is easy to impress or mislead map-users with colourful and attractive maps. ES maps are of high political, societal and economic relevance (Chapter 7). Therefore their compilation should closely follow the logics and the well-founded knowledge from graphic semiology. Based on the diversity of ES map-makers, map-users, the complex topics to be displayed and their high societal relevance, ES maps need to be designed with care. Well-constructed maps can properly communicate and explain complex ES phenomena.

Bertin J (1967) Semiologie Graphique. Les diagrammes, les reseaux, les cartes. With Marc Barbut [et al.]. Paris: Gauthier-Villars. (Translation 1983. Semiology of Graphics by William J. Berg).
Dent B, Torguson J, Hodler T (2008) Cartography: Thematic Map Design. 6th edition. McGraw-Hill Science/Engineering/Math.
Monmonier M (1996) How to lie with maps. 2^nd ed. The University of Chicago Press.
Muehrcke PC (2005) Map Use: Reading, Analysis, and Interpretation. 5th ed. J P Pubns.
Wood D (1992) The Power of Maps. The Guilford Press.

Mapping tools have evolved impressively in recent decades. From early computerised mapping techniques to current cloud-based mapping approaches, we have witnessed a technological evolution that has facilitated the democratisation of Geographic Information Systems (GIS). These advances have impacted multiple disciplines including ecosystem service (ES) mapping. The information that feeds different mapping tools is also increasingly accessible and complex. In this chapter, we review the evolution of mapping tools that are shaping the field of ES mapping together with the different sources of information that exist at this point. We discuss briefly the suitability of these approaches for mapping different ES types and for different scientific and policy aims. Finally, we elaborate on the integration of multiple tools (from desktop applications to sensor, web-based, or mobile devices) and on the future developments of these methods and the possibilities they may open for ES mapping.

ES mapping has achieved rapid progress in a very short time frame. To our knowledge, the first peer-reviewed ecosystem service maps were published in 1996 and, since then, a large number of ad hoc mapping studies have been conducted and a variety of tools have been developed to systematise ES mapping. The progress we have witnessed corresponds to advances in computing power, modelling and GIS, the recognition of a plurality of ES approaches (i.e., participatory mapping (Chapter 5.6.2) and biophysical modelling (Chapters 4.1 and 4.4), and the consensus that ES maps provide a direct connection between ES and the landscape and therefore with policy (Chapter 7.1).

Computing power and data availability that support GIS analysis have evolved substantially in recent years. Several freeware GIS platforms have been developed, such as QGIS (Quantum GIS), GRASS GIS (Geographic Resources Analysis Support System GIS), SAGA (System for Automated Geoscientific Analyses), and gvSIG (Generalitat Valenciana Sistema de Informacion Geografica) that provide similar functionality to the popular commercial ArcGIS software from ESRI (a list of GIS software is available here).

Specific modelling approaches for mapping ES have been developed by different institutions worldwide, resulting in a wide variety of possibilities for ES analysts' use (Table 1, also see chapter 4.4). Most of these tools are openly available to the public and are constantly evolving. Training for the potential users of these tools is of importance for their accessibility and use for decision support. The operational time necessary for their application to case studies ranges from hours (simple spreadsheet-based tools) to several months (advanced software tools).

Table 1.Download as CSV 

List of the most common ES mapping tools.

Tool	Platform	Scale2	Source
Integrated Valuation of Ecosystem Services and Tradeoffs (InVEST)	ArcGIS/Stand-alone	Municipal to provincial	http://www.naturalcapitalproject.org/invest/
Artificial Intelligence for Ecosystem Services (ARIES)	Graphical User Interface (GUI)/Web-based	Municipal to provincial	http://aries.integratedmodelling.org/
Multiscale Integrated Models of Ecosystem Services (MIMES)	Simile software	Village/farm to global	http://www.afordablefutures.com/orientation-to-what-we-do/services/mimes
Social Values for Ecosystem Services (Solves)	ArcGIS	Municipal to provincial	http://solves.cr.usgs.gov/
Land Utilisation Capability Indicator (LUCI)	ArcGIS	Village/farm to provincial	http://www.lucitools.org/
Integrated Model to Assess the Global Environment (IMAGE)	Set of models	Global	http://themasites.pbl.nl/models/image/index.php/Welcome_to_IMAGE_3.0_Documentation
Co$ting Nature	Web-based, Google Earth	Municipal to provincial	http://www.policysupport.org/costingnature
Ecosystem Valuation Toolkit	Web-based	Municipal to provincial	http://esvaluation.org/
ESM-App	Android Smartphone app	Municipal to provincial	http://www.ufz.de/index.php?en=33303

The use of GIS in ES mapping can take three general approaches: (1) analysis tools built into GIS software packages; (2) disciplinary biophysical models applied for ES assessment (e.g., hydrological models such as the Soil and Water Assessment Tool, SWAT or Variable Infiltration Capacity model, VIC for water-related ES); and (3) integrated modelling tools designed specifically for ES assessment (e.g., InVEST, ARIES). The first approach is applicable for simple land cover-based analyses and indicator-based ES mapping (see Chapter 5.6.4) that have been used for example in Mapping and Assessment of Ecosystems and their Services (MAES). The second approach is appropriate for more complex model-based analyses of services that integrate expertise from specific disciplines (e.g., ecology for crop pollination or hydrology for flood regulation mapping). The third approach extends the second one by utilising modelling tools that can assess trade-offs and scenarios for multiple services.

Several ecosystem service valuation databases have been developed as well, such as The Economics of Ecosystems and Biodiversity (TEEB) Valuation Database and the Ecosystem Valuation Toolkit and these might be used to create ES maps. The Ecosystem Services Partnership (ESP) Visualisation Tool is a database consisting of ES maps prepared by different researchers intended to promote synthesis of mapping studies (see chapter 7.9).

In-depth assessment of the different mapping tools is necessary to understand which one will best fit the user?s ES mapping context: time and data availability, mapping skills, types of services to map, accuracy required, expected impact in decision-making and overall study aims. This means that no tool fits all criteria perfectly. Some highly complex models can provide policy support in regions with considerable time, data and personnel resources. Other approaches exist that allow ES to be mapped with more limited budgets and shorter time frames. The intended use of the maps (i.e., for raising awareness or direct use in policy-making) will also influence the decision on which tools to use (see Chapter 5.6.1).

In many cases, the type of ES under assessment will determine the mapping approach or tools to use. Services such as water regulation usually require modelling approaches that integrate meteorological databases, vegetation, soils and topographic data (Chapter 5.5.1), while others such as cultural identity might require a participatory mapping approach (Chapters 5.5.3 and 5.6.2). Other services such as food production might use complex agricultural models or indicator-based approaches (Chapter 5.5.2). However, the complex nature of ES and the inter-linkages between provisioning, regulating and cultural services have led to the use of different tools for each ecosystem service.

It is also important to consider how different mapping tools account for accuracy, reliability and uncertainty. Accuracy is established through successful calibration, reliability through successful application in different contexts and uncertainty through methods that estimate and transparently communicate uncertainty. These aspects have not been adequately covered in the past and still need to be developed for several tools. Greater transparency in the presentation of results and associated uncertainties (Chapter 6) is needed so that informed decisions can be made about the extent to which ES maps can be used for different purposes and which tools are best applied in different contexts and locations.

Several challenges lie ahead for mapping ES. These are related to the progress that is currently underway in research and monitoring, remote sensing, sensor networks, data storage, data and knowledge integration, data harmonisation and sharing, database and tool maintenance and crowdsourcing, among others.

On the technical side, the accumulation of a growing quantity of data raises the challenge of effective storage and analysis of large amounts of data and is leading to an increased emphasis on machine learning, pattern recognition (in complex data or remote sensing products), and data mining. Initially high data storage requirements were addressed by large data storage and supercomputer facilities, but falling costs of distributed solutions have pushed computing towards scalable clusters of computers, grids and cloud computing, all aimed at increasing demand-driven computational power. Some ES modelling approaches using grids include: Tropical Ecology Assessment and Monitoring (TEAM) Network, Web-based Data Access and Analysis Environments for Ecosystem Services, ARIES, enviroGRIDS and biodiversity virtual e-laboratory (BioVel). The advantage of grids/clouds is that they are on-demand, self-service approaches, so the user can unilaterally obtain the necessary computing capabilities, such as server time and network storage, without having to interact with each service's provider. Cloud-based modelling tools and interfaces (e.g., OpenMI) will enable the joint development of and access to modelling and visualisation tools.

The ongoing development and maintenance of ES mapping tools (including free open-source software) require adequate funding. Further integration of ES mapping tools with policy will contribute to ongoing developments in the field and a tailored approach towards decision-making aims.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. or any other Government or by the authors of this article.

Bagstad KJ, Semmens DJ, Waage S, Winthrop R (2013) A comparative assessment of decision-support tools for ecosystem services quantification and valuation. Ecosystem Services 5: 27-39.
Bagstad KJ, Reed JM, Semmens DJ, Sherrouse BC, Troy A (2015) Linking biophysical models and public preferences for ecosystem service assessments: a case study for the Southern Rocky Mountains. Regional Environmental Change: 1-14.
Bateman IJ, Jones AP, Lovett AA, Lake IR, Day BH (2002) Applying Geographical Information Systems (GIS) to environmental and resource economics. Environmental & Resource Economics 22: 219-269.
Crossman ND, Burkhard B, Nedkov S, Willemen L, Petz K, Palomo I, Drakou E, Martin-Lopez B, McPhearson T, Boyanova K, Alkemade R, Egoh B, Dunbar MB, Maes J (2013) A blueprint for mapping and modelling ecosystem services. Ecosystem Services 4: 4-14.
Drakou EG, Crossman ND, Willemen L, Burkhard B, Palomo I, Maes J, Peedell S (2015) A visualization and data-sharing tool for ecosystem service maps: Lessons learnt, challenges and the way forward. Ecosystem Services 13: 134-140.
Eade JDO, Moran D (1996) Spatial economic valuation: Benefits transfer using geographical information systems. Journal of Environmental Management 48: 97-110.
Klug H, Kmoch A (2015) Operationalizing environmental indicators for real time multi-purpose decision making and action support. Ecological Modelling 295: 66-74.
Malinga A, Gordon L, Jewitt G, Lindborg R (2015) Mapping ecosystem services across scales and continents - a review. Ecosystem Services 13: 57-63.
Nelson E, Mendoza G, Regetz J, Polasky S, Tallis H, Cameron RD, Chan KMA, Daily GC, Goldstein J, Kareiva PM, Lonsdorf E, Naidoo R, Ricketts TH, Shaw RM (2009) Modeling multiple ecosystem services, biodiversity conservation, commodity production and tradeoffs at landscape scales. Frontiers in Ecology and the Environment 7(1): 4-11.
Schroter M, Remme RP, Sumarga E, Barton DN, Hein L (2015) Lessons learned for spatial modelling of ecosystem services in support of ecosystem accounting. Ecosystem Services 13: 64-69.
Stoll S, Frenzel M, Burkhard B, Adamescu M, Augustaitis A, Bae?ler C et al. (2015) Assessment of ecosystem integrity and service gradients across Europe using the LTER Europe network. Ecological Modelling 295: 75-87.

Ecosystems are defined by the UN Convention on Biological Diversity (CBD) as 'a dynamic complex of plant, animal and micro-organism communities and their non-living environment interacting as a functional unit'. Ecosystems are therefore, by definition, multi-functional. Each ecosystem provides a series of services for human well-being either directly, for example, as food and fibre, or more indirectly by, for example, providing clean air and water, preventing floods or providing recreational or spiritual benefits.

Ecosystems contain a multitude of living organisms that have adapted to survive and reproduce in a particular physical and chemical environment, i.e. their natural condition. Anything that causes a change in the physical or chemical characteristics of the environment has the potential to change an ecosystem's condition, its biodiversity and functionality and, consequently, its capacity to provide services. Up to the present, ecosystem service (ES) assessments have been based on ecosystem extent and spatial distribution as basic input parameters. The inclusion of condition assessment would add value in terms of ecosystem quality. The provision of timber, for example, not only depends on the availability of forests, but also on the species composition and age class distributions of the forests. Pollination services might be highest in grass- and croplands but are also highly influenced by plant species diversity of these ecosystems which again are also triggered by nutrient content and management.

The concept of ecosystem mapping and conditions assessment can be applied at all spatial and temporal scales. Spatial explicitness is important to characterise ecosystems in terms of their natural conditions determined by climate, geology, soil properties, elevation etc. and, in terms of their physical and chemical conditions, how they are influenced by anthropogenic pressures. Local or regional assessments require more detailed information for adequate decision support. Usually national and continental mapping is less detailed but provides important information at the strategic level. In any case, active stakeholder involvement is recommended to design and adapt the assessments for successful implementation into the decision process.

Should no map of ecosystems or habitats be available, a feasible proxy has to be developed as shown in Figure 1. The basic geometry and main classes in appropriate spatial resolution can be derived directly from satellite images (see e.g., http://www.earthobservations.org/geoss.php) or from existing land cover / land use maps.

Figure 1.

Work flow for ecosystem mapping and condition assessment.

For policy-relevant information, the map should be re-classified using an ecosystem typology which represents the most important types of their human management to make best use of their services, e.g. by agriculture, forestry, fisheries, water management, nature protection or territorial planning. These management lines are also usually implemented in the respective legislations which are important cornerstones in the decision-making process. In case further geometric refinements are required, these can be performed by, for example, integrating more detailed information about rivers and lakes, green linear elements such as hedgerows or detailed maps of urban areas or protected areas.

If needed, such a basic map can be further refined thematically by providing more detailed information about the natural characteristics of the ecosystems and their biodiversity. GIS-based, so-called envelope- or niche-modelling as developed for habitat or climate change impact studies, allows the combination of non-spatially referenced species or habitat information with a set of environmental parameters such as elevation, soil, geology, climate, phenology, potential natural vegetation etc. to delineate the most likely areas of ecosystem presence.

This probability mapping of ecosystem presence depends on the accuracy of the descriptors of its natural boundaries (e.g. alpine meadows or calcareous broadleaf forests) and the availability and quality of the respective data to delineate and map these boundaries. Further enhancement can be performed by attributing statistical information e.g. crop yields or forest inventory data to the respective ecosystem classes.

Mapping of ecosystem types provides information on the natural conditions. To assess the current capability of ecosystems to provide services for human well-being requires information about their current conditions which are induced by human activities.

For decision support, the most comprehensive and informative approach for the assessment of ecosystem conditions should include direct mapping and assessments in combination with information about the direct and indirect pressures which induce these conditions. This approach provides information on both the current environmental state and expected changes due to constant, increasing or decreasing pressures. Additionally, important information for risk assessments can be derived. Time lags between pressures and changes in ecosystem conditions are often triggered by buffering processes which indicate the resilience of species and ecosystems to the different types of stress factors affecting their condition.

For better understanding of the different processes affecting ecosystem condition and the link to human activities, the DPSIR (Drivers, Pressures, State or Condition, Impact, Response) approach is often used (Figure 2). Drivers to cover our demand for ES and other natural resources induce pressures which affect ecosystem conditions. The impacts should create (policy) responses which should again change the drivers and the way we manage our environment to cope with negative impacts. The DPSIR approach should be considered not as absolute but relative to the ecosystem processes under consideration. The nutrient conditions of agro-ecosystems, for example, are the pressures for freshwater ecosystems and both conditions are pressures for marine ecosystems.

Figure 2.

DPSIR framework for assessing ecosystem condition.

Pressures affect ecosystem conditions either by concentration (e.g. ozone) or by accumulation (e.g. nitrogen and pollution load). The Millennium Ecosystem Assessment 2005 identified five different anthropogenic main pressures affecting ecosystem conditions: habitat change, climate change, invasive species, land management and pollution/nutrient enrichment.

Human pressures are either direct, i.e. mainly from land use, or indirect, i.e. by air pollution or anthropogenic climate change. Important for ecosystem conditions are the strength of the pressure signal, its persistence if cumulative and its change over time. Time-series of observed changes in pressures are, therefore, important to analyse the causal connectivities between pressures and current condition for each ecosystem type and each spatial unit. The trend in pressures also provides a first insight into the expected changes in the near future. Decreasing observed trends may indicate further improvement of ecosystem conditions and vice versa, i.e. important information for decision-making about measures to mitigate and adapt to positive or negative effects.

Box 1. European ecosystem map

For the implementation of the EU Biodiversity Strategy to 2020, Member Countries and European institutions perform ecosystem assessments in their territories. The European map is based on a proxy following the scheme as outlined in Figure 1 based on Corine land cover (CLC). Geometry of the basic map was further refined using the High Resolution Layers of Copernicus land services² and re-classified to eight aggregated ecosystem types: urban, cropland, grassland, forests and woodland, heathland and shrubs, sparsely vegetated land, wetlands and rivers and lakes. For Europe's seas, only a very simplified classification mainly based on EUSeaMap sea-floor mapping and bathymetry data is currently implemented. The basic version was thematically enhanced using the non-spatially referenced habitat information of the European Nature Information System (EUNIS) database in combination with a set of environmental parameters to delineate the most likely areas of ecosystem presence.

Figure 3.

Ecosystem map of Europe Version 2.1 (higher resolution map can be downloaded at: http://www.eea.europa.eu/data-and-maps/data/ecosystem-types-of-europe).

In practice, information on ecosystem conditions is often insufficient for appropriate mapping and assessments. Another problem is the mapping and assessment of the combined effects of pressures on the ecosystem condition. Usually spatially explicit maps of the different pressures and their gradients across the area under investigation can be produced but knowledge about the combined effects on biodiversity and ecosystem structure and function is still insufficient. So in many cases, proxy indicators have to be used to indicate the current ecosystem condition as illustrated in Box 2.

Box 2. Mapping ecosystem condition

Figure 4 shows an example how the ecosystem condition can be derived by combining ecosystem mapping, reported data of the European Habitat Directive and statistical data. The combination of information in different units often requires re-scaling from absolute to relative values, e.g. from 'low' to 'high'.

Figure 4.

Map of European cropland conditions.

Ecosystem type mapping and condition assessments have to be further improved making use of new information and data flows from research, reporting and other sources.

A major issue is the lack of detailed information on how the ecosystem condition affects ecosystem service delivery. The delivery of ecosystem services depends on the biological, physical and chemical processes and the biodiversity involved (Chapters 2.2 and 2.3) but there are few quantitative data to model and assess how these processes and functional traits are affected by pressures such as pollution, management or climate change and their combined effects. Further research is needed to fill these gaps and improve our knowledge about the relationships between pressures - ecosystem conditions, related biodiversity and ecosystem service capacity.

De Groot RS, Alkemade R, Braat L, Hein L, Willemen L (2010) Challenges in integrating the concept of ecosystem services and values in landscape planning, management and decision making. Ecological Complexity 7: 260-272.
EEA (2016) Mapping and assessing the condition of Europe's ecosystems: progress and challenges. EEA report 03/2016 http://www.eea.europa.eu/publications/mapping-europes-ecosystems accessed 31 May 2016.
Harrison PA, Berry PM, Simpson G, Haslett JR, Blicharska M, Bucur M, Dunford R, Egoh B, Garcia-Llorente M, Geam N, Geertsema W, Lommelen E, Meiresonne L, Turkelboom F (2014) Linkages between biodiversity attributes and ecosystem services: A systematic review. Ecosystem Services 9: 191-203.
Maes J, Teller A, Erhard M et al. (2013) Mapping and Assessment of Ecosystems and their Services. An analytical framework for ecosystem assessments under action 5 of the EU biodiversity strategy to 2020. Publications office of the European Union, Luxembourg http://ec.europa.eu/environment/nature/knowledge/ecosystem_assessment/pdf/MAESWorkingPaper2013.pdf accessed 12 December 2015.
MAES information platform: Mapping and Assessment of Ecosystems and their Services (MAES): http://biodiversity.europa.eu/maes accessed 31 May 2016.
Millennium Ecosystem Assessment (2005) Ecosystems and human well-being: Synthesis, Millennium Ecosystem Assessment, Island Press, Washington, DC, USA: http://www.maweb.org/en/index.aspx accessed 31 May 2016.
Potschin M, Haines-Young R, Fish R, Turner RK (Eds.) (2016) Routledge Handbook of Ecosystem Services. Routledge London and New York 629 pp.

Landscape metrics have been used to derive indicators in landscape ecology and related disciplines for decades. More than one hundred metrics have been developed for the purpose of describing processes and landscape functions in the form of mathematical terms. After a very enthusiastic time, the focus at present is on meaningful, simpler measures that can be applied in practice. Meanwhile, landscape metrics play a crucial role not only in science, but also in practical issues, such as spatial planning or biodiversity monitoring. Most frequently applied metrics are used to discover biodiversity or landscape fragmentation. Although great advances have already been made, new metrics continue to be developed. Regularly used metrics are further tested and updated regarding their interpretation. This subject is not without controversy. A question which is often raised in research circles is:

Which role can landscape metrics play within the set of indicators for ES mapping and assessment?

The following sections address this question. Regarding the ES cascade model (Chapter 2.3) and following on from it, landscape structures support biodiversity and ecosystem functions that are the basis for the final provision of ecosystem services (ES) to humans. The crucial question is whether landscape metrics applied to land use / land cover maps can provide direct or indirect indications on the provision of ES.

So far, landscape metrics have been applied to indicate cultural ES (e.g. recreation, landscape aesthetics) and regulating ES (e.g. soil erosion, biological pest control). However they are predominantly applied to measure ecological functioning (biodiversity, connectivity, soil quality) and land use processes (land consumption, fragmentation, urban sprawl).

Within this chapter, we review the know-ledge of pattern-related challenges in ES mapping, using the examples of habitat connectivity and scenic attraction. We contribute to a better understanding of the reasons for challenges in mapping structure-dependent ES and we demonstrate some methods for addressing them.

Landscape metrics are tools which can be used to bridge the methodological gap between landscape structure and ES provision. They take the visible spatial manifestation of land use patterns into account. Composition and configuration of patches (homogeneous units of one property, e.g. land use type) are key features of maps. Hence, landscape metrics and mapping are inherently interrelated. Table 1 provides an overview of selected landscape metrics which are applicable for mapping and assessment of ES.

Table 1Download as CSV 

Examples for suitable landscape metrics indicating biodiversity and ES (provisioning, regulating, cultural; following CICES (2013)), without claim to completeness.

Structure/landscape metric	Process/function	Mapping target
		Dimension of Biodiversity
Shannon's diversity index, Patch density ?	Pattern heterogeneity and variety ?	Landscape diversity
Shape index ?	Natural conditions ?	Species diversity
Proximity index, Nearest neighbour index ?	Isolation, Habitat connectivity ?	Species diversity
Effective mesh size ?	Fragmentation ?	Species diversity
		Provisioning service
Total patch area (of arable land) ?	Food and fodder production ?	Food and fodder
Total patch area (of forested/arable land) ?	Biomass production ?	Biomass
Total patch area of lakes ?		Food (fish)
		Regulating service
No. / length of landscape elements (hedges, tree lines) ?	Soil erosion due to water runoff ?	Mass flow
Edge length (of hedges, forests and other ecotones) ?	Habitat provision forpollinators (fringe structures) ?	Pollination
Shannon's diversity index / Heterogeneity of agricultural areas ?	Population development ?	Pest control
		Cultural service
Total patch area (of water), Edge length of waters ?	Attraction, Complexity ?	Landscape aesthetics
Shape index Hemeroby index ?	Complexity and Natural conditions ?	Landscape aesthetics
No. of landscape elements ?	Legibility, mystery ?	Landscape aesthetics

Landscape metrics quantify physical landscape structures which themselves determine processes and functions. Although some landscape structures can be measured and related to the provision of specific ES, direct functional interpretation of single metrics regarding ES remains limited. For the assessment and interpretation of landscape metrics, for example, a normative assessment basis is required which relates the current situation of landscape structure to a reference or target situation of ES provision. However, landscape structure is important information in a more complex evaluation of ES. Landscape metrics have therefore to be considered as meaningful parameters together with others in ES mapping and evaluation. A sound application of landscape metrics is possible considering two dimensions of biodiversity, species and landscape diversity. Many species and species communities rely on specific landscape structures or landscape elements and their interrelations.

Provisioning services strongly depend on the extent of managed land and the land use intensity. However, for quantification of productivity or food provision, further information, for example on soil quality or soil management, is essential to derive a valid estimation on food provision. Regarding regulating services, landscape metrics also comprise the potential to provide supplementary information. Indices like edge length or the number of landscape elements can quantify some preconditions for functions and services. Although modelling and/or measurement of species abundance or mass flows (qualitative data; Chapter 4.1) cannot be replaced by structural indicators, they have to be considered as one important part of the required information.

Strong interrelations between indices of biodiversity and landscape aesthetics can be identified as potential of landscape metrics application in mapping ES. Cultural services such as the potential of landscapes for human recreation are interrelated with structural aspects. However, landscape aesthetics is just one of many spiritual, experiential and educational services. Landscape metrics can therefore serve as a complementary mapping and assessment method for cultural ES.

Landscape metrics have been applied in several ES mapping and assessment studies. Two examples illustrate how they can be related to ecological integrity, considered as the basis for any ES provision (Box 1) and scenic attraction, as an example for cultural ES (Box 2). In a similar manner, spatial structures strongly determine regulating ES. The regulation of soil erosion, for example, can be estimated using the number and the spatial arrangement of landscape elements, such as hedgerows. These elements reduce slope lengths which is one driving factor for soil erosion. However, suitable landscape metrics, such as the number of patches or edge length, have not been frequently used for assessment and mapping of regulating ES.

Box 1. Example for the application of landscape metrics at the regional scale: evaluation of the landscape structures' impact on biodiversity

One approach, how landscape metrics can contribute to ES mapping and assessment, is illustrated in Figure 1. Six metrics were applied to an administrative planning region (3,434 km?) in middle Saxony, Germany to evaluate ecological integrity as the precondition for biodiversity and the cultural service landscape aesthetics. They were implemented into the land use change simulation software GISCAME as supplementary indicators. In this software, the basic evaluation of ES is based on land use types. An additional landscape structure add-on makes the impact of composition and configuration visible and assessable.

The example focuses on habitat connectivity which is a landscape function related to biodiversity (see Table 1). Using the "moving window" method, which is independent from any administrative or geographical zoning, combined with a cost-distance analysis, local landscape pattern were examined across space and interpreted. The size of the moving window is determined by the action radius of a target species. On the basis of the degree of hemeroby of land use types, near-to-nature areas were identified, as well as core habitat areas and functionally connected areas. The latter were defined as potential habitat areas which are too small and not compact enough to provide habitat core areas. Nevertheless, they are close enough to another core area to be appropriate habitats for species moving through a landscape. Such functionally connected areas were also considered as part of the habitat network. (Semi-)natural areas were considered as isolated and therefore not contributing to the habitat network if they were separated by roads, urban areas and similar land use types acting as barriers.

The map in has been classified according to the functional interpretation of a land use map. It can serve scientists as well as spatial planners to identify i) the share of land which contributes to a habitat network and ii) its spatial distribution. This information allows spatially explicit conclusions on priority areas for enhancement of the connectivity and on the overall state of habitat connectivity as one influencing factor of biodiversity.

Box 2. Example for the application of landscape metrics at the national scale: application of ES to estimate the cultural ES scenic attraction

Based on the natural amenities and features, a model for assessing the scenic attraction of landscapes is presented. It is a suitability analysis of an area for nature-based recreation, assuming that certain features of the landscape have a positive or negative impact on the attraction of the landscape and recreation. In this model, landscape metrics are used for several parameters. The relief diversity, the proportion of open space, the hemeroby Index, the density of forest-dominated ecotones, the density of water edges (without coasts), the coastlines and the proportion of unfragmented open space greater 50 km? were selected.

The relief diversity (ratio 3D / 2D) reflects not only the maximum height difference (relief energy), but also the cumulative differences in altitude. A low proportion of open space indicates urban or densely built-up areas which can decrease the natural attraction of the landscape by the strong influence of technical artefacts. In congruence with the hemeroby index, the natural condition is an important factor for the attraction of landscapes. With the density of ecotones dominated by trees and shrubs and the density of water edges, landscape diversity and structure are taken into account. This parameter characterises mainly the variety and edge effects. Since the coasts play a very important role in terms of attraction and recreation, they are represented by their own parameters - coastlines. Finally, the disturbing effect of fragmentation by the transport network is considered with the parameter 'proportion of unfragmented open spaces greater than 50 km?'.

All data used were based on the official land use data of the state and federal German survey authorities (ATKIS Basis DLM or land cover model LBM-DE) in vector format collected in 2010. The indicator of the scenic attraction was calculated based on a 5-km grid (standardised according to EU INSPIRE directive).

To determine the five classes of scenic attraction, the standard deviation from the nationwide average was used. The reason behind this approach is mainly to use no fixed scale, but starting from the average values of the scenic attraction, to be able to make statements as to whether an area is rather less, or rather more scenically attractive. Landscapes which are significantly affected by anthropogenic impacts and thus often are particularly fragmented, intensively farmed or settled, can be found in the class "less attractive". Average attractive landscapes already meet recreational functions in a regional context, while very or particularly attractive landscapes represent targets for nature-related tourism and are mostly well-known nationwide.

Monitoring the development of scenic attraction using this aggregated indicator would provide decision-makers with indications as to where the scenic attraction is particularly reduced or has improved. The information derived from the aggregated, landscape metrics-based indicator reveals that individual changes affect the landscape values in their sum. Furthermore, spatial information on the scenic attraction can be used to avoid encroachments in scenic highly attractive areas and thus to achieve better management.

Furthermore, small-scale landscape elements such as ecotones at forest borders, single-trees, hedgerows including field margins are important for the regulating of ES pollination.

Landscape metrics and ES mapping are inherently related topics since landscape metrics quantify spatial characteristics of landscape patterns. Therefore, we recommend the application of landscape metrics in the context of ES mapping and also ES assessment. These indices have the power to support the identification and monitoring of spatial characteristics of landscapes which have implications on the perfor mance of biodiversity and several ES.

Some dimensions of biodiversity and cultural ES can be comprehensively indicated by landscape metrics.

The validity and verifiability of landscape metrics, however, is limited. They quantify and illustrate processes and/or functions, which can serve as surrogates for specific ES. Due to such indirect links to ES, landscape metrics should only be used as supplementary indicators for ES assessments. In the case of landscape aesthetics and recreation, they can be more directly linked to the ES provision. However, landscape metrics describe structural aspects of ES (which are important and should not be forgotten), but usually additional information (e.g. data on quality of land use) is necessary. Still, they have a great significance in terms of mapping.

The spatial interpretation of land use maps with the help of landscape metrics serves as a valuable method for communicating ES-related issues. With regard to the current application for ES mapping and assessment in science and practice, we foresee a large capacity for future application of landscape metrics, especially in practice. The benefit of using landscape metrics for ES mapping is currently below its estimated potential.

Burkhard B, Kandziora M, Hou Y, Muller F (2014) Ecosystem Service Potentials, Flows and Demands - Concepts for Spatial Localisation, Indication and Quantification. Landscape Online 34: 1-32.
Botequilha Leitao A, Ahern J, McGarigal K (2006) Measuring Landscapes. A Planner's Handbook. Island Press; 2nd edition.
Frank S, Furst C, Koschke L, Makeschin F (2012) A contribution towards a transfer of the ecosystem service concept to landscape planning using landscape metrics. Ecological Indicators 21: 30-38.
Haines-Young R, Potschin M (2013) Common International Classification of EcosystemServices (CICES), version 4.3. Report to the European Environment Agency EEA/BSS/07/007 (download: www.cices.eu).
McGarigal (2015) FRAGSTATS HELP. V. 4.2, 21 April 2015, available online: http://books.umass.edu/landeco/research/fragstats/documents/fragstats.help.4.2.pdf
Walz U, Stein C (2014) Indicators of hemeroby for the monitoring of landscapes in Germany. Journal for Nature Conservation 22(3): 279-289.
Walz U (2015) Indicators to monitor the structural diversity of landscapes. Ecological Modelling 295: 88-106.

Abson DJ, von Wehrden H, Baumgartner S, Fischer J, Hanspach J, Hardtle W, Heinrichs H, Klein AM, Lang DJ, Martens P, Walmsley D (2014) Ecosystem services as a boundary object for sustainability. Ecological Economics 103: 29-37.
Crossman ND, Burkhard B Nedkov S, Willemen L, Petz K, Palomo I, Drakou EG, Martin-Lopez B, McPhearson T, Boyanova K, Alkemade R, Egoh B, Dunbar MB, Maes J (2013) A blueprint for mapping and modelling ecosystem services. Ecosystem Services 4: 4-14.
Dick J, Maes J, Smith RI, Paracchini ML, Zulian G (2014) Cross-scale analysis of ecosystem services identified and assessed at local and European level. Ecological Indicators 38: 20-30.
Hauck J, Gorg C, Varjopuro R, Ratamaki O, Maes J, Wittmer H, Jax K (2013) Maps have an air of authority: Potential benefits and challenges of ecosystem service maps at different levels of decision making. Ecosystem Services 4: 25-32.
Martnez-Harms MJ, Balvanera P (2012) Methods for mapping ecosystem service supply: A review. International Journal of Biodiversity Science, Ecosystems Services and Management 8: 17-25.

Ecosystem services (ES) arise when ecological structures and ecological processes directly or indirectly contribute to human well-being and meet a certain demand from people. This flow of ES from ecosystems to society is well represented by the ES cascade concept (see Chapter 2.3). Ecosystems provide the necessary structure and processes that underpin ecosystem functions which are defined as the capacity or potential to deliver services. ES are derived from ecosystem functions and represent the realised flow of services in relation to the benefits and values of people. This model is useful for quantifying ES. Consider the following example: wetlands (an ecosystem or a structure) provide habitat for bacteria which break down excess nitrogen (denitrification, a process). This results in the removal of nitrogen from the water (a service) resulting in better water quality (a benefit). People can value increased water quality in multiple ways (e.g., by expressing their willingness to pay for clean water). Each of these different steps can be quantified using biophysical, economic or social valuation methods.

This chapter focuses on biophysical quantification which is the measurement of ES in biophysical units. Biophysical units are used to express, for example, quantities of water abstracted from a lake, area of forest or stocks of carbon in the soil. Looking at the ES cascade, it seems evident that biophysical quantification focuses, in particular, on the measurement of ecosystem structures, processes, functions and service flows (also known as the left side or the supply side of the cascade). Benefits and values (also known as the right side or demand side of the cascade) are more often measured using social (see Chapter 4.2) or economic units (see Chapter 4.3). Nonetheless, benefits and values can sometimes be expressed in biophysical units as well. Consider again the above example of water purification in wetlands. The benefit from this ecosystem service is clean water and this can be expressed as the concentration of pollutant substances.

To quantify ES along the different components of the ES cascade, we need to address two questions: what do we measure and how do we measure (Figure 1)? For the purpose of this chapter, we assume that the question as to why we measure (e.g., policy questions, scope of an ecosystem assessment) has been answered.

Figure 1.

Biophysical quantification of ecosystem services (Icons by Freepik).

The first question is addressed in the scientific literature by developing and proposing indicators. Ecosystem service indicators are used to monitor the state or trends of ecosystems and ecosystem service delivery within a determined time interval. In recent years a substantial indicator base has been developed world wide to assess or measure ES.

Once an indicator is proposed or selected for inclusion in an ecosystem assessment, the second question becomes important: how can we measure the service or the indicator in biophysical terms or units? Which methods or procedures should be applied to come to an reasonable estimate of the quantity of service provided?

ES indicators are information that efficiently communicates the characteristics and trends of ES, making it possible for policy-makers to understand the condition, trends and rate of change in ES.

Different indicators can be used to measure or indicate a single ecosystem service. The choice for an indicator depends on many factors including the purpose, the audience, its position on the ES cascade, the spatial and temporal scale considered and the availability of data.

Purpose and target audience are important criteria for selecting or designing indicators for ES. It makes a difference if indicators are used to inform policy makers, journalists, conservation and land managers, scientists or students. Not everybody has an equal understanding of the flow of ES which is indeed a relatively complex concept. Therefore, indicators are sometimes expressed in relative terms by setting a reference value equal to, for instance, 100 and by calculating other values relative to this reference. This facilitates interpretation for some user groups. Of equal importance is the purpose of an indicator. Why is it used? Many ES indicators are proposed to report the state and trends of ES under different biodiversity policies from global to local scale. But such indicators are not necessarily useful for application by spatial planners or for scientific support to river basin management. Consider pollination, a regulating ecosystem service. A scientist could be interested in the diversity and density of different bee and bumblebee populations; a farmer may wish to know how far he can rely on wild pollination to help pollinate his fruit trees; a biodiversity policy officer may need to know if, at national scale, pollination services are declining or increasing. Clearly, these stakeholders have different information requests which require different indicators with different biophysical units although pollination is the common denominator.

The above example also illustrates the importance of spatial and temporal scales. The issue of scale is frequently presented in all textbooks on ecology as biodiversity and the ecological processes it supports (and thus also the delivery of ES) are heavily dependent on time and space. Processes are influenced by different time cycles (day-night, seasons) and take place at different rates (see also Chapter 5.3). The self-purifying capacity of water is, for instance, highly dependent on the velocity at which water flows. Water purification services, for example, which can be measured by the amount of pollutant removed, differ between fast running streams and stagnant lakes with the latter ecosystems having, in general, a higher capacity (more time) to remove nitrogen but a lower capacity to clean organic pollution. Also spatial scale matters. Bees and bumblebees deliver their pollination services within a distance of a few hundred metres whereas the storage of carbon in trees operates at almost global scale. Indicators and, in particular, their units of measurement have to consider the scale at which ES are relevant. Sometimes indicators are designed to be scale independent. This means they can be upscaled or downscaled, a very useful technique for mapping.

An important question often raised in literature on ES is: should indicators measure the stock and the flow? A service flow refers to the actual use of the actual benefits people receive from ecosystems. A stock refers to the capacity of ecosystems to deliver those benefits. Flows are always expressed per unit of time. Timber production serves as a good example to illustrate the difference between an indicator which measures the stock and an indicator which measures the flow. Timber production is often measured by quantifying the harvest (how much timber is cut, usually expressed in a volume of wood per unit area and per unit of time, for example, m³/ha/year). Sometimes timber production can also be indicated by the available timber stock which can be harvested. This difference is subtle for the case of timber. If the stock is harvested, stock becomes flow. However, for other services, the difference between stock and flow is important because indicators for stock and flow cannot always be expressed in the same units. Wetlands have a certain capacity to clean water but it is not always straightforward to express this capacity in terms of pollutant removal (e.g., amount of nitrogen removed or immobilised in the sediment in kg/ha/year). Often the size of the wetland (in ha) is used as proxy to indicate this capacity. The rationale is that larger wetlands have more capacity to purify water than smaller wetlands. In this context, the concept of ecosystem condition is important as well (see Chapter 3.5). Not only the quantity (spatial extent) of an ecosystem is important to assess the physical values of ES capacity, ecosystem quality or ecosystem condition is also an important determinant of ecosystem delivery. Changes in ecosystems through degradation can thus alter the flows of ES and should thus be measured as well by indicators.

A final remark on indicators relates to composite indicators or indices which aggregate different sorts of information into a single number. Usually such indicators are made for specific purposes or to inform on particular challenges with a single value. In a similar context for ES, such indicators exist but usually they are composed of normalised versions of indicators for single services which are summed or aggregated. They cannot be quantified directly but depend on separate quantification of their individual components.

This chapter does not provide a list with indicators for ES for the simple reason that there are hundreds of indicators available. Many countries and regions have developed ES indicator sets; the setting of global or regional biodiversity targets has also spurred the development of indicators. Furthermore, the application of the ES concept for planning, natural resources management and conservation has created additional indicators. Therefore we list in Table 1 some important initiatives where readers can find a selection of indicators, organised from global to sectorial initiatives.

Table 1.Download as CSV 

Examples of sources, websites and key publications for ecosystem service indicators.

Scale	Location	Publication
Global		Measuring Nature's Benefits: A Preliminary Roadmap for Improving Ecosystem Service Indicators (http://pdf.wri.org/measuring_natures_benefits.pdf) http://www.bipindicators.net/ (report ISBN 92-9225-376-X) Measuring ecosystem services: Guidance on developing ecosystem service indicators (ISBN: 978-92-807-4919-5) http://es-partnership.org/community/workings-groups/thematic-working-groups/twg-3-es-indicators/; A Global System for Monitoring Ecosystem Service Change (doi: 10.1525/bio.2012.62.11.7)
Sub-global	European Union	website: http://biodiversity.europa.eu/maes/mapping-ecosystems; article: doi: 10.1016/j.ecoser.2015.10.023
National	Finland	website: http://www.biodiversity.fi/ecosystemservices/home; article: doi: 10.1016/j.ecolind.2015.03.041
	Canada	Website: https://www.ec.gc.ca/indicateurs-indicators/
	Switzerland	Website: http://www.bafu.admin.ch/publikationen/publikation/01587/index.html?lang=en
	Germany	article: Towards a national set of ecosystem service indicators: Insights from Germany (doi: 10.1016/j.ecolind.2015.08.050)
	Spain	Website: http://www.ecomilenio.es/informe-de-resultados-eme/1760; Article: doi: 10.1371/journal.pone.0073249

In summary, ES indicators express what to measure when quantifying ES in a biophysical manner. Good ES indicators come with information on their place on the ES cascade, on the available data, on the targeted audience and the objective and on whether they assess a stock or a flow.

Indicators must be measured but how is this done for ES? Some of the above given examples already provide the answer. The number of bees on a farmland, the timber harvest from a forest or the denitrification in a wetland can all be monitored or measured with different methods or devices. Yet measuring stocks or flows of ES is less evident than it seems. Here we present three approaches which can be considered to quantify biophysical stocks and flows of ES: direct measurements, indirect measurement and (numerical) modelling.

Direct measurements of an ecosystem service indicator is the actual measurement of a state, a quantity or a process from observations, monitoring, surveys or questionnaires which cover the entire study area in a representative manner. Direct measurements of ES deliver a biophysical value of ES in physical units which correspond to the units of the indicator. Direct measurements quantify or measure a stock or a flow value. Direct measurements are also referred to as primary data.

Examples of direct measurements of ES (see also Table 2) are counting the number of visitors visiting a national park (nature based recreation); measuring the total volume of timber in a forest stand (timber production); monitoring the release of nitrous oxides of a reed bed or deposition of sulphur dioxide on leaves (water and air filtration); recording the crop yield of a farm (crops); measuring the volumetric capacity of a flood plain (flood control); monitoring over time the improvement of water quality (water purification); measuring the abstraction of water from ground water layers (water provision) or asking citizens how many times they visit a forest to pick berries, mushrooms or chestnuts (wild food products). When the spatial extent or relative surface area of ecosystems is used to approximate ES, also botanical and forest inventories, permanent plots or any other direct observation on the terrain can be used as proxy. In certain cases remote sensing can be considered also as direct measurement.

Table 2.Download as CSV 

Examples of different methods to measure ecosystem service indicators

Ecosystem services	What to measure	How to measure (method)
(CICES class)	Indicator	Direct	Indirect	Model
Cultivated crops	Crop yield (tonne/ha/year)	Crop statistics (obtained through official reporting)	Remote sensing of crop biomass using NDVI and aerial photo analysis for long temporal changes Coupling structural observations with remote sensing information	Crop production models
Reared animals and their outputs	Livestock (heads/ha)	Livestock statistics (head counts obtained by reporting)
Wild plants, algae and their outputs	Wild berry yield (tonne/ha/year)	Field observations and surveys of people harvesting wild fruits		Species distribution models; ecological production model
Animals from in-situ aquaculture	Fish yield (tonne/ha/year)	Aquaculture statistics (obtained through official reporting)		Fish production models
Water (Nutrition)	Water abstracted (m3/year)	Water statistics (obtained through official reporting)	Remote sensing of water bodies and soil moisture	Water balance models
Biomass (Materials)	Timber growing stock (m3/ha) and timber harvest (m3/ha/year)	Forest stand measurements and forest statistics	Remote sensing of forest biomass using NDVI	Timber production models
(Mediation of waste, toxics and other nuisances)	Area occupied by riparian forests (ha)	Site observations	Earth observation land cover data
	Nitrogen and Sulphur removal in the atmosphere or in water bodies (kg/ha/year)	Measurement of deposition of NO2 and SO2; field measurement of denitrification in water bodies	Remote sensing of canopy structure (leaf area index)	Transport and fate models for N and S
Mass stabilisation and control of erosion rates	Soil erosion risk (tonne/ha/year)	Field measurements of soil erosion		Soil erosion models (RUSLE)
Flood protection	Area of floodplain and wetlands (ha)	Site observations	Elevation models and data; aerial photo analysis; remote sensing of land cover	Modelling water transport
Pollination and seed dispersal	Pollination potential; number and abundance of pollinator species (number/m2)	Field sampling of pollinator species; counts of bee hives		Species distribution models; ecological modelling of habitat suitability
Decomposition and fixing processes	Area of nitrogen fixing crops (ha)	Field surveys; crop statistics (obtained through official reporting)		Crop production models
Global climate regulation by reduction of greenhouse gas concentrations	Carbon storage (in soil or aboveground biomass) (tonne/ha); carbon sequestration (tonne/ha/year)	On-site measurements of carbon stock and carbon fluxes	Remote sensing of vegetation	Carbon cycle models
Physical and experiential interactions	Visitor statistics (number/year)	Visitor data and questionnaires of visitors	Monitoring parking lots, mapping trails or camping sites	Modelling potential use of nature reserves by people

These examples of direct measurement share a number of characteristics. They are time and resource consuming and thus costly, mostly suitable for carrying out at site level or local scale and they measure tangible flows of ES, in particular for provisioning ES. Direct measurements are also feasible in case of a clearly defined service providing species (or areas) such as pollination, bird watching or biological control.

As many of these indicators are effectively measured for other reasons, it is not always needed to set up expensive measurement schemes. Most provisioning ES including crops, fish, timber and water are recorded by national and regional governments. Furthermore, certain species groups and taxa are monitored to assess trends in biodiversity.

TESSA (http://tessa.tools/) is a toolkit for rapid assessment of ES at site level which provides many procedures and suggestions for on-site measurement of ES.

Direct measurements and the use of primary data are the most accurate way to quantify ES but they become impractical and expensive beyond the site level or they are simply not available for all ES.

Therefore the next step to consider for biophysical quantification is indirect measurements.

Indirect measurements of ES deliver a biophysical value in physical units but this value needs further interpretation, certain assumptions or data processing, or it needs to be combined in a model with other sources of environmental information before it can be used to measure an ecosystem service. Indirect measurements of ES deliver a biophysical value of ES in physical units which are different from the units of the selected indicator.

In many cases, variables that are collected through remote sensing qualify as indirect measurement. Examples for terrestrial ecosystems are land surface temperature, NDVI (Normalised Difference Vegetation Index), land cover, water layers, leaf area index and primary production. Examples for marine ecosystems include sea surface temperature, chlorophyll A concentration and suspended solids. Many of these data products do not measure stocks or flows of ES but they are highly useful to quantify global climate regulation as well as all those ES which depend directly on the vegetation biomass of ecosystems to regulate or mediate the environment. Soil protection and water regulation, for example, are strongly driven by the presence of vegetation which can be inferred from earth observation datasets. Local climate regulation can be inferred from spatially and temporally explicit patterns of surface temperature. Air filtration by trees and forest is directly related to the canopy structure which, in turn, can be measured by the leaf area index. In addition, micro-climate regulation in cities (temperature reduction during heat waves through evapotranspiration and provision of shade) can be approximated by measuring the total surface area of urban forest.

A specific role is reserved for land cover and land use data which are used for both direct and indirect quantification of ES. Detailed and accurate information on the extent of ecosystems or of ecosystem service providing units, constitute an essential data basis for all ecosystem assessments. Importantly, land data can also be used to quantify demand for ES.

Not all indirect measurements are provided by earth observation. The density of trails and camping sites may provide an indirect estimate of recreation and tourism (Table 2).

Indirect measurements, in particular earth observation, offer substantial advantages. They provide consistent sources of information often with global coverage and they are regularly updated which makes them suitable for natural capital accounting and monitoring trends.

ES modelling can be used to quantify ES if no direct or indirect measurements are available. This is virtually always the case in any ecosystem assessment. With ES modelling, we understand the simulation of supply, use and demand of ES based on ecological and socio-economic input data or knowledge. Models can vary from simple expert based scoring systems to complex ecological models which simulate the planetary cycles of carbon, nitrogen and water. More details are also available in Chapter 4.4

In the context of biophysical quantification, models can be used for spatial and temporal gap filling of direct and indirect measurements, extrapolation of direct and indirect measurements, modelling ES for which there are no measurements available or for scenario analysis.

For regulating services, modelling is sometimes the only option in order to quantify actual ecosystem service flows. This is particularly evident when ecosystems are regulating or mediating stocks and flows of soil, carbon, nitrogen, water or pollutants. Consider soil protection - also termed as erosion regulation or erosion control - which is the role ecosystems and vegetation plays in retaining soil or avoiding soil being eroded as a result of wind or run-off water. Soil erosion can be measured directly on sites which are prone to erosion, usually cropland on slopes. However, estimating the quantity of soil that is not eroded due to the protective cover of vegetation cannot be measured. It can however be modelled by comparing the amount of soil erosion with a model which simulates the presence of vegetation with a model where the protective vegetation cover is deliberately set to zero or to parameters which correspond to parameters for cropland or bare soil. The difference between these two models results in an estimate of avoided soil erosion and can represent the realised service flow. A similar rationale applies to water purification, air quality regulation or other services which exert control on the fate and transport of abiotic and organic material.

Ecosystem service assessments have increasingly been used to support environmental management policies, mainly based on biophysical and economic indicators. Therefore ES assessments have to integrate data and information on biophysical ecosystem components, including biodiversity, with socio-economic system components and the societal and policy contexts in which they are embedded.

Quantification of ES using biophysical methods have been used for a number of perspectives and for a variety of purposes, including landscape management, natural capital accounting, awareness raising, priority setting of projects or policies and policy instrument design. However, transferring the outcomes of the biophysical assessments to policy is not straightforward and some additional work is required to ensure a minimum degree of consistency and avoid over-simplistic conclusions.

Different methods are relevant at different policy levels (ranging from international, EU, national, regional and local scales). Existing literature frequently acknowledges that, in these cases, the interrelationship between different scales must be taken into consideration, which can pose significant challenges. Broad framings for these methods include the work done globally of the Inter-governmental Platform on Biodiversity and Ecosystem Services (IPBES) and the Mapping and Assessment of Ecosystems and their Services (MAES) in the context of the EU Biodiversity Strategy. The initial methodological work on biophysical methods will be the basis for the assessment of the economic value of ES and promote the integration of these values into accounting and reporting systems.

"You can't manage what you don't measure". This well-known expression is also valid for ES which is, in essence, a concept to guide and support the management of natural resources, ecosystems and socio-ecological systems. ES represent the flows of material, energy and information from ecosystems to society. Accurate measurement of these flows as well as the extent and the condition of ecosystems which support these flows is therefore key to base decisions, to monitor progress to biodiversity targets and to create a sound knowledge base for natural capital.

Boerema A, Rebelo AJ, Bodi MB, Esler KJ, Meire P (2016) Are ecosystem services adequately quantified? Journal of Applied Ecology. DOI: 10.1111/1365-2664.12696.
De Araujo Barbosa CC, Atkinson PM, Dearing JA (2015) Remote sensing of ecosystem services: a synthetic review. Ecological Indicators 52: 430-443.
Kareiva P, Tallis H, Ricketts TH, Daily GC, Polasky S (2011) Natural Capital: Theory and Practice of Mapping Ecosystem Services. Oxford University Press, Oxford.
Mononen L, Auvinen AP, Ahokumpu AL, Ronka M, Aarras N, Tolvanen H, Kamppinen M, Viirret E, Kumpula T, Vihervaara P (2016) National ecosystem service indicators: Measures of social-ecological sustainability. Ecological Indicators 61: 27-37.
Peh KS-H et al. (2013) TESSA: A toolkit for rapid assessment of ecosystem services at sites of biodiversity conservation importance. Ecosystem Services 5: e51-e57.
Pettorelli N, Owen HJF, Duncan C (2016) How do we want satellite remote sensing to support biodiversity conservation globally? Methods in Ecology and Evolution 7: 656-665.

Any evaluation of ES requires an integrated analysis, taking into account the supply and demand of ES and their biophysical, socio-cultural and economic value dimensions (see Chapters 4.1, 4.2 and 4.3, respectively). Recent literature has acknowledged that many of the contributions on ES valuation still use the term 'value' exclusively in a monetary sense, ignoring the broader contributions of ecosystems and biodiversity to society in terms of cultural, therapeutic, artistic, inspirational, educational, spiritual or aesthetic values.

To fill this scientific gap, literature on socio-cultural valuation approaches has grown in the last ten years, mostly related to cultural ES (Figure 1). The recent increase in the number of scientific papers on socio-cultural valuation of ES coincides with the creation of the Intergovernmental Platform of Biodiversity and Ecosystem Services (IPBES) in 2012. Some of the challenges addressed by IPBES are related with socio-cultural valuation of ES, such as the inclusion of different knowledge-systems or the recognition of value pluralism.

Figure 1.

Trends in the scientific literature exploring socio-cultural valuation approaches for cultural ES. Note: this illustration is not representing the total number of published papers on cultural services valuation, but the timeline of publications of the most relevant papers which focus on six cultural ES: non-extractive recreation and tourism (e.g. outdoor recreation, ecotourism), (2) extractive recreation and tourism (e.g. sport fishing, recreational hunting), (3) local ecological knowledge, (4) scientific knowledge and environmental education, (5) spiritual interactions with nature and (6) aesthetic experience.

Despite the increase in the number of publications, socio-cultural valuation approaches have not yet formalised a common methodological framework. Designing a methodological framework, able to explore ways of representing cognitive, emotional and ethical responses to nature, alongside ways of expressing preferences, needs and the desires of people in relation to ES, is very much needed. In this context, the present chapter aims to contribute to this challenge through the review of socio-cultural valuation methods that have been frequently applied in ES literature.

Socio-cultural valuation is defined in this chapter as an umbrella term for those methods that aim to analyse human preferences towards ES in non-monetary units. Under this umbrella, terms such as 'psycho-cultural valuation', 'social valuation', 'deliberative valuation', 'qualitative valuation' and 'subjective assessment' represent valuation approaches that aim to uncover individual and collective values and perceptions of ES without relying on market logic and monetary metrics.

There are multiple approaches to uncover socio-cultural values of ES depending on data availability and the purpose of the valuation. In this chapter, we will focus on seven methods that are frequently used in literature.

Preference assessment is a direct consultative method that assesses the individual and social importance of ES by analysing motivations, perceptions, knowledge and associated values of ES. Data is collected through free-listing exercises, ecosystem service ranking, rating, or other selection mechanisms. Techniques for weighting the preferences related to impacts on the ecosystem service of different management alternatives such as multi-criteria analysis are examples of integrated preference assessment valuation.

In the same manner, but aiming at a more quantitative indicator of socio-cultural values of ES, the time use method creates hypothetical scenarios for willingness to give up time (WTT). This method estimates the value of ES by asking people how much time they are willing to dedicate for a change in the quantity or quality of a given ecosystem service. This method is not only a non-monetary metric, but also a way of measuring the willingness to actively contribute to nature conservation through practical actions.

Photo-elicitation surveys seek to uncover the socio-cultural value of ES by translating people's visual experiences, perceptions and preferences of landscapes into ecosystem service values. The use of photo-elicitation surveys has proven to be a useful technique for eliciting socio-cultural values of ES as it uses a communication channel (i.e. photographs) which is easily understood by multiple social actors (for instance see Chapter 7.3.3).

Narrative methods differ from the previous three as they are mainly used to collect qualitative data. By using narrative methods (e.g. structured, semi-structured and unstructured interviews, focus groups, participant observation, content analysis, voice and video recording of events, artistic expression, etc.), participants can articulate the plural and heterogeneous values of ES through their own stories and direct actions (both verbally and visually).

Three other approaches, frequently used in socio-cultural valuation, focus on the integration of knowledge systems, disciplines and diverse data. Participatory mapping of ES (or sometimes referred to as participatory geographical information systems or review and standarized PGIS, see Box 1) assesses the spatial distribution of ES according to the perceptions and knowledge of stakeholders via workshops and/or surveys. PGIS facilitates the participation of various stakeholders (e.g. community members, environmental professionals, NGO representatives, decision-makers, etc.) integrating their perceptions, knowledge and values in maps of ES (see Chapter 5.6.2).

Box 1. Participatory mapping of ecosystem service flows in a National Park (Sierra Nevada, Spain)

Participatory GIS seeks to produce ecosystem service maps in regions of data scarcity while engaging stakeholders through the mapping process. These two aims were pursued in the process developed in Sierra Nevada to map ecosystem service flows. In a two day workshop, 20 participants mapped the supply and demand (i.e. Service Provision Hotspots and Service Benefiting Areas) of 11 ES. Results showed the importance of protected areas to deliver ES and allowed the elaboration of concrete policy proposals for the protected area and its surrounding landscape. Regarding ecosystem service supply, potential restoration areas and areas that require a value enhancement strategy were identified. Ecosystem service demand maps showed the need of a multi-scale strategy for protected area management beyond protected area boundaries to be able to manage the demand that affects the ecosystem within the protected area.

Participatory mapping of ES developed by experts (i.e. managers and scientists) in Sierra Nevada Protected Area.

Scenario planning combines various tools and techniques (e.g. interviews, brainstorming or visioning exercises in workshops, often complemented with modelling) to develop plausible and internally consistent descriptions of alternative futures, where values of ES can be elicited. Assumptions about future events or trends are questioned and uncertainties are made explicit to establish transparent links between changes in ES and human well-being.

Deliberative methods comprise various tools and techniques to engage and empower non-scientific participants. These methods (e.g. valuation workshops, citizens' juries, photo-voice, etc.) invite stakeholders and citizens to form their preferences for ES together through an open dialogue. Deliberative methods can address ethical beliefs, moral commitments and social norms and are often used in combination with other approaches (e.g. mapping or monetary valuation).

The diversity of socio-cultural methods described above is determined by different methodological requirements (Table 1) and the ability of the different methods to provide different outputs and to uncover different types of values (Table 2). Regarding methodological requirements, socio-cultural methods can be clustered into three different groups: (1) methods that require multiple observations as they are quantitative methods and are usually developed in collaboration with scholars from the same field (i.e. preference assessment, time-use and photo-elicitation), (2) methods based on qualitative data that are usually applied in collaboration with non-academic stakeholders (i.e. narratives), (3) methods that are able to gather qualitative and quantitative data by collaborating with scholars from other fields and non-academic stakeholders (for instance PGIS, participatory scenario planning and deliberative valuation), also called integrated approaches (Table 1). This third group of methods has been applied to uncover ES values at national scales (and international in the case of scenarios) while the first two groups are not usually applied at such broad scales. Further, the third type of methods can contribute to social learning and knowledge co-production as it fosters discussion between different stakeholder groups regarding the importance of different ES (deliberative valuation), their spatial distribution (PGIS) and the future trends of ES and their implications for human wellbeing (participatory scenario planning).

Table 1.Download as CSV 

Methodological requirements of socio-cultural methods for valuing ES. Methods are evaluated according to their suitability to value ES at different spatial scales and to uncover quantitative or qualitative data -(***) high , (**) moderate , (*) low - and according to the level of requirements in terms of data, collaboration, time and economic resources -(H) high , (M) medium , (L) low -. Source: Kelemen et al. (2015).

SOCIO-CULTURAL METHODS	SPATIAL SCALE			DATA			COLLABORATION			RESOURCES
	Local	Regional	National	Amount of data	Qualitative	Quantitative	Researchers' own field	Researchers' other field	Non-academic stakeholders	Time	Economic
Preference assessment	***	***	*	H	**	***	H	H	L	M	M
Time use	***	***	*	H	*	***	H	M	L	M	M
Photo-elicitation surveys	***	***	*	H	**	***	H	H	L	M	M
Narratives	***	***	*	M	***	**	L	L	H	M	M
Participatory GIS (PGIS)	***	***	**	M	***	***	L	H	H	M	M
Scenario planning	***	***	**	M	***	**	L	H	H	L	M
Deliberative valuation	***	***	**	M	***	***	L	H	H	M	M

Table 2.Download as CSV 

Main socio-cultural methods are presented in relation to their capacity to integrate different types of values-(***) high , (**) moderate , (*) low, (-) not appropriate - and according to their capacity to integrate values -(H) high , (M) medium , (L) low -. Source: Kelemen et al. (2015).

SOCIO-CULTURAL METHODS

IPBES values

TEEB values

Total Economic Value

Intrinsic

Relational

Instrumental

Ecological

Socio-cultural

Monetary

Direct use values

Indirect use values

Existence values

Bequest values

Option values

Integrative Potential

Preference assessment

*

**

**

*

***

-

**

**

**

**

*

M

Time use

*

**

**

*

**

*

**

**

**

*

*

M

Photo-elicitation surveys

*

*

**

**

***

*

**

**

**

**

**

M

Narratives

**

**

**

***

***

-

-

-

*

*

*

L

Participatory GIS (PGIS)

*

***

*

***

***

*

**

*

***

***

**

H

Scenario planning

*

***

***

**

**

**

*

**

**

***

***

H

Deliberative valuation

**

***

**

**

***

*

*

**

**

**

**

H

Degree of values captured by all methods

L

M

M

M

H

L

L

L

M

M

M

PGIS is also the most suitable method to provide spatial outputs, although preference assessment, time use and photo-elicitation may also contribute with spatially explicit results by estimating representative values for different geographical areas. PGIS is particularly suited to identify ecosystem service benefiting areas, i.e. places where use or demand of ES converge (see Chapters 5.2 and 5.6.2).

Despite all developments regarding socio-cultural valuation of ES, the question of how socio-cultural valuation methods can elicit the broad range of values associated with nature is still relatively unexplored. Following the conceptual definitions provided for value categories in the Total Economic Value (TEV), the Economics of Ecosystems and Biodiversity (TEEB) and the IPBES, an integrative approach to socio-cultural valuation methods has the capacity to uncover most of the different value categories (Table 2). Broadly speaking, Table 1 shows that some methods are more specific towards certain value types (e.g. narrative methods), while other methods are generally able to capture multiple values, but not specifically designed for any value type in particular (e.g. participatory scenario planning or deliberative valuation). All value types are appropriately covered by one or more methods, but all methods have blind spots, which imply bias and conditional application. Consequently, using multiple methods is necessary to cover all values types.

The resulting analyses reflect the extent to which diverse valuation methods capture specific value types or have integrative potential, as well as which set of complementary methods can be applied to capture multiple values.

A key similarity amongst socio-cultural methods is the assumption that values of ES are rooted in individuals and, at the same time, shaped by individuals' social and cultural context. In fact, socio-cultural approaches have the capacity to elicit collective and shared values of ES through participatory and deliberative techniques that go beyond the aggregation of individual preferences. Socio-cultural valuation methods aim at valuing ES in a considered way by discovering the psychological, historical, cultural, social, ecological and political contexts and conditions, as well as social perceptions that shape individually held or commonly shared values.

Variability among methods makes socio-cultural valuation capable of flexible adaptation to specific worldviews and decision contexts. Key aspects of this variability include (Figure 2):

Figure 2.

Variability among socio-cultural valuation methods in relation to three axes: type of values elicited, type of rationality attributed to value providers, and the dominant approach of handling data.

1. The type of values elicited: methods focusing on the value to individuals versus methods focusing on the value to society. Values can be considered at the level of the individual (what is considered useful, important, good or morally acceptable by a person) and at higher levels of societal organisation, including a group, a community or the society as a whole (Figure 2). The latter type includes social and cultural values and refers to the fact that societies hold shared principles and virtues, as well as a shared sense of what is worthwhile and meaningful. Shared social values influence individual values because all of us are part of and have been socialised within, a specific community and social context. Valuation methods differ in terms of focusing on personal (individual) understandings of value, or eliciting those value dimensions that are shared by a group of people and culturally embedded within a society.

2. The type of rationality attributed to participants (value providers): self-oriented versus others-oriented methodological approaches. We can distinguish between individual (I) and collective (We) rationality as the two main rules of thumb behind reasonable actions (Figure 2). When following "I" rationality, we consider individual benefits and costs of personal actions and choose the most beneficial option for ourselves. On the other hand, following "We" rationality means that before acting, we consider what is good and bad for our community/society and how our actions can impact others. Therefore, "I" rationality refers to self-oriented actions and choices, while "We" rationality refers to other-regarding actions and choices.

3. The process of including participants (value providers) in valuation: observation, consultation or engagement methods. There are three options to gain knowledge on preferences, depending on whether preferences (values) are considered as pre-existing or in the process of formation. Preferences can be observed and reported when participants have a direct relation with the subject of valuation (e.g. they frequently use or enjoy some ES). However, not having a direct relation with the subject of valuation does not necessarily mean that participants do not attribute value to it. To explore these social preferences, participants can be consulted or asked via questionnaires or interviews about their perceptions of ES. If preferences are not expected to exist a priori, or are in the process of formation (i.e. participants do not have a priori knowledge about, or have not faced others' perceptions of certain ES), we can also engage participants in a joint preference formation process through deliberative valuation, participatory scenario planning or PGIS.

4. The dominant approaches to handling data: predominantly quantitative, predominantly qualitative and mixed methodological approaches. All three types of methods can be used to collect quantitative, as well as qualitative data. Quantitative data can be collected in numerical form from large populations and, if representative, can provide results that are applied, in a general sense, from local to regional or even broader spatial scales. Quantitative data can be collected both at individual and group level and then aggregated to generalise the results from the sample to larger populations. Qualitative data allow an in-depth understanding of values and underlying motivations, but usually for a much smaller (and often non-representative) sample. Qualitative data can be collected at the individual and group level in the form of narrative arguments (mainly words, but also pictures, drawings, etc.). Due to the heterogeneity of types of data, aggregation is often impossible and other means of synthesis have to be used (e.g. narrative methods or deliberation). In practice, quantitative and qualitative approaches can be placed along a continuum (Figure 2) and, in many cases, they are used in a mixed and complementary approach.

Ecosystem service assessments have increasingly been called to support environmental planning, mainly based on biophysical and economic indicators. However, the expectations of decision-makers in relation to how these assessments can support decision-making are not always fulfilled. Moreover, few studies have included the socio-cultural dimension of ES, despite its being considered a research priority. Overlooking the socio-cultural dimension might obscure human-nature relationships and hinder the mainstreaming of ES across societal sectors and in decision-support.

Integrated valuation aims to clarify the interdependencies between the multiple values associated with different ES (see also Box 2 for an example). The biophysical dimension, i.e. an ecosystem's capacity to supply services, determines the range of potential uses by society which also influences its socio-cultural and monetary values. Socio-cultural values might also have an influence on monetary values because individual and social motivations determine the 'utility' a person obtains from a particular service. Conversely, monetary values have social interpretations and the process of monetary valuation is value-articulating in itself. These interdependencies between value dimensions and the different information provided by them, justify combining the different value domains to properly inform environmental decision-making processes.

Box 2. Socio-cultural valuation of ES in Hungary (Homokhatsag)

The major aim of this ES study was to help local stakeholders and decision-makers move towards a more sustainable landscape management system. To this end, in-depth and semi-structured interviews and focus groups were applied. We carried out narrative methods to understand the institutionalised mechanisms affecting farmers' choices that are often in conflict with nature conservation.

Moreover, we carried out deliberative valuation to understand how farmers relate to biodiversity and whether it has different meanings and values to different groups of farmers. A preference assessment survey was carried out to mobilise community members and collect information on their knowledge, opinion and feelings related to ES. This was then channelled into a participatory scenario planning process, combined with modelling, to enable stakeholders and experts to explore alternative future options and choose the most desirable one(s) together. This long lasting research process was able to highlight multiple dependencies between local inhabitants and their surrounding environment. We could identify plural and heterogeneous values and their possible changes across time and space.

Focus group with local stakeholders using visual stimuli to elicit socio-cultural values of ES and their spatial distribution in the landscape.

In this section, we formulate several propositions regarding how socio-cultural valuation methods can provide support in decision-making:

1. Socio-cultural approaches help broaden the valuation scope and capture multiple values that complement other valuation methods. Socio-cultural valuation methods can be used to identify how values and perceptions toward ES differ among stakeholders and offer insights into the motivations for conserving nature and the symbolic, cultural and spiritual values that are frequently invisible in other valuation approaches. Further, socio-cultural valuation methods can address relational values that are preferences, principles and virtues associated with nature-human relationships. For example, deliberative methods allow the consideration of ethical beliefs, moral commitments and social norms.

2. Socio-cultural valuation methods can cover different spatial scales. Values derived from large representative samples of a population can be transferred to other locations when the social, cultural and ecological conditions are similar and aggregated to larger scales than the original study. Given the emphasis of socio-cultural valuation methods on social formation and context-dependency of values, some approaches such as value transfer, aggregation and scaling are less common than in economic valuation where assumptions of pre-existing individual preferences encourage comparisons across contexts. In addition, a number of socio-cultural valuation methods are applied at local scales to assess certain values in depth.

3. Socio-cultural valuation methods are a useful tool to identify how plural values are interlinked. These help identify plural and heterogeneous values that are relevant for different people (e.g. different socio-demographic profiles, different cultures or cosmologies), at different temporal scales (e.g. seasons of the year) and different choice situations (individual versus group). Socio-cultural valuation methods can reveal how plural and heterogeneous values are interlinked and contribute to human wellbeing.

4. Socio-cultural methods are more appropriate in situations of social conflict than other valuation methods. Aiming for an in-depth understanding of human-nature relationships, some socio-cultural methods integrate different forms of knowledge (e.g. expert or technical knowledge and experiential and local knowledge) held by different social actors. Sometimes, the interests of one stakeholder group might be in conflict with the interests of other stakeholders and power relations might operate between them. In that case, socio-cultural valuation can support the identification of conflicts arising from different perceptions, needs and uses of ES, as well as power inequities in the access to ES.

5. Socio-cultural preferences can serve as indicators of the impact of different management options on the ecosystems' capacity to deliver services. Socio-cultural preferences are often associated with ecosystem service bundles. They are helpful in identifying ecosystem service synergies and trade-offs resulting from stakeholders' diverging interests and knowledge. Social preferences for ES can be used as indicators of present and future pressures on landscapes and land-use change. For example, multi-criteria analysis can combine a biophysical ecosystem service assessment with people's willingness to trade off one ecosystem service for another, establishing a ranking order of landscape management alternatives that can be used in priority-setting.

In summary, socio-cultural valuation methods can provide decision-support in the form of awareness-raising, value and knowledge recognition, value conflict identification and priority-setting. They also help bring different voices and stakeholders into the decision-making process.

Chan K, Balvanera P, Benessaiah K, Chapman M, Diaz S, Gomez-Baggethun E, Gould R, Hannahs N, Jax K, Klain S, Luck G, Martin-Lopez B, Muraca B, Norton B, Ott K, Pascual U, Satterfield T, Tadaki M, Taggart J, Turner NJ (2016) Why Protect Nature? Rethinking Values and the Environment. PNAS 113: 1462-1465.
Garcia-Llorente M, Martin-Lopez B, Iniesta-Arandia I, Lopez-Santiago CA, Aguilera PA, Montes C (2012) The role of multi- functionality in social preferences toward semi-arid rural landscapes: an ecosystem service approach. Environmental Science and Policy 19-20: 136-146.
Gomez-Baggethun E, Martin-Lopez B (2015) Ecological Economics perspective in ecosystem services valuation. In: Martinez-Alier J, Muradian R (Eds) Handbook of Ecological Economics, 260-282. Edward Elgar, London.
Iniesta-Arandia I, Garcia-Llorente M, Aguilera P, Montes C, Martin-Lopez B (2014) Socio-cultural valuation of ecosystem services: uncovering the links between values, drivers of change and human well-being. Ecological Economics 108: 36-48.
IPBES (2015) Preliminary guide regarding diverse conceptualisation of multiple values of nature and its benefits, including biodiversity and ecosystem functions and services, IPBES 15 Deliverable 3 (d).
Kelemen E, Garcia-Llorente M, Pataki G, Martin-Lopez B, Gomez-Baggethun E (2014) Non-monetary techniques for the valuation of ecosystem services. OpenNESS Syntehsis Papers No. 6.
Kenter J, O'Brien L, Hockley N, Ravenscroft N, Fazey I, Irvine K, Reed M, Christie M, Brady E, Bryce R and Church A (2015) What are shared and social values of ecosystems? Ecological Economics, 111: 86-99.
Kovacs E, Kelemen E, Kaloczkai A, Margoczi K, Pataki G, Gebert J, Malovics G, Balazs B, Roboz A, Kovacs E, Mihok B (2015) Understanding the links between ecosystem service trade-offs and conflicts in protected areas. Ecosystem Services 12: 117-127.
Martin-Lopez B, Iniesta-Arandia I, Garcia-Llorente M, Palomo I, Casado-Arzuaga I, Del Amo DDG, Gomez-Baggethun E, Oteros-Rozas E, Palacios-Agundez I, Willaarts B, Gonzalez JA, Santos-Martin F, Onaindia M, Lopez-Santiago C, Montes C (2012) Uncovering ecosystem service bundles through social preferences. PLoS ONE 7(6): e38970.
Oteros-Rozas E, Martin-Lopez B, Gonzalez JA, Plieninger T, Lopez CA, Montes C. (2014) Socio-cultural valuation of ecosystem services in a transhumance social-ecological network. Regional Environmental Change 14: 1269-1289.
Plieninger T, Bieling C, Ohnesorge B, Schaich H, Schleyer C, Wolff F (2013) Exploring futures of ecosystem services in cultural landscapes through participatory scenario development in the Swabian Alb, Germany. Ecological and Society 18(3): 39.
Santos-Martin F, Martin-Lopez B, Garcia-Llorente M, Aguado M, Benayas J, Montes C (2013) Unravelling the Relationships between Ecosystems and Human Wellbeing in Spain. PLoS One 8: e73249.
Teddlie C, Tashakkori A (Eds) (2009) Foundations of mixed methods research: Integrating quantitative and qualitative approaches in the social and behavioural sciences. Sage Publications Inc.

Economic quantification of ES attempts to measure the human welfare derived from the use or consumption of ES. Economic quantification or valuation is one way to assess and communicate the importance of ES to decision-makers and can be used in combination with other forms of information (e.g. bio-physical or social quantification - see Chapters 4.1 and 4.2). The comparative advantage of economic valuation is that it conveys the importance of ES directly in terms of human welfare and uses a common unit of account (i.e. money) so that values can be directly compared across ES and across other goods and services that are consumed by society.

The aim of this chapter is to introduce the key concepts underlying economic quantification of ES and to provide an explanation for the various economic methods that can be applied.

In this section, we provide definitions of the various concepts of economic value that may be encountered when quantifying ES.

In neo-classical welfare economics, the economic value of goods or services is the wellbeing derived from its production and consumption, usually measured in monetary units. In a perfectly functioning market, the economic value of goods or services is determined by the demand for and supply of those goods or services. Demand for goods or services is driven by the benefit, utility or welfare that consumers derive from it. Supply of goods or services is determined by the cost to producers of producing it. The left-hand panel in Figure 1 provides a simplified representation of demand (marginal benefit) and supply (marginal cost) for goods traded in a market at quantity 'Q' and price 'P'. Area 'A' represents the consumer surplus which is the gain obtained by consumers because they are able to purchase a product at a market price that is less than the highest price they would be willing to pay (which is related to their benefit from consumption and represented by the demand curve). The producer surplus, depicted by 'B', is the amount that producers benefit by selling at a market price that is higher than the lowest price that they would be willing to sell for (which is related to their production costs and represented by the supply curve). The area 'C' represents production costs which differ among producers and/or over the scale of production. The sum of areas A and B is labelled the 'surplus' and is interpreted as the net economic gain or welfare resulting from production and consumption with a quantity of Q at price P.

Figure 1.

Demand and supply for ES.

It is important to recognise that, when we make a decision to allocate resources to produce particular goods or services, we are also deciding not to allocate those resources to produce alternative goods or services. The goods or services that we give up is called the "opportunity cost" of our decision. Opportunity cost can be defined as the value of the foregone next best use of resources. This is an important concept in the context of ES since it is often the value of the alternative use of resources (e.g. agriculture, timber extraction, aquaculture) that drives ecosystem loss.

In the case that ES are not traded in a market, the interpretation of the welfare derived from their provision can also be represented in terms of surplus. The right-hand panel of Figure 1 represents the supply and demand of a non-marketed ecosystem service. In this case, the ecosystem service does not have a supply curve in the conventional sense that it represents the quantity of the service that producers are willing to supply at each price. The quantity of the ecosystem service that is 'supplied' is not determined through a market at all but by other decisions regarding ecosystem protection, land use, management, access etc. The quantity of the ecosystem service supplied is therefore independent of its value. This is represented as a vertical line. For the most part, bio-physical indicators of ES measure the quantity supplied but not the welfare obtained. The demand curve for non-marketed ES is still represented as a downward sloping line since marginal benefits are expected to decline with quantity (the more that we have of a service, the lower the additional welfare of consuming more). In this case, consumers do not pay a price for the quantity (Q) that is available to them and the entire area under the demand curve (D+E) represents their consumer surplus. It is useful to keep this picture in mind when considering the economic quantification of ES.

The concept of Total Economic Value (TEV) of an ecosystem is used to describe the comprehensive set of utilitarian values derived from it. This concept is useful for identifying the different types of value that an ecosystem provides. TEV comprises of use values and non-use values. Use values are the benefits that are derived from some physical use of the resource. Direct use values may derive from on-site extraction of resources (e.g. fuel wood) or non-consumptive activities (e.g. recreation). Indirect use values are derived from off-site services that are related to the resource (e.g. downstream flood control, climate regulation). The option value is the value that people place on maintaining the option to use an ecosystem resource in the future given the uncertainty that they would actually use it. Non-use values are derived from the knowledge that an ecosystem is maintained without regard to any current or future personal use. Non-use values may be related to altruism (maintaining an ecosystem for others), bequest (for future generations) and existence (preservation unrelated to any human use) motivations. The constituent components of TEV are represented in Figure 2. It is important to understand that the "total" in Total Economic Value refers to the aggregation of different sources of value rather than the sum of all values derived from a resource. Accordingly, many estimates of TEV are for marginal changes in the provision of ES but "total" in the sense that they take a comprehensive view of sources of value.

Figure 2.

The components of Total Economic Value.

The concept of welfare value is used in most economic assessments of ES but it is not used in the system of national accounts (SNA) that is used to calculate gross domestic product (GDP) and other economic statistics. The SNA uses the concept of exchange value which is a measure of producer surplus plus the costs of production. In the left-hand panel of Figure 1, this is represented by areas B and C or equivalently P times Q. Under the concept of exchange value, the total outlays by consumers and the total revenue of the producers are equal. For national accounting purposes, this approach to valuation enables a consistent and convenient recording of transactions between economic units. In the context of comparing the values of ES with values in the system of national accounts, it is therefore necessary to value the total quantity of ES at the market prices that would have occurred if the services had been freely traded and exchanged. In other words, it is necessary to measure exchange value and not welfare value.

The differences between the concepts of welfare value and exchange value are the inclusion of consumer surplus (A) in the former and the inclusion of production costs in the latter (C). The concept of welfare value corresponds to a theoretically valid measure of welfare in the sense that a change in value represents a change in welfare for the producers and/or consumers of the goods and services under consideration. The concept of exchange value does not correspond to a theoretically valid measure of welfare and a change in exchange value does not necessarily represent a change in welfare for either producers or consumers.

A variety of methods have been developed for quantifying the economic value of ES. These valuation methods are designed to span the range of valuation challenges raised by the application of economic analyses to the complexity of the natural environment. An important distinction exists between methods that produce new or original information generally using primary data (primary valuation methods) and those that use existing information in new policy contexts (value transfer methods).

Table 1 provides an overview of primary valuation methods, typical applications and limitations and indicates which primary valuation methods can be used to value which ecosystem service. The reader should be aware of the important distinction between primary valuation methods, i.e. the difference between revealed preference methods (those that observe actual behaviour of the use of ES to elicit values) and stated preference methods (those that use public surveys to ask beneficiaries to state their preferences for, generally, hypothetical changes in the provision of ES). Revealed preference methods may be favoured since they reflect actual behaviour but are limited in their applicability to some ES. Stated preference methods, on the other hand, rely on responses recorded in surveys or experiments but are more flexible in their application.

Table 1.Download as CSV 

Primary economic valuation methods.

Valuation method	Approach	Application to ES	Example ecosystem service
Market prices	Prices for ES that are directly observed in markets.	ES that are traded directly in markets.	Timber and fuel wood from forests; Recreation at national parks that charge an entrance fee.
Public pricing	Public expenditure or monetary incentives (taxes/subsidies) for ES as an indicator of value.	ES for which there are public expenditures.	Watershed protection to provide drinking water; Purchase of land for protected areas.
Defensive expenditure	Expenditure on protection of ecosystems.	ES from protected ecosystems.	Nutrient filtration by protected wetlands
Replacement cost	Estimate the cost of replacing an ES with a man-made service.	ES that have a man-made equivalent.	Coastal protection by dunes; water storage and filtration by wetlands.
Restoration cost	Estimate cost of restoring degraded ecosystems to ensure provision of ES.	Any ES that can be provided by restored ecosystems.	Coastal protection by dunes; water storage and filtration by wetlands.
Damage cost avoided	Estimate damage avoided due to ecosystem service.	Ecosystems that provide storm or flood protection to houses or other assets.	Coastal protection by dunes; river flow control by wetlands.
Net factor income	Revenue from sales of environment-related good minus cost of other inputs.	Ecosystems that provide an input in the production of marketed goods.	Filtration of water by wetlands; commercial fisheries supported by coastal wetlands.
Production function	Statistical estimation of production function for marketed goods including an ES input.	Ecosystems that provide an input in the production of marketed goods.	Soil quality or water quality as an input to agricultural production.
Hedonic pricing	Estimate influence of environmental characteristics on price of marketed goods.	Environmental characteristics that vary across goods (usually houses).	Urban open space; air quality.
Travel cost	Use data on travel costs and visit rates to estimate demand for recreation sites.	Recreation sites	Outdoor open access recreation.
Contingent valuation	Ask people to state their willingness to pay for an ES through surveys.	All ES	Species loss; natural areas; air quality; water quality landscape aesthetics.
Choice modelling	Ask people to make trade-offs between ES and other goods to elicit willingness to pay.	All ES	Species loss; natural areas; air quality; water quality; landscape aesthetics.
Group / participatory valuation	Ask groups of stakeholders to state their willingness to pay for an ES through group discussion.	All ES	Species loss; natural areas; air quality; water quality; landscape aesthetics.

Decision-making often requires information quickly and at low cost. New 'primary' valuation research, however, is generally time consuming and expensive. For this reason, there is interest in using information from existing primary valuation studies to make inform decisions regarding impacts on ecosystems that are of current interest. This transfer of value information from one context to another is called value transfer. Value transfer is the use of research results from existing primary studies at one or more sites or policy contexts ("study sites") to predict welfare estimates or related information for other sites or policy contexts ("policy sites"). Value transfer is also known as benefit transfer but since the values that are transferred may be costs as well as benefits, the term value transfer is more generally applicable.

In addition to the need for expeditious and inexpensive information, there is often a need for information on the value of ES at a different geographic scale from that for which primary valuation studies have been conducted. So, even in cases where some primary valuation research is available for the ecosystem of interest, it is often necessary to extrapolate or scale-up this information to a larger area or to multiple ecosystems in the region or country. Primary valuation studies tend to be conducted for specific ecosystems at a local scale whereas the information required for decision-making is often needed at a regional or national scale. Value transfer therefore provides the means to obtain information for the scale that is required.

The number of primary studies on the value of ES is substantial and growing rapidly. This means that there is a growing body of evidence to draw on for the purposes of transferring values for informed decision-making. With an expanding information base, the potential for using value transfer is improved.

Value transfer can potentially be used to estimate values for any ecosystem service, provided that there are primary valuations of that ecosystem service from which to transfer values. The use of value transfer is widespread but requires careful application. The alternative methods of conducting value transfer are described here.

Unit value transfer uses values for ES at a study site, expressed as a value per unit (usually per unit of area or per beneficiary), combined with information on the quantity of units at the policy site to estimate policy site values. Unit values from the study site are multiplied by the number of units at the policy site. Unit values can be adjusted to reflect differences between the study and policy sites (e.g. income and price levels).

Value function transfer uses a value function estimated for an individual study site in conjunction with information on parameter values for the policy site to calculate the value of an ecosystem service at the policy site. A value function is an equation that relates the value of an ecosystem service to the characteristics of the ecosystem and the beneficiaries of the ecosystem service. Value functions can be estimated from a number of primary valuation methods including hedonic pricing, travel cost, production function, contingent valuation and choice experiments.

Meta-analytic function transfer uses a value function estimated from the results of multiple primary studies representing multiple study sites in conjunction with information on parameter values for the policy site to calculate the value of an ecosystem service at the policy site. A value function is an equation that relates the value of an ecosystem service to the characteristics of the ecosystem and the beneficiaries of the ecosystem service. Since the value function is estimated from the results of multiple studies, it is able to represent and control for greater variation in the characteristics of ecosystems, beneficiaries and other contextual characteristics. This feature of meta-analytic function transfer provides a means to account for simultaneous changes in the stock of ecosystems when estimating economic values for ES (i.e. the "scaling up problem"). By including an explanatory variable in the data describing each "study site" that measures the scarcity of other ecosystems in the vicinity of the "study site", it is possible to estimate a quantified relationship between scarcity and ecosystem service value. This parameter can then be used to account for changes in ecosystem scarcity when conducting value transfers at large geographic scales.

These three principal methods for transferring ecosystem service values are summarised in Table 2. The choice of which value transfer method to use to provide information for a specific policy context is largely dependent on the availability of primary valuation estimates and the degree of similarity between the study and policy sites (in terms of biophysical and socio-economic characteristics and context). In cases where value information is available for a highly similar study site, unit value transfer may provide the most straightforward and reliable means of conducting value transfer. On the other hand, when study sites and policy sites are different, value function or meta-analytic function transfer offers a means to systematically adjust transferred values to reflect those differences. Similarly, in the case where value information is required for multiple different policy sites, value function or meta-analytic function transfer may be a more accurate and practical means for transferring values.

Table 2.Download as CSV 

Value transfer methods.

	Approach	Strengths	Weaknesses
Unit value transfer	Select appropriate values from existing primary valuation studies for similar ecosystems and socio-economic contexts. Adjust unit values to reflect differences between study and policy sites (usually for income and price levels).	Simple	Unlikely to be able to account for all factors that determine differences in values between study and policy sites. Value information for highly similar sites is rarely available.
Value function transfer	Use a value function derived from a primary valuation study to estimate ES values at policy site(s).	Allows differences between study and policy sites to be controlled for (e.g. differences in population characteristics).	Requires detailed information on the characteristics of policy site(s).
Meta-analytic function transfer	Use a value function estimated from the results of multiple primary studies to estimate ES values at policy site(s).	Allows differences between study and policy sites to be controlled for (e.g. differences in population characteristics, area of ecosystem, abundance of substitutes etc.). Practical for consistently valuing large numbers of policy sites.	Requires detailed information on the characteristics of policy site(s). Analytically complex.

The representation of economic values on maps involves estimating variable combinations of supply and demand across spatial units and plotting the resulting values. Spatial units in a value map can include land parcels (e.g. polygons representing ownership), ecosystem patches (e.g. polygons representing distinct ecosystems of different type), ecosystem units (e.g. raster grids of ecosystem type), grid cells (e.g. raster grids with land use/land cover), or beneficiaries (e.g. people plotted using residential or activity location). In most cases, spatial units are used to represent the ecosystem that supplies the ecosystem service, but mapping values by the location of beneficiaries can be useful in some decision-making contexts (e.g. for representing the distributional consequences of changes in ecosystem service provision across communities; or for designing payment mechanisms for ES).

In mapping ecosystem service values, each spatial unit is treated as a separate sub-market for the ecosystem service. Methods for mapping ecosystem service values can address spatial variations in supply, demand, or a combination of both determinants.

general terms, bio-physical methods (see Chapter 4.1) are used to estimate the spatially variable quantities of ES supplied (e.g. probability of flood damage, quantity of clean water, area of recreational space, tonnes of carbon stored) and economic methods are used to estimate spatially variable marginal values per unit of ecosystem service provided and consumed. Mapping economic values therefore necessarily involves linking maps of biophysical ecosystem supply with economic valuation methods.

In addition to the quantification of human welfare derived from ES in monetary units, economic quantification of ES encompasses the recording or estimation of production and consumption statistics in physical units. Indeed, the measurement of physical units of production and consumption of ES is a necessary a step in the process of quantifying economic value. Economic quantification may, however, stop short of estimating values and directly report production and consumption statistics as useful information to support decision-making.

To a great extent, the production of ES is quantified using bio-physical methods (see Chapter 4.1). For most ES, however, it is also necessary to use insights and methods from economics to measure the quantities that are actually used (i.e. to quantify utilised services as opposed to potential services). This generally involves measuring the extent of demand for ES in terms of the population and preferences of beneficiaries.

The physical units in which production and consumption statistics are reported are specific to each ecosystem service. For example, non-timber forest products (NTFPs), such as wild honey, may be measured in kilograms; water extracted for consumption is measured in kilolitres or megalitres, carbon sequestration is conventionally measured in tonnes of carbon or CO₂; and recreational use of natural open space may be measured in numbers of visits - all usually expressed per unit of time over which the flow of service is recorded (e.g. per year). In very few cases will the quantity of an ecosystem service be explicitly observed and recorded in a systematic and accessible way (i.e. there is generally not an equivalent of the business activity surveys conducted for the SNA). In most cases, it is necessary to estimate the level of production and consumption using some form of bio-economic modelling.

Bagstad KJ, Johnson GW, Voigt B, Villa F (2013) Spatial dynamics of ecosystem service flows: a comprehensive approach to quantifying actual services. Ecosystem Services 4: 117-125.
Bouma JA, van Beukering PJH (Eds) (2015) Ecosystem Services: From Concept to Practice. Cambridge University Press.
Brander LM, Brauer I, Gerdes H, Ghermandi A, Kuik O, Markandya A, Navrud S, Nunes PALD, Schaafsma M, Vos H, Wagtendonk A (2012) Using meta-analysis and GIS for value transfer and scaling up: Valuing climate change induced losses of European wetlands. Environmental and Resource Economics 52: 395-413.
Brander LM (2013) Guidance manual on value transfer methods for Ecosystem Services. United Nations Environment Programme. ISBN 978-92-807-3362-4.
Brander LM (2015) . Economic valuation of Ecosystem Services in the Eastern Plains Landscape, Modulkiri, Cambodia. Report for WWF Cambodia.
Brouwer R et al. (2009) Economic Valuation of Environmental and Resource Costs and Benefits in the Water Framework Directive: Technical Guidelines for Practitioners. AquaMoney.
Bryan BA, Crossman ND (2013) Impact of multiple interacting financial incentives on land use change and the supply of Ecosystem Services. Ecosystem Services 4: 60-72.
DEFRA (2013) Guidance for policy and decision makers on using an ecosystems approach and valuing Ecosystem Services. Department for Environment, Food & Rural Affairs https://www.gov.uk/ecosystems-services.
DEFRA (2007) An introductory guide to valuing Ecosystem Services (2007) Department for Environment, Food and Rural Affairs.
Freeman AMI (2003) The Measurement of Environmental and Resource Values. Resources for the Future, Washington D.C.
Johnston RJ, Rolfe J, Rosenberger RS, Brouwer R (Eds) (2015) Benefit transfer of environmental and resource values: A handbook for researchers and practitioners. Springer. ISBN 978-94-017-9929-4.
Pascual U, Muradian R, Brander L, Gomez-Baggethun E, Martin-Lopez B, Verma M (2010) The economics of valuing Ecosystem Services and biodiversity. In Kumar (Ed.) The Economics of Ecosystems and Biodiversity: Ecological and Economic Foundations. Earthscan, London and Washington.
Pearce D et al. (2001) Economic Valuation with Stated Preference Techniques Summary Guide. Department of Transport, Local Government and Regions, London.
Schagner JP, Brander LM, Maes J, Hartje V (2013) Mapping ecosystem service values: Current practice and future prospects. Ecosystem Services 4: 33-46.

Computer models are simplified representations of the environment that allow biophysical, ecological, and/or socio-economic characteristics to be quantified and explored. Modelling approaches differ from mapping approaches (Chapter 5) as (i) they are not forcibly spatial (although many models do produce spatial outputs); (ii) they focus on understanding and quantifying the interactions between different components of social and/or environmental systems and (iii) by changing parameters within models, they are capable of exploring both alternative scenarios and internal model dynamics.

When applied to the assessment of ecosystem services (ES), models are important tools which can quantify the relationships that underpin ES supply, demand and flows and, in some cases, produce maps representing these factors. Furthermore, as models can explore scenarios, trade-offs that result from different scenarios can be assessed. This chapter provides a broad overview of different types of models that have been applied to ES assessments and discusses, with examples, the ways that these models have the potential to be used in practice.

In the context of ES, there are a number of ways of distinguishing between different types of models. Here, we distinguish between individual models focussing on single ES and modelling frameworks that can assess multiple ES within the framework of a single modelling tool.

A model is a simplification of reality that represents the relationship between two or more sets of factors and uses one set of factors to predict values of the other: y = x² is a simple model where the variable y can be predicted from variable x by performing the square function on x. However, there are many different types of models, most of which are considerably more complex.

When used in ES research, models are generally used to predict either ES themselves, or underlying environmental aspects from which ES are derived. ES models use diverse types of input variables, but commonly include measurements of environmental parameters (e.g., tree heights, river flows, species counts), survey responses or scores given by scientists or stakeholders (e.g., from questionnaire responses or interviews) or the outputs from another model (e.g., outputs from a climate model may provide precipitation inputs to a water flow model).

Computer modelling predates the popularisation of the ES concept and models have a decades-long history of use within the environmental sciences. As such, there are a large number of models that can be used to assist in ES assessment, though older biophysical models may be less deliberately beneficiary-orientated than more modern ES models.

It is beyond the scope of this chapter to provide a full overview of modelling per se. In the following text, we describe five general types of models that are used for ES assessment.

The previous section provided a brief overview of the types of models that exist to provide information to help with ES assessment. We next focus on how these models can be used, i.e., on which aspects of the environment do models provide information and how does this help us better assess ES?

In the following sections, we consider four main applications of models. First, models of the natural environment that do not provide direct information about ES, but can assess underlying ecosystem structures and functions from which ES can be understood. Second, models that are focussed on ES, both on individual services and those intended to enable assessment of a suite of different ES. Third, modelling systems that take an integrated approach to ES, which allows for an assessment of trade-offs and synergies between groups of services. Finally, models designed to explore ES with decision-makers or stakeholders.

Not all ES are as easy to model as others. In general, provisioning and regulating services have a longer history of being modelled than cultural ES. In fact, modelling cultural ES tends to be limited to analyses of services with relatively tangible physical aspects to their provision, such as recreation, tourism and, to some extent, aesthetic beauty. This is because factors with greater social or cultural meaning are considerably harder to tie to environmental parameters. It is in situations such as these that participatory approaches come to the fore (see Chapter 5.6.2 for a discussion on the use of participatory approaches for mapping cultural ES).

Care should be taken when interpreting model outputs as ES, as these outputs often represent proxies rather than the actual ES of interest. A clear example is carbon sequestration which is often used as a proxy for the ES of climate regulation, but there are many others (e.g. the distance to locally accessible green space as a proxy for recreation provision). It is very important to understand exactly what the output represents: is it evaluating the underlying ecosystem structure and function only, or does it provide a direct benefit with concrete beneficiaries? Furthermore, does it quantify actual service provision (as directly used by beneficiaries) or potential ES provision (that could be taken up by the beneficiaries, if they had demand for and accessibility to the ecosystems supplying the service)?

Whether ES supply or demand is modelled is another consideration. For some ES, both can be modelled and overlaid to identify mismatches between the two (e.g., air pollution filtration by trees can be modelled using forest data and compared with a map of human exposure to pollutant levels). However, it is often far better to apply another model that accounts for ES flows via service-specific flow mechanisms, rather than to just identify in situ supply-demand mismatches.

Though less commonly mentioned, it is of course possible with the same caveats to model ecosystem disservices or their converse - the natural benefits that control disservices.

Modelling is being widely applied in the field of ES. There are a large number of modelling approaches and a wide range of existing models that can be used for ES assessment. Modelling has considerable potential to evaluate both the ecosystem structure and function underlying ES and the supply and demand for ES themselves. Furthermore, modelling provides the potential to explore the impacts of environmental change and management on the future provision of ES through scenarios, making them vital tools for ES decision support.

Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. or any other Government or by the authors of this article.

Bagstad KJ, Semmens DJ, Waage S, Winthrop R (2013) A comparative assessment of decision-support tools for ecosystem services quantification and valuation. Ecosystem Services 5: 27-39.
Christin ZL, Bagstad KJ, Verdone MA (2016) A decision framework for identifying models to estimate forest ecosystem service gains from restoration. Forest Ecosystems 3: 3.
Dunford RW, Smith A, Harrison PA, Hanganu D (2015) Ecosystem Services in a changing Europe: adapting to the impacts of combined climate and socio-economic change. Landscape Ecology 30(3): 443-461.
IPBES (2016) Summary for policymakers of the methodological assessment of scenarios and models of biodiversity and ecosystem services of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Ferrier S, Ninan KN, Leadley P, Alkemade R, Acosta LA, Akcakaya HR, Brotons L, Cheung W, Christensen V, Harhash KA, Kabubo-Mariara J, Lundquist C, Obersteiner M, Pereira H, Peterson G, Pichs-Madruga R, Ravindranath NH, Rondinini C, Wintle B (Eds. ) Secretariat of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services, Bonn, Germany, 32 pp.
Schroter M, Remme RP, Sumarga E, Barton DN, Hein L (2015) Lessons learned for spatial modelling of ecosystem services in support of ecosystem accounting. Ecosystem Services 13: 64-69.
Wainright J, Mulligan M (Eds) Environmental Modelling: Finding Simplicity in Complexity, 2nd Edition, Wiley, 494 pp.

The complexity of natural systems and the interactions between nature and society impedes the use of state-of-the-art, data-driven, process-based techniques for ecosystem service (ES) modelling. Instead, simplified, pragmatic approaches can be used to provide initial estimates of ecosystem service delivery. Although simplification leads to an increase in model output uncertainty, many modelling approaches, however complex, often do not take uncertainties into account. Despite their apparent simplicity, Bayesian Belief Networks (BBN) do take uncertainty into account and, as a result, are worthy of attention.

Bayesian Belief Network models are graphical probabilistic models that conceptualise the system being represented as a chain of causal relations, visualised as a Directed Acyclic Graph (DAG). Such a graph consists of nodes that represent the system's variables and arrows that represent causal relations amongst them. Variables are typically discrete and relations amongst them are quantified through probabilistic rules, captured as conditional probability distributions. These distributions can be derived from data, from expert knowledge or a combination of both.

An example BBN that enables an analysis of how our estimate of wood production would change, given information about land use and soil type, is provided in Figure 1. The first step of the model development process consists of selecting suitable variables and defining putative causal relations. By assuming that land use and soil type are the most important drivers that determine the production of wood, the model's variables are restricted to 'soil suitability', 'land use' and 'wood production'. By assuming that soil type and land use both influence wood production and that both variables are independent, the structure of the graph is defined (Figure 1). To implement the model, probability distributions need to be defined: unconditional ones for the input nodes, conditional ones for the others. By combining the information captured in the model's conditional probability tables (CPTs) with the initial probability distributions of the network's input nodes, probability distributions for other nodes can be calculated based on Bayes' theorem, which describes the conditional probability of an event. There are a number of software tools available that enable users to make these calculations automatically. The calculated probability distributions are represented as so-called belief bars in the model (Figure 1).

Figure 1.

A model example illustrating the structural components of a Bayesian Belief Network: (a) the directed acyclic graph (DAG) and (b) the conditional probability tables (CPT).

The application of BBNs generally consists of inserting new information or evidence in one or more nodes of the model and, subsequently, analysing the resulting belief changes. This new information can be deterministic or probabilistic depending on whether the information implies that a state is exactly known or not. Figure 2 provides two examples of inserting deterministic evidence in the model that was introduced in Figure 1. When evidence is inserted in one of the input variables of the model, the model will be run in predictive mode and will predict effects of input changes (Figure 2a). Knowing that the soil suitability is high, our belief in high wood production will increase substantially. Our belief in the 'zero' state, however, will not change, as we still know that 10% (this information has not changed) of the forests are reserves and, thus, do not produce wood. When evidence is inserted in the output node, models are run in diagnostic mode and will predict causes instead of effects (Figure 2b). If we know that wood production is high, we can infer that soil suitability will be high with a high probability. Moreover, based on the inserted information, we can infer that the forest stand being considered is definitely not a forest reserve.

Figure 2.

Predictive and diagnostic belief updating in a Bayesian Belief Network model.

Although BBN models have been used since the 1980s, applications were restricted to medical diagnosis, where BBNs were used to combine probabilistic information on disease occurrence with probabilistic information on symptom development to support the process of reaching a diagnosis. Late in the 1990s, BBNs were introduced in the environmental modelling domain, predominantly because of their ability to explicitly account for uncertainty, an important aspect when natural processes are being modelled. A second important reason for their adoption was the potential of the technique to integrate expert knowledge in the modelling process. Expert knowledge can be used to develop the network structure or to populate the model's CPTs with subjective probabilities, which are also referred to as beliefs. This functionality is especially useful in case variables need to be included for which no supporting data are available, a frequently occurring problem in ecosystem service modelling. A final strength of the modelling technique is its graphical nature. Due to this feature, BBNs are transparent models that are relatively easy to grasp. This means that non-expert stakeholders can be involved in model development.

Another important advantage in the context of ecosystem service modelling is that BBNs fit extremely well in the 'ecosystem services cascade' which has been used as a basis for many ecosystem service studies (see Chapter 2.3) (Figure 3). The idea that ecosystem benefits are generated through services, services through functions and functions through the biophysical structure of the environment, closely resembles a chain of causal relations which can be easily modelled in a BBN.

Figure 3.

General Bayesian Belief Network structure for ecosystem service modelling.

Although a linear representation of the ecosystem service production process might facilitate system understanding, in reality most ecosystem service delivery processes are non-linear and involve a range of feedbacks which BBNs cannot easily take into account. Developing several models for successive time steps of a system and chaining them afterwards is a 'workaround' that mimics feedbacks with BBNs. Such time-sliced models, however, often become very complex and lack transparency. Another important drawback of BBN is the obligation to discretise continuous variables so that they can be represented as having states in a node. To minimise information loss discretisation methods need to be chosen carefully. Nevertheless, for some applications, nodes with discrete states are more easily understood than continuous variables. If discretisation thresholds are chosen in accordance with the aim of the model, discretisation does not necessarily lead to information loss. Where information loss is not acceptable, more complex software packages are available that enable the use of continuous variables in BBNs.

The drawbacks discussed above all suggest that BBNs are less suitable for modelling complicated processes than some other approaches. Thus, for well-studied services, the added value of using BBNs instead of process-based, validated models is low. The real strength of BBNs lies, however, within the integration of well- and poorly-studied services in one integrated model. Such integrated models might deliver additional insights into trade-offs and synergies among services. In addition, their graphical nature can facilitate stakeholder involvement and social learning, two objectives that are difficult to attain using conventional modelling approaches.

To determine which variables need to be included in the model, a variety of knowledge sources can be used. Domain experts can be consulted to select variables that are important for biophysical modelling of service provision, while stakeholders can be consulted to include social interests, i.e. the ecosystem services that are considered important in the study area, or the values associated with them. The type of endpoint being modelled is also an important aspect to consider when including ES in the model. These endpoints can be biophysical quantities, scores that represent social values or monetary values of generated benefits. Modelling the full ES cascade, up to the final benefits, can be attained by integrating studies from different research fields such as economics and sociology.

To select input variables, two things need to be considered, namely the management options that need to be evaluated by the model and whether spatial application of the model is desired. For spatial modelling applications, spatial data on the model's input nodes or on their proxies need to be available. To make management evaluation possible, all variables that are influenced through management and that impact ecosystem service delivery need to be included.

Discretisation of continuous variables is the next step in the model development process. In general, the number of states needs to be kept as low as possible. As the number of states directly affects the complexity of the model's CPTs, less effort needs to be invested in CPT quantification in case the number of states is kept low. Thus, a balance needs to be found between reducing model complexity and minimising information loss. Additionally, in case CPTs are learned from data, the number of states needs to be restricted to ensure that sufficient information is available for all state combinations.

To develop the structure of the model, experts are often consulted. To integrate local knowledge in the model structure, stakeholders can also be consulted. As stakeholders and experts are generally not aware of a model's technical restrictions (e.g. the fact that feedback loops cannot be included), modellers need to guide the model development process and, if necessary, adjust the structure afterwards. Although data can be used to create model structures and estimate probabilities using learning algorithms, relations that are defined through this process are not necessarily a result of causality and, therefore, they are sometimes hard to interpret.

Quantification of CPTs is the final step towards implementing the model. A broad range of knowledge sources can be consulted for this, including expert and stakeholder knowledge, empirical equations, simulations with existing models, literature data and field data. Although data might seem the most objective way to quantify a CPT, datasets are often not sufficient to fully quantify a model's CPTs. In these situations, experts can be consulted to define prior CPTs and data can be used to update CPTs; this is a typical Bayesian workflow. Aside from CPT quantification, the probability tables of the input nodes also need to be quantified. If input nodes represent spatial variables, histograms of spatial datasets can be used to populate these probability tables.

To increase the credibility of a Bayesian Belief Network, model validation needs to be performed. To evaluate a model's predictive performance, a broad range of validation metrics are available, similar to those extensively used in other modelling domains. The predictive performance of BBNs, however, is generally low compared to other techniques. While most models only focus on performing one specific task optimally, BBNs try to approximate the joint probability distribution over all variables, mostly at the expense of their predictive performance. Predictive performance is, therefore, generally not the most important aim of a BBN model, especially in the field of ecosystem service modelling. Other evaluation criteria include the ability of the model to describe a system, to enhance social learning and to facilitate decision-making. To evaluate those aspects, evaluation through experts and stakeholders might be more appropriate. The consulted experts and stakeholders for model evaluation are preferably not those consulted during model development.

To perform the above tasks, a range of software packages are available. Frequently used software packages in the ecosystem services modelling domain are 'Netica' and 'Hugin'. They both provide a user-friendly graphical user interface for model development that can potentially be used with stakeholders. Most packages also include algorithms that can be used to train and validate models using existing datasets. Furthermore, through application programming interfaces (API), software packages can be extended with all kinds of tools. Following this approach, BBNs can, for example, be coupled to geographical information systems (GIS). This is an important functionality when BBNs are used for ecosystem service mapping.

As illustrated in this chapter, BBNs have much potential for modelling and mapping ES. They operate at an intermediate level of complexity which makes them especially useful where the volumes of available data and knowledge are not sufficient for empirical or process-based modelling. Additionally, BBNs are useful tools to help structure the available knowledge into comprehensible ways that can support social learning and stakeholder participation in ecosystem service modelling and management studies.

Cain J (2001) Planning improvements in natural resource management. Guidelines for using Bayesian networks to support the planning and management of development programmes in the water sector and beyond. Wallingford: Centre for Ecology and Hydrology.
Haines-Young R (2011) Exploring ecosystem service issues across diverse knowledge domains using Bayesian Belief Networks. Progress in Physical Geography 35(5): 681-699.
Landuyt D, Van der Biest K, Broekx S, Staes J, Meire P, Goethals PLM (2015) A GIS plug-in for Bayesian belief networks: Towards a transparent software framework to assess and visualise uncertainties in ecosystem service mapping. Environmental Modelling & Software 71: 30-38.
Landuyt D, Broekx S, D'hondt R, Engelen G, Aertsens J, Goethals PLM (2013) A review of Bayesian belief networks in ecosystem service modelling. Environmental Modelling & Software 46(0): 1-11.
Potschin M, Haines-Young R (2016) Defining and measuring ecosystem services. In: Potschin M, Haines-Young R, Fish R, Turner RK (Eds) Routledge Handbook of Ecosystem Services. Routledge, London and New York, 25-44.

Node - Graphical representation of the system variables in a Bayesian Belief Network model.

State - A value, discrete class or qualitative level to which a variable can be assigned. Each variable in a Bayesian Belief Network model has a set of states it can manifest.

Probability distribution - The set of probabilities assigned to the states of a variable that express the probability that the variable equals one of its states. This set of probabilities always sums up to 1.

Arrow - Graphical representation of the causal relations amongst the system variables in a Bayesian Belief Network. Each arrow flows from a parent node to a child node.

Conditional probability - The probabilities that quantify the model's causal relations and express the probability distribution of a child node given the status of its parent nodes.

Conditional probability table or CPT - A table that contains probability distributions over a node's states conditional on all possible combinations of its incoming nodes' states.

Directed acyclic graph or DAG - Graphical representation of the system being modelled by means of nodes that represent system variables and arrows that represent causal relations among the system variables.

Expectations towards ES science and application are very high. The global socio-ecological challenges which researchers are aiming to tackle are both urgent and important. Still, the amount of trust and public resources going to ES studies and mapping is relatively high compared to their current impact on solving real world problems.

Applied ecology and sociology are considered complex fields, combining several disciplinary frameworks, ways of thinking and related methods. ES, at the crossroads of applied ecology, economy, sustainability science and social sciences, can be defined as "super-complex". Super-complex or so-called wicked problems require engagement of several theoretical disciplines and practitioners in actual implementation from the very onset of the problem-solving process.

What looks like just a simple ES map is often a complex combination of selected quantitative data, proxies and expert estimates, qualitative judgements, theoretical assumptions, technical choices and communicative visual goals (see Chapters 3.3 and 6.4). The quality of the actual mapping process directly determines the qualities of the map in all its aspects (credibility, relevance, clarity, usefulness; see Chapter 5.4). Creating a map which lives up even to minimal real world application ambitions obliges the involvement of 'experts' to legitimise, clarify, improve and validate maps to be relevant for any specific application context.

Delineating who is an expert and who is not is not straightforward. From the above, it is clear that, when solving real world problems, merits of diploma and discipline are not enough by far. A bright GIS technician is certainly a required expert, but without complementary input from the ecology expert, the modeller, the economist and the social scientist, there is actually nothing to map or to interpret. Also, without the local or topical experts to put the socio-economic and natural science theories into a specific context, maps will be hard to validate. Moreover, without the expert who connects specific policy demands, cultures and know-how, implementation of the maps into actual solution strategies will rarely happen. And finally, deciding on societal importance of issues or values of specific ES to decide upon trade-offs requires input from policy makers and/or the direct end-users of these services.

All these types of knowledge are indispensable for the mapping process, and not necessarily related to education level or strictly technical skills. The central idea is that all experts - or knowledge-holders - need to be thoughtfully engaged.

From the point of view of a technical mapping project, involving experts is often regarded as costly, tedious and complicated. We will show that structural expert involvement will add value to the whole process of map creation and effective problem-solving. Three examples of selective engagement are discussed here. The section, following this discussion, returns to address the more profound expert engagement of trans-disciplinary research.

Mapping ES in the context of real world problem-solving needs to go further. Structural engagement of experts departs from a different paradigm. The underlying principle is that there is no de facto distinction between experts and laymen, or between stakeholders and researchers. All people involved in, or potentially affected by, the ES mapping project are stakeholders as well as experts in a certain aspect.

Such a trans-disciplinary viewpoint has two immediate consequences: first, the researchers mandated to perform the mapping project depart from a humble attitude (see Chapter 5.4). Second, experts/stakeholders outside of the actual project team are 'promoted' to the level of potentially indispensable knowledge-holders and project-owners. These include people commissioning the project, topical experts on certain ES, technical experts on different methods, experts on local or thematic context into which the mapping project is framed and people actually depending on ES.

The above does not mean, of course, that every mapping project should involve large numbers of experts throughout the project in order to be effective. The actual number of experts is not the issue here, but it is their competence, diversity, qualification and role they have in the project. In the following section, a theoretical illustration of a mapping project's cycle is presented. This example imagines an ideal project without issues of policy restrictions or budgetary constraints.

• Clear goals. Being effective requires the right product, produced in the right way. Clearly formulated goals are essential.
• Diversity. The best people should be identified with the diverse skills and knowledge types needed. Consider them equal regardless of their diplomas and promote this attitude.
• Facilitation. Do not think that a trans-disciplinary process will run itself. Project facilitation is a skill, and skilled people will be needed to keep the process running smoothly.
• Parsimony. Do not overdo it. Weigh costs and efforts against stakes. Be pragmatic when needed, but without forsaking the project goals. Adapt unrealistic goals to more realistic objectives.
• Testing and evaluation.
• Do not expect that your team will produce a perfect product at the end of the project. Look for the weaknesses in the project and address them. Test maps as soon as possible and avoid the trap of self-evaluation. The sooner a weakness or failure is identified, the greater chance there will be of finalising your project with a high level of success and impact.

Bradshaw GA, Borchers JG (2000) Uncertainty as information: narrowing the science-policy gap. Conservation Ecology 4(1): 7.
Cornell S, Berkhout F, Tuinstra W, Tabarae JD, Jager J, Chabay I, de Wit B, Langlais R, Mills D, Moll P, Otto IM, Petersen A, Pohl C, van Kerkhoff L (2013) Opening up knowledge systems for better responses to global environmental change. Environmental Science & Policy 28: 60-70.
Drescher M, Perera AH, Johnson CJ, Buse LJ, Drew CA, Burgman MA (2013) Toward rigorous use of expert knowledge in ecological research. Ecosphere 4: 1-26.
Gunderson LH, Holling CS (Eds.)(2002) Panarchy. Island Press, Washington Covelo London.
Hay I (2010) Qualitative Research Methods in Human Geography. 3rd Edition. Oxford University Press.
Jacobs S, Burkhard B, Van Daele T, Staes J, Schneiders A (2015) The Matrix Reloaded: A review of expert knowledge use for mapping ecosystem services. Ecological Modelling 295: 21-30.
Seidl R et al. (2013) Science with society in the anthropocene. Ambio 42(1): 5-12.
Voinov A, Seppelt R, Reis S, Nabel JEMS, Shokravi S (2014) Values in socio-environmental modelling: persuasion for action or excuse for inaction. Environmental Modelling and Software 53: 207-212.

Ecosystem services (ES) originated as a concept that reflects the value of nature for humans and provides additional reasons for protection and sustainable management of ecosystems (see Chapter 2.3). Many ES face spatially explicit pressures or rely on anthropogenic contributions such as technology, energy or knowledge. ES maps can help to uncover risks for ecosystem health, unsustainable use of potentials to provide a service, harmful impacts on a landscape, impaired spatial flows of ES as well as mismatches between ES supply and demand (see Chapter 5.2). Such information can indicate where to improve ES provision and where to prioritise nature and biodiversity conservation.

Multiple components play a role in ES provision and use which can be mapped, assessed and monitored. ES can be mapped and assessed using quantitative indicators or qualitative estimations. ES mapping and assessment include ecosystem properties and conditions, ES potential, ES supply, ES flow and ES demand which we generically define in the next sections.

The framework presented here aims to depict different aspects of ES important for mapping. Our framework bridges variously interconnected ecosystems and socio-economic systems, including the interactions between their components. Figure 1 highlights aspects of ES which can be considered relevant for mapping. ES are generated in the context of different aspects or components, which are interrelated, but can be mapped separately.

Figure 1.

Mapping aspects of ES (own illustration, adapted version of the the ES cascade by Haines-Young & Potschin (see Chapter 2.3), Wolff et al. 2015, Bastian et al. 2013). Bold grey: subjects relevant for mapping; dashed: may be mapped; thin: additional aspects for which mapping could be developed.

Ecosystem properties and conditions provide the ecological basis for ES potentials which, together with human inputs, form a capacity of a social-ecological system to provide ES (ES supply). ES flows (i.e. the actual use of ES) can be a fraction of this supply, or be higher in case stocks are depleted or ecosystems are unsustainably used. Demand for ES steers ES flows, i.e. without a demand for a service, there is no actual use. This demand can, however, be higher than actual flow, for example, in cases where societal preferences for specific services remain unsatisfied. Within the socio-economic system, benefits arise from several kinds of ES use depending on the demands of concerned people. Feedbacks from the socio-economic system such as land use change, landscape maintenance or environmental pressures, affect the ecosystem and thereby the ES supply. The following sections explain these terms in detail.

Definition: Properties describe the character, structure and processes of an ecosystem. Conditions refer to the integrity and health status of an ecosystem which determine its ability to generate ES (see Chapter 3.5). Land use or land cover provide the basis of many ES maps. Beyond that, ecosystem properties such as soil type, slope gradient and inclination, climate conditions and the position in relation to a shoreline or within a watershed are properties that essentially control the supply of many ES. Features of landscape structure like density of certain objects, edge conditions, connection and shape of areas can also be very important. Ecosystem conditions, however, comprise much more: for instance, the load of pollutants, species composition and health may be crucial preconditions for ES.

Delimitation: Properties and condition reflect both the natural ecosystem state and the type of ecosystem as result of a specific land use. Since the condition for ES supply differs between specific ES, the scope of related assessments has to be defined very carefully per ES.

Necessity and applicability: Indicators for ecosystem properties and conditions should be applied to different protection goods or land use classes. They are relevant because they provide the spatial and physical preconditions for ES (see Chapter 2.2). ES potentials can, for example, give a reference point for planning and scenarios (see Chapter 7.2). Both the individual patches' land use and land cover and the configuration and arrangement of such patches, are important for ES supply. Therefore, the landscape structure with its mosaic of patches should be considered (see Chapter 5.2).

Possible indicators: Land cover can provide an essential database for ES mapping. The CORINE land cover dataset is often used in European studies (see Chapter 3.5). At national level, land use data from land survey or habitat mapping often are available. Additional data need to be integrated in more detailed evaluations (see Example 1).

Example 1. Wood-dominated ecotones and non-fragmented forests

Large contiguous areas of woodland are vital for nature protection by offering habitats for animals and plants and provide people with areas for relaxation. The size of uninterrupted woodland, not dissected by roads and railways, is an important criterion for ecosystem conditions.

Ecotones are transitional areas between habitats. As such, they are home to a particularly rich variety of species, not only those of the adjacent communities but also species that have become specialised to the ecotone itself. In open landscapes, such elements are important as habitat for pollinating insects and for other beneficial organisms. At the same time, a landscape with high proportions of such elements is very attractive for human recreation. In this context, landscape configuration with ecotones is an indicator for ecosystem condition. The calculation of the perimeter of forest-dominated ecotones takes account of all hedges, tree rows and the margins of small copses as well as all forest margins (see Walz 2015).

Ecosystem properties and conditions are directly linked to the state of biodiversity. A high level of biodiversity - in most cases - underpins the supply of multiple ES (see Chapter 2.2).

Definition: ES potential describes the natural contributions to ES generation. ES capacity is often used synonymously. ES potential measures the amount of ES that can be provided or used in a sustainable way in a certain region given current land use and ecosystem properties and conditions. It is recommended to regard this potential for a sufficiently long time period.

Delimitation: The (natural) ES potential is often supplemented by human system inputs to generate ES supply (see Section Human inputs). The actual provision (co-production) of ES (flow) sometimes includes large human efforts, is strongly dependent on technological refinement and can be very difficult to determine.

Necessity and applicability: In terms of ES potential, the ecological carrying capacity and resilience need to be considered. ES potential allows the distinction between a realised ES and the opportunities and limits of use which is often meaningful for planning purposes, scenarios and management issues. Sometimes, an indicator for ES potential can help to better understand and calculate physical indicators for regulating ES supply.

Possible indicators are, for example, the Muencheberg Soil Quality Rating (SQR), metrics for relief diversity and the share of water bodies as part of landscape aesthetics as well as proxies for processes such as groundwater recharge rates.

ES potential is particularly applicable for planning, management and predictive research purposes. Since it is conceptualised hypothetically and for the long term, ES potential should not be assessed for short time periods (such as for only one season). Preferably, ES potential should be orientated on natural regeneration rates. Direct human interventions such as fertilisation, technical energy inputs or breeding and genetic engineering should not be considered as contribution to ES potentials. In contrast, land use type (grassland, field, forest, settlement) and the consequences of long-lasting or very strong impacts such as mining have to be considered naturally. A distinction of a real 'natural' state that contributes to ES is not straightforward.

Definition: Supply is the provision of a service by a particular ecosystem, irrespective of its actual use. It can be determined for a specified period of time (such as a year) in the present, past, or future.

Delimitation: The amount of ES supply depends on natural conditions and often on human inputs (see below), such as land management contributions, knowledge and technology. Though there are some ES without human co-production, they may nevertheless depend on ecosystem preservation. ES supply also includes stocks of natural assets as starting points of the flows of material, energy, information and organisms as results of both ecosystem potential and human co-production.

Necessity and applicability: ES supply is a central subject to be mapped and can be considered a complement to ES demand (see below).

Possible indicators are average yields of crops, wood regrowth in forests, flood retention in catchments or floodplains, amount of carbon stored in soil and vegetation, relative reduction of noise or pollutants, aesthetics of scenery.

Definition: Flow is a measure for the amount of ES that are actually mobilised in a specific area and time. Driven by a demand for a service, ES supply is turned into ES flows (Figure 1). In case both ES supply and demand are quantified using the same dimension and unit, a quantitative comparison is possible (supply-demand budget calculation). Flow can, in a more tangible meaning, also involve a movement of material, energy or information across space. In case supply and demand are not spatially congruent, flow maps can show spatial connections between Service Providing and Service Benefiting Areas (SPA - SBA; see Chapter 5.2).

Delimitation: Service flow can be constrained by an inadequate ES supply which would lead to exceedance of the ES potential. This again may lead to an overuse of given ES potentials, degradation of natural capital or to unmet ES demand.

Necessity and applicability: ES flow maps can unfold spatial mismatches between ES providers and beneficiaries. If there are essential natural processes supporting these interactions between providers and beneficiaries, ES flow mapping gives insights to Service Connecting Areas (SCA; see Chapter 5.2). Their conditions such as possible barriers or other features shaping the flow are items that can be mapped meaningfully.

Possible indicators are fish catch, timber logging, bioenergy gain, groundwater extraction (by wells), flood peak reduction, visitor numbers.

ES flow should particularly be included in integrative supply-demand assessments. There is a broad range of process models (see Chapter 4.4), expert knowledge (see Chapter 4.6) or monetary valuation methods (see Chapter 4.3) which can be applied here.

Definition: Demand is the need for specific ES by society, particular stakeholder groups or individuals. It depends on several factors such as culturally-dependent desires and needs, availability of alternatives, or means to fulfil these needs. It also covers preferences for specific attributes of a service and relates to risk awareness. Demand links ES to particular beneficiaries. This means that without a demand for a service, there is no flow. Beneficiaries express demand and can have the power to translate this demand into an actual ES use. Demands for some ES (such as several regulating ES) might be uncovered, or certain groups of society might be unaware that they actually benefit from an ES.

Delimitation: Demand can be different from flow which measures the actual extraction of a service within a region. Demand can, for example, be higher than flow within that particular region. This means, when demand is realised, it could be fulfilled through services that come from another region. For instance, many provisioning ES (e.g. food, timber, energy) can be imported. The demand for carbon sequestration (ES climate regulation) can be fulfilled by a region with a high potential to sequester carbon or cultural ES such as recreation can be actively used in another region through travel (see Chapter 6.2). The phenomenon of regionally-unmet demand is common to many ES and so far we have only started to understand the long-distance effects between different regions caused by inter-regional ES use. (Regional) demand could also be lower than flow, in case ES are exported. Demand is then expressed by other social-ecological systems while ES flow takes place in the region of interest.

Necessity and applicability: Demand can change over time and can show an uneven pattern across space. As a result, it makes sense to map demand independently from potential, supply and flow. Regional demand can exceed the (regional) supply considerably and, through an increased flow, this could result in unsustainable regional levels of extraction or use of a service so that flow could exceed ES potential. As a consequence, local ecosystems are at risk of overuse or ecosystems in other parts of the world are degraded by land use change (ES footprint).

Possible indicators are vulnerability of people or value of endangered assets for flood risk, desirable attributes for recreation, accessibility and travel costs of visitors, socio-economic valuation and stakeholder perceptions.

Demand involves human preferences which can be determined through questionnaires, but also involves basic needs (e.g. unpolluted air) and actually used ES (e.g. flood protection at a riverside) even when people are not aware of them. Aspects of risk aversion can be based on assumptions, be modelled or by enquiries (stated preferences). In the case of provisioning ES, the beneficiary could be a farmer who benefits from an intact agricultural ecosystem. It could, however, also be the regional population that formulates the demand for locally-produced food.

Definition: Human inputs encompass all anthropogenic contributions to ES generation such as land use and management (including system inputs such as energy, water, fertiliser, pesticides, labour, technology, knowledge), human pressures on the system (e.g. eutrophication, biodiversity loss) and protection measures that modify ecosystems and ES supply.

Delimitation: Human inputs often emerge as harmful impacts to ecosystems caused by monocultural land use, land use change or intensification. Today, most ecosystems and the services they provide are used and influenced by humans.

Necessity and applicability: Humans perform multiple roles in ecosystems acting as managers, but also as co-producers, distributors or beneficiaries of ES.

Possible indicators are land use type and intensity, load of pollutants, material or energy input (such as nitrogen), effort of landscape maintenance, further contributions to ES.

Human impacts are accompanied in many cases by substantial losses of biodiversity.

Particular attention should be paid to human inputs since they may alter ES supply considerably and this impact differs spatially. Not only targeted land use activities influence the integrity of ecosystems, but also the utilisation and improvement of ES can impact other services as well. Resulting ES trade-offs (see Chapter 5.7) are important to review, but are often hard to map.

Depending on the scope of application, ES maps can show different contextual aspects of ES which are spatially heterogeneous in a different way and therefore relevant for ES mapping. Depending on data availability and the policy question or information needs at hand, mapping of one or two of these aspects might be sufficient. It is recommended to map only such aspects that can be derived from reliable data. When monitoring or systematic balance over time is requested, data and indicators have to be double-checked for comparability which can also depend on methods or technology of data collection and on appropriate indicator selection.

Ala-Hulkko T, Kotavaara O, Alahuhta J, Helle P, Hjort J (2016) Introducing accessibility analysis in mapping cultural ecosystem services. Ecological Indicators 66: 416-427.
Albert C, Galler C. Hermes J, Neuendorf F, von Haaren C, Lovett A (2015) Applying ecosystem services indicators in landscape planning and management: The ES-in-Planning framework. Ecological Indicators 61, Part 1: 100-113.
Bastian O, Syrbe R-U, Rosenberg M, Rahe D, Grunewald K (2013) The five pillar EPPS framework for quantifying, mapping and managing ecosystem services. Ecosystem Services 4: 15-24.
Burkhard B, Kandziora M, Hou Y, Muller F (2014) Ecosystem service potentials, flows and demand - concepts for spatial localisation, indication and quantification. Landscape Online 34: 1-32.
Fischer A, Eastwood A (2016) Coproduction of ecosystem services as human-nature Interactions - An analytical framework. Land Use Policy 52: 41-50.
Grunewald K, Herold H, Marzelli S, Meinel G, Syrbe R-U, Walz U (2016) Assessment of ecosystem services at the national level in Germany - illustration of the concept and the development of indicators by way of the example wood provision. Ecological Indicators 70: 181-195.
Jones L, Norton Z, Austin AL, Browne D, Donovan BA, Emmett ZJ, Grabowski DC, Howard JPG, Jones JO, Kenter W, Manley C, Morris DA, Robinson C, Short GM, Siriwardena CJ, Stevens J, Storkey RD, Waters G, Willis F (2016) Stocks and flows of natural and human-derived capital in ecosystem services. Land Use Policy 52: 151-162.
Liu J, Yang W, Li S (2016) Framing ecosystem services in the telecoupled Anthropocene. Frontiers in Ecology and the Environment 14: 27-36.
Nedkov S, Burkhard B (2012) Flood regulating ecosystem services - Mapping supply and demand, in the Etropole municipality, Bulgaria. Ecological Indicators 21: 67-79.
Remme RP, Edens B, Schroter M, Hein L (2015) Monetary accounting of ecosystem services: A test case for Limburg province, the Netherlands. Ecological Economics 112: 116-128.
Termansen M, McClean CJ, Jensen FS (2013) Modelling and mapping spatial heterogeneity in forest recreation services. Ecological Economics 92: 48-57.
Villamagna AM, Angermeier PL, Bennet EM (2013) Capacity, pressure, demand, and flow: A conceptual framework for analyzing ecosystem service provision and delivery. Ecological Complexity 15: 114-121.
Walz U (2015) Indicators to monitor the structural diversity of landscapes. Ecological Modelling 295(1): 88-106.
Wolff S, Schulp CJE, Verburg PH (2015) Mapping ecosystem services demand: a review of current research and future perspectives. Ecological Indicators 55: 159-171.

It is an important feature of natural and cultivated ecosystems that they are not evenly distributed across landscapes, coastal or marine areas and they also vary over time. Ecosystem services (ES) are usually generated by ecological processes within their area of influence such as catchments, habitats, natural regions or land use units. This suggests the need for site-specific assessments. Therefore, each ecosystem service should not only be assessed through considering underlying ecosystem types but also with respect to:

• underlying natural regional conditions (geology, landform configuration, soil, climate, etc.),
• its positional relations to main types of landscape (urban, agrarian, near-nature),
• the configuration (landscape structure) of the corresponding units (catchments, natural regions, etc.) with natural resources or land uses,
• the relations between the ecosystem providers of a service and groups of people who make use of it (i.e. beneficiaries) and
• the use, management and maintenance of the respective ecosystem.

The holistic approach presented here presumes that complex ecological systems underlie the production of most ES, which can be envisioned as SPUs, or Service Providing Units. For mapping purposes, such an SPU should be regarded as a spatial unit. This opens the way for applying landscape-scale geographic assessment methods based on landscape units corresponding to the area of influence (see above).

In order to avoid terminological confusion with different SPU variants, we term the spatially defined complexes as Service Providing Areas (SPA) (see Text Box 1). SPAs are a promising basis for an inclusive approach of ES at the landscape scale.

As the service providing areas defined above include entire ecosystems, their constituent populations and underlying biophysical characteristics, the best way of capturing them spatially is as ecological spatial units (e.g. living spaces, water bodies or soil areas) or as area of influence of the respective processes (e.g. catchment areas, flood plains). From this point of view, such biophysically-delineated areas are more suitable for analysis than administrative units.

In the spatial analysis framework, however, not only the SPAs are of interest, but also the regions to which their benefits accrue. For example, one might ask: where is the benefit of a given ecosystem service needed? In addition to the service providing areas (SPAs), Service Benefiting Areas (SBAs) should thus be defined in which beneficiaries receive the service (see Text Box 1). In a spatial framework, urban areas, rural settlement areas and especially administrative units could be considered as SBAs. Factors such as population density, social facilities (e.g. schools, hospitals but also parks for recreation) and built structures (residential, commerce or industry buildings) or the number and size of the households, are important as indicators (e.g. per household measures of demand for specific ES).

The service providing and service benefiting areas may overlap, but significant spatial differences are also possible (see example in Text Box 2). If the service providing and service benefiting areas are not adjacent, the properties of the connecting space can have an influence on the provision of the service (see Text Box 2). We include such an interstitial space between service providing and service benefiting areas in our considerations under the term Service Connecting Area (SCA) (cf. Fig. 1).

Figure 1.

Types of spatial relations of Service Providing Areas (SPA), Service Benefiting Areas (SBA) and Service Connecting Areas (SCA) (adapted and extended from Fisher et al. 2009; Syrbe and Walz 2012).

The following fundamental types of relations between the service providing and the service benefiting areas can be distinguished (Fig. 1):

1. 'in situ': the two area types are identical, i.e. the ES are supplied and in demand in the same area (e.g. the population uses the groundwater of its settlement area),
2. 'central demand': the surrounding area provides for / impacts on a central demand area (e.g. a settlement benefits from supply of fresh and cold air which is generated by open spaces in the surrounding),
3. 'omni-directional': the service benefiting area surrounds a service providing area independent of direction (e.g., farmland benefits from hedges as a living space for beneficial insects),
4. 'directional without dependency on a slope': the service benefiting area is situated "behind" the service providing area, protected as it were with respect to the predominant impact direction (e.g. a residential area protected against traffic noise by a forest),
5. 'directional downslope': the service benefiting area is situated downhill (downriver) from the service providing area, i.e. the service is dependent on gravitational processes (e.g. cold air, water, avalanches) and
6. 'spatially separated': e.g. drinking water, food production, recreational areas. There can be different connective methods, e.g. natural hydrologic flow within watersheds, infrastructure (pipes/aqueducts) or road/trail networks.

The relation types d, e and f can especially exhibit considerable service connecting areas.

Once SPAs, SBAs and SCAs are defined for each ecosystem service (see Table 1 for examples), they can be described in greater detail according to their properties such as structure, type and characteristic of the spatial situation. The comprehensive characterisation of a service-providing area should contain at least the following information:

Table 1.Download as CSV 

Examples for Ecosystem Services, which depend on lateral or vertical landscape processes, with associated Service Providing Areas (SPA), Service Benefiting Areas (SBA), Service Connecting Areas (SCA) (adapted from Syrbe and Walz 2012). * Type of service: P - provisioning services, R - regulating services, H - habitat services, C - cultural services

Ecosystem Services *	Service providing area (SPA)	Service connecting area (SCA)	Service benefiting area (SBA)
P groundwater recharge	Arable land, wood, grassland, wetlands and other open land in a groundwater basin	Groundwater flow paths (with possible contaminated sites and risk areas for the protection of groundwater)	Settlement areas, irrigated areas
P drinking water	Headwaters and catchment areas	Bodies of groundwater, streams, rivers, (pipelines); (see Chapter 6.2)	Settlement areas, industry (for production, less for cooling)
P fodder for grazing animals	Grassland and forage crops	Pastoral paths	Farms
R protection against snowdrift, storm	Forest, road trees, shrubs, hedges	Embankments at roads and railway lines	E.g. roads, railway lines and runways
R erosion prevention - by wind - by water	Forest, hedges, bushes, trees and shrubs (grassland, permanent crops)	Field edges, gullies	Areas under cultivation, water reservoirs
R flood prevention	Forest, ponds, wetlands, etc. in flood generation areas	Floodplains above benefiting areas	Built-up area in the floodplain
R local climate regulation (cold/fresh air)	Open land, parks above cities	Slopes (with or without obstacles) around a city	City in valley
R noise reduction	Roadside greenery, wood, ramparts	Areas (if appropriate buildings) around the source of noise	Residential and recreational area
R avalanche and landslide prevention	Forest above residential or recreational areas	Slope area	Residential or recreational areas below steep slopes
R pollination	Nesting habitats of insects	Radius of flight and foraging habitat	Farms with crops requiring pollination
R pest control	Nesting habitats of predators	Foraging habitat	Crop land
R stream water purification	Surface water bodies, wetlands	Water catchments	Residential or recreational areas
C appreciated scenery	Viewsheds (areas which can be seen from a particular site)	Line of sight, open country	Settlements and touristic infrastructure
C recreation activities	Surface water bodies, mountains, wood	Road and path network between SPA and SBA	Touristic lodging units

1. a site characterisation and classification of the potential for providing the service with respect to the required natural processes and their dynamics,
2. an analysis of the human usage patterns also regarding their internal structure, for example, through landscape metrics and
3. the consideration of the conditions regarding location and neighbourhood according to the respective processes.

The comparison and the positional relations of the service-providing areas to the associated service-benefiting and service-connecting areas form another focal point of the spatial investigation. The characteristics of the service-providing areas are primarily founded on the natural sciences, since they relate to the beneficial natural resources and - if applicable - to those processes which ensure their regeneration. Furthermore, the analysis needs to examine whether investments (protection or management measures) are necessary in order to preserve the service supply capacity. If so, the kind and frequency of the maintenance measures should be determined and the necessary cultivation rules should be known. If the natural capital is reducible (by consumption), the natural capacity to regenerate has to be determined in order to adapt the consumption to the regeneration rate if a sustainable resource management is to be achieved.

The characterisation of the service-benefiting areas also includes further analysis from the social sciences. Especially, users' demands have also to be incorporated into analysis of SBAs. Depending on the area of investigation, the demands, preferences and values of the benefiting population groups represent indicators for the demand for the ES. The size of a population group is an important basis for determining and assessing the service. However, whether threshold values should or must be defined is also crucial for the assessment. This can be the case, for example, with respect to people endangered by natural disasters (e.g. restriction of construction areas because of flood hazards) or to unsustainable resource harvest rates (see above). Moreover, limited or widely demanded resources require clear rules for accessing them in order to avoid "free-rider effects" (benefiting from a service without contributing to it) and misinvestments. The type of access (private, common or public) to a resource and the possibility of excluding people from such access determine the marketability or non-marketability of the ES.

Even if a service connecting area does not exist separately because service-providing and service-benefiting areas overlap, an analysis of the connection properties is useful, since horizontal transfer processes are influenced by landscape characteristics. If there is an interstitial space between the service-providing and the service-benefiting area, this connecting space first needs to be determined more closely, which can at times be difficult. This can be modelled for example using the transport and transformation paths of substances, energy, biota and possibly also information.

Depending on the type of the ecosystem service that is being assessed, very different spatial units can be considered for service providing, benefiting and connecting areas (Table 2). Moreover, certain actors whose actions significantly contribute to the benefit may participate in the service provision or the transfer of benefits. Stimulating their economic interest (remunerating instead of disadvantaging them) is an essential goal of the ES approach.

Table 2.Download as CSV 

Example for the suitability of landscape units for designation of provision, benefiting and connecting areas.

Nature and origin of the service	Spatial unit
Generated by specific species	Suitable habitats
Based on biophysical resources	Natural regions
Depending on specific landscape mosaic	Spatial unit with comparable spatial features
Generated by an abiotic process	Area of influence of this process
Dependent on specific land management practices	Management units
Rooted in history and culture	Units of the historic cultural landscape
Hydrologic services	Water catchment areas
Demand for ecosystem services by people	Administrative units

1. Examples of different types of spatial units for capturing and assessing individual ecosystem services are:
2. Single surfaces (patches), landscape elements such as arable field sections or forests provide the spatial basis of reference most frequently used for the overall assessment.
3. Administrative units are useful if data are analysed which have such units as a reference basis, such as socio-economic data or legal framework conditions.
4. Water catchment areas are typical morphological units which can be delimited well using GIS on the basis of a digital elevation model (see Chapter 3.4). They therefore represent the common reference units for hydrologic services.
5. Natural units should be drawn upon if natural properties such as soil conditions, surface forms, climate, geology or vegetation mainly determine a service.
6. Landscape units, delimited not only according to natural conditions but also according to land use, are usable for most services, especially for site-scale assessment of ES. These units serve as a reference basis for assessing different aspects of diversity, for determining the spatial heterogeneity and as a spatial framework for practical management.

A decided advantage of a spatial-structural approach is that it makes it possible to understand ES beneficiaries and flows. As soon as it is possible to determine the beneficiary of a service, the benefit of such a service can also be identified. This especially applies when the provision and the use of such services do not spatially overlap. Only this knowledge makes it possible to design incentive systems and fair payment for the providers when they deliver this service (see Chapters 7.2 and 7.3). This is also the prerequisite for ES' availability in the long term.

Bagstad KJ, Villa F, Batker D, Harrison-Cox J, Voigt B, Johnson GW (2014) From theoretical to actual ecosystem services: Mapping beneficiaries and spatial flows in ecosystem service assessments. Ecology and Society 19(2): 64.
Bastian O, Grunewald K, Syrbe RU (2012) Space and time aspects of ecosystem services, using the example of the EU Water Framework Directive. International Journal of Biodiversity Science, Ecosystem Services & Management: 1-12.
Burkhard B, Kandziora M, Hou Y, Muller F (2014) Ecosystem Service Potentials, Flows and Demands - Concepts for Spatial Localisation, Indication and Quantification. Landscape online 34: 1-32.
Fisher B, Turner RK, Morling P (2009) Defining and classifying ecosystem services for decision making. Ecological Economics 68(3): 643-653.
Grunewald K, Bastian O (Eds) (2015) Ecosystem Services. Concept, Methods and Case Studies. Berlin, 312 pp.
Syrbe RU, Walz U (2012) Spatial indicators for the assessment of ecosystem services: Providing, benefiting and connecting areas and landscape metrics. Ecological Indicators 21: 80-88.

Mapping ecosystem services (ES) is often seen as a static three-dimensional problem where space (x, y) and the value of a given ecosystem service (z) are referred as the main factors of analysis. A wide group of examples that follow this approach populate current scientific papers, books and technical reports. The issue with these assessments is that they often consider that the value of a given ecosystem service in a particular place is (a) stable in time or (b) it already encapsulates the effects of the underlying ecological processes/cycles.

Under a spatial notation (x, y, z), ecosystem service supply is represented by a magnitude, a spatial distribution or configuration and an extent. Although perceptive, this approach does not consider that specific ES are often supplied in different moments in time (e.g. pollination, food production and flood regulation) and generate benefits that can be equally temporally displaced (e.g. in flood regulation there is a lag of time between the accumulated decrease of runoff [superficial water flow] by percolation and the actual reduction of the downstream flood plain). This results from the fact that ecological processes/cycles vary through time and, because most ES (namely, production and regulating services) depend on specific ecological processes/cycles, ecosystem service supply is also dynamic. These dynamics can be illustrated by focussing on a specific ecosystem service provider, e.g. a deciduous tree (Figure 1).

Figure 1.

Example of a within-year ecosystem service supply cycle considering a deciduous tree as the focus of ecosystem service supply.

From January to December, the life cycle of a deciduous tree allows for the supply of a relatively large number of ES. Starting in spring, this single tree represents an important support for bird nesting contributing to habitat quality while, at the same time through photosynthetic processes, it captures carbon and other atmospheric pollutants, thus improving local air quality. As time passes, it gives shade for picnics in the summer but it also promotes heat absorption and reduces the albedo effect which helps to reduce heat waves in cities or the probability of fires in forests. At the end of the summer, rain season starts and the same tree contributes to control soil erosion by reducing the erosivity power of precipitation. In the autumn, the leaves fall contributing to local soil fertility. The landscape changes to autumn colours which inspire poets, painters and mountaineers. When winter arrives, the same tree that in summer absorbed heat, now lets the solar radiation pass and thus improves local heat regulation. When isolated, this deciduous tree has a rather narrow potential to supply all of these ES but, when part of a community (e.g. integrated in a deciduous forest), this potential is multiplied and new ES can emerge.

This example serves to show the dynamics and complementarity of ecosystem service supply through time. It also highlights the need to include temporal variations in the assessment of ES, as the likelihood of misrepresenting ecosystem service supply in static assessments is considerable. In fact, time dependency of ES correspond to a very broad and complex issue that includes various time scales, ranging from very short timescales (within a day or a year) to several years, decades or centuries depending on the ecosystem service under assessment. Properly selecting the scale of assessment is fundamental and it mainly depends on the objectives of the assessment and the ecological cycle/process under study. For example, in flood protection, it is possible to focus on hourly variations (if a peak flood is considered and the capacity of vegetation to reduce runoff velocity is studied), monthly variations (if the purpose is to identify hotspots of ecosystem service supply), or yearly variations which can be projected through centuries (if the purpose is to study probability of flood events or projection of trends).

Another dimension of complexity is also the significant mismatch between the potential for ecosystem service supply and the actual ecosystem service supply. This mismatch is also linked to temporal issues.

Consider soil protection as a regulating ecosystem service. In this context, vegetation cover protects soils from being eroded. If vegetation is removed, for instance by harvesting crops, there is an enhanced erosion risk. If we evaluate the temporal dynamics of vegetation cover (here representing the potential to supply soil erosion prevention) and the actual ecosystem service supply (avoided erosion), these variables have two very different temporal distributions resulting in a supply and demand mismatch.

As for the ecosystem service supply, the demand for ES is also dynamic. It usually correlates with the cycles of environmental impact (in the case of regulating services), production cycles (e.g. the requirement for pollination services according to crop cycles), specific consumer demands (e.g. the increase in codfish or turkey demand during the Christmas period), recreation cycles (e.g. the increase in the demand for hiking areas during the summer time), amongst others. The potential differences between the dynamics of demand and supply of ES are among the drivers for over-exploitation of ecosystems making the evaluation of temporal dynamics even more significant.

Assessing ecosystem service potential, supply and demand (see Chapter 5.1) requires a thorough understanding of ecological cycles and ecosystem service mechanisms. Both of these are dynamic and entail the recognition that an ecosystem service is dependent on multiple simultaneously occurring processes with different (often competing) objectives and that ecosystem service supply is secured by different ecosystem service providers with their own specific ecological cycles, targets and trends.

This recognition is critical when assessing ecosystem service supply but it also depends on the objectives of the assessment and on the research question that is being addressed. Within a static approach, the indicators of ecosystem service supply portray a snapshot (an image of a single moment in time). These indicators often neglect the existence of ecological or environmental cycles and dynamics or assume that these are already encompassed within the results obtained. Although these indicators can eventually be used as state or impact indicators they often lack the ability to produce a good representation of ecosystem service supply that is suitable for policy support, land management assessments or other forms of decision-making.

One of the reasons for this, is the inability of static indicators to capture the influence of particular management practices on the overall ecosystem service supply. Or at least this is often only true when using long cycles and when a direct relation between ecosystem service supply and the accumulated effects of specific impacts (e.g. the effect of intensive ploughing on soil erosion) is effectively established. In this example, a static impact prevention indicator can be used to illustrate the spatial distribution of ecosystem service supply but it gives little information regarding the underlying process.

To effectively assess ecosystem service supply, it is essential to implement methodological approaches that consider indicators that vary over time and space. Many examples of these approaches can be found in literature (see the "Further reading" section in this chapter) and more recently StDMs (stochastic dynamic methodologies) are being used to highlight the influence of specific land management strategies on the ecosystem condition and the related ecosystem service supply.

Independently of the chosen method, there are three major dimensions to be considered when implementing a dynamic assessment of ecosystem service supply: i) the significant temporal amplitude of the underlying ecological cycles; ii) co-dependency processes and their impact on the provision of multiple ES; and iii) seasonality.

Ecological processes develop within a wide range of temporal cycles from short- to long-term. Therefore, correctly assessing ES strongly depends on identifying the relevant temporal amplitude that allows the capture of the full extent of ecosystem service supply. Another aspect for consideration is the determination of the relevant temporal amplitude to identify the effect of specific drivers on ecosystem service supply. In some cases, within the same ecological process, one has to look at both the short- and long-term cycles in order to understand the contribution of ecosystem service supply to society and the influence of different drivers. Good examples come from assessing the contribution of ES to mitigate a particular flood event versus determining the mitigation effect in the case of extreme, long-term, events (e.g. a 0.01 probability event such as a "100-year flood").

At the same time, many ecological processes have "multiple" co-dependency relationships between themselves. This dependency is often determined by the cycle of one or more ecosystem components and it is also reflected in ecosystem service supply. For example, both flood regulation and soil erosion prevention depend on the processes by which water percolates into the soil and is retained by vegetation. Although using different processes and interactions, vegetation plays a significant role in the supply of these two different ecosystem services. As in this example, these co-dependence effects often do not necessarily happen at the same time and are therefore often overlooked by ecosystem service supply analysis.

As an example, crop yield depends strongly on water (from infiltration) and nutrient availability (e.g. from nitrification), but the service supply from these three services occurs and has to be quantified at different moments in time.

Related to this is the seasonality of ecosystem service supply and its related benefits. Previously illustrated in Figure 1, the intensity and frequency of ecosystem service supply depends strongly on the seasonality of the ecological processes underlying a given ecosystem.

All of these different aspects contribute to undermine ecosystem service supply quantification and its analysis. When assessing disturbance or recovery dynamics, an assessment of ecosystem service supply should consider at least one or more of these different aspects in order to produce consistent results and to enable the illustration of specific dynamics of change.

Ecosystems evolve over time as they are affected by and react to different human and environmental drivers of change. This evolution can result in cumulative effects for the ecosystem (e.g. the cumulative effect of soil sedimentation in wetlands) but can also allow determination of the influence of specific drivers in relation to specific ecological functions. Here lies the value of trend analysis, the contribution for understanding the past and current development pathways in order to create knowledge about the future of ecosystems.

Current assessments of ES do not always favour the use of time series. This often comes from data limitations regarding the use and availability of contemporary datasets for all system components but also and more importantly, the availability of temporal datasets with an amplitude and a frequency that is relevant for the processes under study. A common limitation is related to the availability of comparable time series of soil datasets or the existence and availability of biodiversity data with relevant thematic, temporal and spatial extent.

Nonetheless, the use of trend analysis corresponds to one of the most valuable tools to identify the determinants of change. Examples of this can be seen through literature (see Further reading for references) using long time series to illustrate the effects of policies, land management, forest fires, amongst others. Figure 2 presents an illustration of a time series of land cover and land use change for a montado landscape in the South of Portugal from which it is possible to calculate long term trends. Such data is of critical importance for understanding changes in ES over time as a result of changes in management and policy implementation.

Figure 2.

Example of land use and land cover change over a period of 61 years in a montado area in the South of Portugal.

At the same time, trend analysis also presents a valuable opportunity to better design and describe future scenarios of ecosystem development. These scenarios are plausible representations of possible future states for one or more components of a system, or as alternative policy or management options intended to alter the future state of these components. Scenario analysis in ecosystem assessments, policy support and decision-making aims at visualising future impacts on biodiversity and ES of global, regional or local changes such as land use change, invasive alien species, over-exploitation, climate change and pollution. Scenario analysis also provides decision support for developing adaptive management strategies and exploring the implications of alternative social-ecological development pathways and policy options. At the same time, scenario analysis and scenario planning have been successfully applied in many local studies, in national assessment and for regional and global assessments (Chapter 5.7.3).

Generally, scenario analysis includes three major phases. The initial step is to define the major tendencies for a specific region or for a specific subject and to analyse the drivers of change that are likely to be involved in the foreseen tendencies. This phase results in a few plausible scenarios. A second phase is to translate these scenarios quantitatively or qualitatively into variables that describe the major drivers of change, such as economic development or demography. These drivers of change are then the input for models that relate these changes to environmental change, such as land use change or climate change, and on biodiversity and ES. A third phase starts with analysing the outcomes of these models and formulate policy options to avoid undesired developments in key variables of biodiversity and ES.

Models used in scenario analysis are typically able to describe dynamic relationships amongst drivers, biodiversity and ES. Often a wide range of models is needed to perform an adequate scenario analysis. Not only models that quantify changes of ES based on changes of land use are needed, but also models that drive these land use changes, such as economic and demographic models. In addition, hydrological and other biophysical models in combination with biodiversity interactions are required if more complex issues are under consideration.

New approaches for scenario analysis are proposed and applied, where stakeholders and local knowledge holders are increasingly involved. Another recent development in modelling for scenario analysis is to understand the feed-back loops from changing ES provision to a change in economic development.

Ecological modelling and particularly process-based ecological modelling, depend on a vast array of ecological, biophysical and anthropogenic datasets to generate relevant results. Although in recent years, earth observation systems have evolved to the point of delivering continuous (temporally and spatially) data for particular ecosystem components (e.g. forest change and extent, tree density, elevation, human density, economic characteristics, precipitation, etc.), many of these lack the ability to be compared or used in a modelling environment due to different resolutions and/or methods/sensors.

Additionally, there is a clear mismatch between the publication date of the variables to be used in a given assessment (e.g. LUCAS soil data from 2009) and the reference date for the assessment itself (for example using vegetation data from 2016 to assess the effect of soil erosion prevention without considering the 7 years' difference between these datasets). In several cases, if any modelling approach is to be implemented, these mismatches cannot be simply overcome and often error propagation assessments should be implemented to minimise unwanted effects.

Independently of the problems or potential caveats related to particular datasets, the temporal resolution (i.e. the amplitude and frequency of data collection) of a given dataset is an important determining factor for dataset selection in trend analysis. Therefore, future ecosystem service supply studies should include the effects of data quality on their results as it can produce important biases in the overall interpretation and decision-making support.

Bateman IJ, Harwood AR, Mace GM, et al. (2013) Bringing ecosystem services into economic decision-making: land use in the United Kingdom. Science (80) 341: 45-50.
Guerra C, Metzger MJ, Maes J, Pinto-Correia T (2016) Policy impacts on regulating ecosystem services: looking at the implications of 60 years of landscape change on soil erosion prevention in a Mediterranean silvo-pastoral system. Landscape Ecology. doi: 10.1007/s10980-015-0241-1.
Kandziora M, Burkhard B, Muller F (2013) Mapping provisioning ecosystem services at the local scale using data of varying spatial and temporal resolution. Ecosystem Services 4: 47-59. doi: 10.1016/j.ecoser.2013.04.001.
Koch EW, Barbier EB, Silliman BR et al. (2009) Non-linearity in ecosystem services: temporal and spatial variability in coastal protection. Frontiers in Ecology and Environment 7: 29-37. doi: 10.1890/080126.
Nelson E, Mendoza G, Regetz J et al. (2009) Modelling multiple ecosystem services, biodiversity conservation, commodity production and tradeoffs at landscape scales. Frontiers in Ecology and Environment 7: 4-11. doi: 10.1890/080023.
IPBES (2016) Methodological assessment of scenarios and models of biodiversity and ecosystem services, Ferrier S, Ninan KN, Leadley P, Alkemade R, Acosta-Michlik LA, Akcakaya HR, Brotons L, Cheung WWL, Christensen V, Allam Harhash K, Kabubo-Mariara J, Lundquist C, Obersteiner M, Pereira HM, Peterson G, Pichs-Madruga R, Ravindranath N, Rondinini C, Wintle BA (Eds.) Secretariat of the Intergovernmental Platform for Biodiversity and Ecosystem Services, Bonn, Germany.

Maps for ecosystem services (ES) are made for a broad set of purposes. These include advocacy (awareness raising, justification, decision support), ecosystem assessment, priority setting, instrument design, ecosystem accounting, economic liability and scientific spatial analysis. Figure 1 illustrates the theoretical relationship between mapping purposes and quality requirements. Requirements concern notably spatial and temporal resolution, scientific accuracy and reliability and ease of understanding. Additional methodological requirements not represented in Figure 1 are the extent of the mapping exercise, the repeatability, the theme of the mapping (e.g. supply, demand, conflict maps etc.) and basics of cartography and mapping semantics (see chapters 3.1 and 3.3). These vary depending on the specific context of the mapping exercise (e.g. community development versus national assessment, see Figure 1).

Figure 1.

Ecosystem services mapping requirements according to purpose.

Figure 1 can be interpreted across purposes for one specific requirement or across requirements for one specific purpose. For example, the expected clarity of a map meant for research use is lower than that aimed at policy advocacy. On the other hand, maps used by research should be highly reliable while those used for awareness raising (advocacy) do not require such high reliability.

Many current mapping applications focus on quantitative valuation and accounting. Typically, these maps are neither meant to be understood by a broad range of stakeholders nor do they necessarily require a high spatial resolution, but they should be highly accurate and reliable. This chapter illustrates this for two specific examples concerning regional assessment and priority setting.

Mapping quality requirements are bound by resource availability and by the risk of decisions based on them. The upper boundary of requirements is set by the principle of parsimony, stating that "among two good solutions, the simplest is always best". This highlights the need for using the least resources or assumptions necessary to solve a problem. In other words, one should not spend excessive (project) time and/or (public) money to map at a greater level of detail than necessary. For example, land use based maps (see Chapter 5.6), that can be produced repeatedly at relatively low costs (in terms of time and money) are sufficiently adequate for most purposes, while more reliable data can sometimes only be obtained at excessively high cost, or involving complex assumptions. Moreover, the time spent on a specific map should be traded off against the urgency of the purpose.

The lower limit of map quality requirements is determined by the societal impact of the decisions based on the mapping. Uncertainty (or absence of information on uncertainty) translates in a societal risk for adverse outcomes if decisions are based on wrong data. Public or policy advocacy for the importance of ES does not require highly accurate or detailed maps. However, communication maps cannot be used for purposes which have more stringent requirements, such as ecosystem accounting or economic liability: the risk for unfair or undesired outcomes is too high or unknown.

This brings us to the issue of the safe operating space for each type of map. Maps with lower requirements cannot be used for purposes which have higher requirements. On Figure 1, this goes both ways: for instance, maps made for scientific purposes need simplification to be clear enough for priority setting, assessment or advocacy, while assessment maps have to be detailed further to obtain the accuracy and reliability required for some scientific purposes.

Maps are instrumental tools that are combined with other types of data and contextual information in order to achieve a certain purpose (see Figure 1). This information can be quantitative and qualitative and is rarely spatially explicit. Knowing how maps will be combined with these non-spatial data and used in a specific context is essential for the mapping process. We illustrate this below by showing how maps are used as part of the diverse information for two common ecosystem service questions: a land use priority setting in a local context and a regional ecosystem assessment.

In this final section, we provide guidelines for critical map-makers to engage in effective ES mapping. While most of these will seem evident, they are rarely applied in practice. Following these guidelines will improve effectiveness of ecosystem service maps to impact actual decision-making and contribute to scientific advance.

• Clearly define the purpose for which mapping is needed. Plenty of maps are created without clear purpose and later applied for the wrong purpose.
• Determine the minimum reliability, accuracy, resolution and clarity required. The risk for undesired outcomes grows if maps are used for higher impact decisions.
• Assess the resources (time and money) needed to meet these requirements. Highly expensive, detailed or complex maps are not necessarily more effective.
• Delineate the safe operating space of your maps. The map-maker, being aware of the power and limitations of maps, bears responsibility to caution against wrong or risky application (see Chapter 6.4).
• Target the form and communication of maps fitted to the process they are used in. Maps are essential for many processes, but project purposes are never just maps.

Box 1. Local example priority setting for land consolidation to optimise ES provision

ES mapping at the local scale is often used to set priorities and guide decision-making to optimise ES provision. This example describes how ecosystem service maps were combined with biophysical models and valuation data to serve a participatory land-consolidation plan for three municipalities in Wallonia, Belgium. It is co-constructed by the administrations, scientists and local stakeholders. The project's objective is to design a replicable methodology, based on hands-on experience in a first case study. Figure 2 describes the methodological framework further.

After selecting a list of locally relevant ES and, based on a typology of ecosystems, biophysical assessment and social valuation are carried out. The biophysical assessment includes mapping and quantification of selected ES based on indicators obtained from a hydrological model and scenario development of potential ecosystem service supply. Social analysis comprises stakeholder analysis, societal valuation according to these stakeholders, participatory validation of the biophysically mapped ES and participatory mapping of ecosystem service demand. These supply and demand maps are then used to guide participatory comparison of land-consolidation actions. For instance, maps of biophysical indicators were compared with demand maps to highlight locations for which there is potential improvement of supply. Technical experts of land consolidation then suggest potential measures (e.g. installation of new hedgerows, creation of new water retention basins, new flower strips along a walkway etc.) to be implemented in the final land consolidation plan. This example clearly demonstrates that maps are used as a central means in combination with various other data, methods and actions, to achieve a broader objective shared by various stakeholders and lead to improved decision-making.

Figure 2.

Methodological framework for integrated valuation of ES to set priorities for land consolidation in Wallonia: Maps are central, necessary parts of a yet broader process (from Baptist et al. 2016).

Box 2. Regional example - regional ecosystem assessment

National and regional ecosystem service assessments seek to assess the state and trends of ES in their region, with the purpose of monitoring their evolution and informing policies. The state of ES comprises information on the demand, the supply, the balance between demand and supply, the use of ES, ecosystem functions underpinning them, drivers of change, impacts on human well-being and governance. Spatial data - also in regions with high data-density - are not available for all aspects of all services and for some aspects the spatial dimension is even irrelevant.

The Flanders regional assessment has assessed demand, supply, balance between these two and interactions between use of services. These statements were based on a detailed review of all data and information in 16 ecosystem service chapters to obtain one single concise table on the state of ES with known reliability. Despite the focus of the chapters on maps, the data underpinning this assessment are only partly spatially explicit and range over different data types which are synthesised in key findings (Figure 3). Although the separate maps can be used to answer specific questions, the context of a regional assessment requires synthesising maps into short conclusive statements or non-spatially explicit indicators for policy communication. Therefore, the statements derived from the 78 maps to inform the regional state assessment were verified and reviewed by all the involved map-makers.

In conclusion, maps which are integrated in communication, decisions or even research will be reduced to quantitative or qualitative findings and combined with other data and information to obtain final outcomes. Mapping will be more effective when engaging in the specific context, by targeting and communicating the maps to the specific purpose and by tuning maps to the diverse information they are combined with.

Figure 3.

Proportion of spatially explicit (distribution available on Flanders scale) data throughout the ecosystem service chapters (left panel) and per data type (right panel).

In many cases, maps are a starting point for an open discussion about what the maps need to indicate and about the assumptions made in the underlying models. Using maps top-down as 'objective data' often discards nuanced reality of a local context and is counterproductive in most real-life decision processes. To effectively apply maps, the ES map-maker needs to involve:

• Interdisciplinary engagement: learn from existing practices and cooperate with other research fields, such as environmental decision support, communication science, participatory processes, etc. to avoid classic pitfalls.
• Trans-disciplinary engagement: consider the use of co-design approaches from the very start. Nowadays, stakeholder involvement is an essential indicator for end-user satisfaction and final uptake of the developed maps and the only reality check the ES-map-maker has.

Ecosystem service mapping can be highly rewarding in terms of impact on real-world decision-making. This requires leaving the comfort zone of single disciplines and clear data layers and finding the right balance between scientific demands, user demands, functionality and available resources. for every mapping project again.

Baptist F, Degre A, Grizard S, Maebe L, Pipart N, Renglet J, Sohier C, Dufrene M, Dendoncker N (2016) Elaboration d'une methodologie d'evaluation des incidences sur l'environnement de l'amenagement foncier s'appuyant sur la notion des services ecosystemiques. Rapport general. Direction Generale Operationnelle de l'Agriculture, des Ressources Naturelles et de l'Environnement, 187 pp.
Gomez-Baggethun E, Barton DN (2013) Classifying and valuing ecosystem services for urban planning. Ecological Economics 86: 235-245.
Guerry AD, Polasky S, Lubchenco J, Chaplin-Kramer R, Daily GC, Griffin R, Ruckelshaus M, Bateman IJ, Duraiappah A, Elmqvist T, Feldman MW, Folke C, Hoekstra J, Kareiva PM, Keeler BL, Li S, McKenzie E, Ouyang Z, Reyers B, Ricketts TH, Rockstrom J, Tallis H, Vira B (2015) Natural capital and ecosystem services informing decisions: From promise to practice. Proceedings of the National Academy of Sciences 112(24): 7348-7355.
Hauck J, Gorg C, Varjopuro R, Ratamaki O, Maes J, Wittmer H, Jax K (2013) Maps have an air of authority: Potential benefits and challenges of ecosystem service maps at different levels of decision making. Ecosystem Services 4, 25-32. doi: 10.1016/j.ecoser.2012.11.003.
Jacobs S, Dendoncker N, Keune H (Eds.) (2013) Ecosystem Services: Global Issues, Local Practices, Elsevier, New York.
Jacobs S, Spanhove T, De Smet L, Van Daele T, Van Reeth W, Van Gossum P, Stevens M, Schneiders A, Panis J, Demolder H, Michels H, Thoonen M, Simoens I, Peymen J (2016) The ecosystem service assessment challenge: Reflections from Flanders-REA. Ecological Indicators 61: 715-727. doi: 10.1016/j.ecolind.2015.10.023.
McIntosh BS, Ascough JC, Twery M, Chewe J, Elmahdi A, Haase D, Harou JJ, Hepting D, Cuddy S, Jakeman AJ, Chen S, Kassahun A, Lautenbach S, Matthews K, Merritt W, Quinnm NWT, Rodriguez-Roda I, Sieber S, Stavenga M, Sulis A, Ticehurst J, Volk M, Wrobel M, van Delden H, El-Sawah S, Rizzoli A, Voinov A (2011) Environmental decision support systems (EDSS) development - Challenges and best practices. Environmental Modelling & Software 26: 1389-1402.
Pavlovskaya M (2006) Theorising with GIS: a tool for critical geographies? Environmental Planning A 38: 2003-2020. doi: 10.1068/a37326.

Joachim Maes, Chiara Polce, Grazia Zulian, Ine Vandecasteele, Carolina Perpina, Ines Mari Rivero, Carlos Guerra, Sara Vallecillo, Pilar Vizcaino & Roland Hiederer

Climate regulation

Ecosystems regulate our climate at various levels. In and around cities, urban forests provide shade during hot summer days and by evaporating water through their leaves, they cool down cities, thus delivering benefits in terms of saved energy costs or lowered ozone production and concentration. On larger spatial scales, forests, wetlands, coastal systems and other ecosystems maintain comfortable atmospheric conditions and regulate climate. Yet, mapping ES which contribute to the regulation of climate, is often narrowed down to mapping carbon storage and carbon sequestration. Climate change science and policy is evidently the reason for this focus. Net primary production is at the basis of this and many other ES and therefore often mapped. Much useful information to map primary production is available through remote sensing, field observations and modelling.

Given the increase in atmospheric carbon and the consequences for climate, terrestrial carbon pools are an important factor in the carbon balance. The terrestrial organic carbon pool (soil and vegetation) is estimated to be 3500 Pg C, most of which (75%) is stored in soil. This is almost fivefold the amount of carbon in the atmosphere. The carbon stored in the soil mainly originates from dead organic material. The main governing factors for the status of the soil organic carbon (SOC) pool are land use/land cover and local climatic conditions. Changes in land use and management practices can lead to imbalances in the flux between carbon pools. Depending on environmental conditions, the SOC pool can act as either a source of atmospheric carbon or a sink, i.e. removing carbon from the atmosphere.

Box 1. Mapping soil protection in Europe

An assessment of soil protection in Europe in 2010: (a) Soil retention at European scale, b) Soil retention in Central Portugal, c) Capacity to avoid soil erosion in Central Portugal and d) Structural impact in Central Portugal. Soil retention (Es) was calculated as soil loss without vegetation cover (structural impact, Y) minus soil loss including the current land use/cover pattern (the mitigated impact), measured in tonns ha^-1 year^-1.

The structural impact is the total soil erosion impact when no ecosystem service is provided. The capacity of a given land cover type to provide soil protection (e) is expressed using values ranging from 0 to 1 for every mapped grid cell. To estimate the capacity, the vegetation per land cover type was computed using the Normalised Difference Vegetation Index (NDVI), the environmental zones and the snow cover. The highest soil retention values corresponded to areas covered by forest, transitional woodland and shrubs (semi-natural vegetation areas) and pastures.

Soil retention is also a function of structural impact (high potential erosion). Expressed differently, soil retention only occurs where soils run the risk of being eroded. In these places, vegetation cover protects the soil against water flows (surface runoff), reduces the structural impact and, therefore, effectively delivers a service.

A close up is presented for the central part of Portugal (Alentejo and Centro Regions). In the Tagus river valley, soil retention is low (light orange areas) due to a low structural impact and the dominant land use type, mainly agriculture. High soil retention (high provision of the service, in dark blue) results from the combination of high structural impact and high capacity to avoid soil erosion, for instance in forested areas. In contrast, if the inherent structural impact is low, the provision of the service (soil protection) is low as well, thus lowering the role of vegetation in soil protection.

Figure 1.

Mapping changes to the SOC pool can be based on the methodology of the International Panel on Climate Change (IPCC). The method uses type of climate, soil, category of land use, management and input practices as factors influencing SOC stocks. For each factor, the relative effect of changes to the SOC pool is provided for different climate/soil regions. When all factors remain unchanged, an equilibrium in the SOC pool is assumed to be reached after 20 years.

Given the factors influencing SOC content, the spatial distribution of SOC stocks is very variable (Figure 2). Most of the global SOC is stored in the northern hemisphere where cool and moist conditions favour plant decomposition into soil organic matter. However, under wet conditions and high productivity of vegetation, organic material may also accumulate in tropical regions, such as in peat lands of south-east Asia. In tropical forests, the amount of carbon stored in the above-ground vegetation exceeds the carbon stored in the soil with the exception of peat lands.

Figure 2.

Spatial distribution of global soil organic carbon density (t C ha-1). Source: FAO and ITPS.

Water regulation

Forests, grasslands and wetlands are ecosystems with a high capacity for regulating the flow of water. This is particularly important for ensuring the supply of a sufficient quantity of water to support the immediate environment whilst avoiding extreme fluctuations in water flows. Where water is not properly regulated by the ecosystem (e.g., in cities, where the natural water cycle is often interrupted by impermeable surfaces), there is a much higher risk of such fluctuations, potentially leading to flooding or water shortages. The provision of water regulation can be mapped by breaking down the process into its various components. Ideally, the landscape should naturally retain and store an adequate amount of water for its needs, whilst limiting the amount of surface runoff - an excess of which may cause flooding further downstream. Water flow through a landscape may be influenced by the following natural processes, all of which contribute to the storage of water and therefore the reduction of surface runoff: interception by vegetation, storage in surface water bodies, infiltration and retention in soil and percolation to groundwater stores.

In addition to these processes, the amount of water which can be retained will also be affected by the slope of the landscape and by the degree of permeability of the soil. Steeper slopes will promote faster surface runoff, whilst flatter areas allow greater time for infiltration of water. Impermeable surfaces (e.g. artificial infrastructure such as roads and buildings) represent a barrier to the infiltration and retention of water, thus promoting surface runoff.

Figure 3 gives an overview of the parameters taken into account to map the water retention as a proxy for the water regulation capacity of the ecosystem.

Figure 3.

Schematic overview of the structure of the indicator for mapping water retention. Parameters in grey are dynamic and thus change over time.

The retention of water in vegetation, surface water bodies, soil and bedrock (groundwater stores) are considered landscape storage factors. Additionally, the influence of slope and surface imperviousness are considered as physical factors altering the actual water retention capacity of the landscape. The contribution of each process to the final indicator is approximated using one or more parameters or characteristics of the landscape. The parameters shaded in grey are those which are changeable over time. The various factors are combined to give the final composite indicator representing relative landscape water retention or, rather, the capacity of the ecosystem to provide water regulation as a service.

Pest control

Agricultural ecosystems are often harmed by pests such as insects (i.e. caterpillars) and small mammals (i.e. moles), significantly reducing the harvested share of crop production. However, nature offers natural fighters against these pests, thus saving farmers billions of dollars annually by protecting crops and reducing the need for chemical control. There are different groups of natural enemies known to play a key role in pest control, such as birds, mammals, spiders, lady bugs and other types of organisms. So, mapping pest control clearly relies on spatial information on the distribution of predator species (species distribution models, see also section on pollination).

We show below an example of mapping potential pest control by birds in agricultural systems (Figure 4). The example is based on species distribution models of 49 bird species, recognised as pest-control providers. Modelled species include the Little Owl (Athene noctua), a known hunter of mice, voles, shrews, moles and rabbits and the Hoopoe (Upupa epops) which has an insect-rich diet.

Figure 4.

Spatial distribution of predatory bird species richness in the European Union. The close-up around Paris shows that species richness is lower near urban areas (mapped in red).

Species distribution models map the probability of species occurrence based on field observations. A probability threshold can be defined for instance at 50% to assume the presence of a certain species. By overlaying all the species occurrence maps of the 49 modelled species, a map of potential pest control by predatory birds is obtained. Higher species richness corresponds to a more diverse community of natural predators and is assumed to exert a greater control on pest populations. Figure 4 shows potential pest control by bird species across Europe. The inset is a close-up around Paris, showing spatial differences in bird species richness, with low values (areas in yellow) in and around the urban areas (in red).

Air quality regulation

Air pollution is one of the main environmental risks for human health and is the main cause of premature deaths. In this context, abatement of pollution has become of major concern especially in areas with high pollutant concentrations, typically urban areas. Maintaining and developing green urban areas can be part of an integrative strategy to help increase air quality in European cities. Trees reduce temperatures in cities by evaporating water and they remove air pollutants and particulate matter via their leaves through dry deposition. Urban trees, green areas and forests surrounding cities have the capacity to remove significant amounts of pollutants thereby increasing environmental quality and human health.

Mapping air quality regulation is based on three types of information: the dry deposition velocity (supply), the removal of air pollutants (flow) and human exposure (demand).

The pollutant dry deposition velocity by vegetation is considered often as a proxy to assess the ecosystems capacity to remove pollutants from the atmosphere. This quantity measures the rate at which pollutants are collected from the atmosphere by tree leaves. The contribution of vegetation is often mapped and modelled using spatially explicit data of the leaf area index (LAI). The LAI is defined as the one-sided green leaf area per unit ground surface area. The larger this area, the more pollutants are captured by trees.

Furthermore, the pollutant removal flux by vegetation, which is estimated as the product of pollutant dry deposition velocity by vegetation and pollutant concentration, is usually considered as a good measure for the ecosystem service flow.

Finally, demand for the service can be mapped using population exposure to pollutant concentrations beyond the limit established within the legislation currently in force.

Maps of the atmospheric concentration of pollutants are essential inputs to map air quality regulation as an ecosystem service. Mostly, they rely on a network of monitoring stations where different pollutants are measured. The measurements collected by different monitoring stations can then be interpolated to obtain maps of concentrations. Several GIS techniques exist to perform interpolation by, for example, kriging and spatial regressions.

Figure 5 presents an example for the Barcelona metropolitan region. In this case, concentrations of NO₂ were estimated using Land Use Regression (LUR) models. The LUR model was built using NO₂ concentration measurements for the year 2013 from the operational monitoring stations as dependent variables and a set of spatial predictor parameters (independent variables) that were considered to be the most relevant for distribution of NO₂ concentrations, related to land cover type, geomorphology, climate and population. The map of unsatisfied demand for air quality regulation was generated from the population living in areas where annual mean concentrations exceed the EU limit value (40 µg/m³ for NO₂).

Figure 5

Indicators for the assessment of air quality regulation of NO₂ in the Barcelona Metropolitan Region: concentration, deposition velocity, removal capacity and population affected.

Benjamin Burkhard & Marion Kruse

Introduction

Material and energy outputs from ecosystems are usually classified as provisioning services. These are tangible goods or services that are directly used, traded or exchanged by all human beings. They can be grouped into nutrition (e.g. cultivated crops, seafood from aquaculture, wild food), materials (e.g. fibres and genetic materials) and energy services (e.g. fuel wood). Some of them, such as cultivated crops and animal outputs, are amongst the most mapped ecosystem services (ES), whereas others, such as genetic materials and energy provided by animals, have been studied or mapped less frequently to date (Figure 1).

Figure 1.

Schematic overview of least and most mapped provisioning services.

Provisioning services are often produced and consumed or used in different places. They are generally transported from the place of production (i.e. supply) to the place of consumption (i.e. demand). It is more common and easier to map the supply, as it is spatially explicit and directly depends on the ecosystem's structure and functioning, whereas the demand is a function of socio-economic drivers.

The economically important crop and animal products, as well as timber and fish products, can be closely associated with agriculture, forestry and fishery/aquaculture and consequently, their related land cover/use types. As they represent traditional economic activities and research focuses that have existed for a very long time, a large body of subject-specific knowledge and data sets are available to quantify ES supply based on these economic sectors. This fact also gives the opportunity to analyse changes and trends of these ES in many regions. These production systems are usually monocultures and require a large amount of human input (Chapter 5.1). On the contrary, wild plants, water and genetic resources are less clearly associated with one specific land cover/use class and are generated in more diverse and semi-natural or natural ecosystems and landscapes.

The production of ES is not only location-specific, but it is also dynamic over time (Chapter 5.3). Examples include crops mainly being grown and harvested in spring-summer period in the temperate zones. In the tropical zone, the growing season lasts all year around. Another example is the dynamic supply of drinking water from mountain regions, which follows the seasonal changes in hydrological and climatological conditions.

Mapping methods for provisioning services

Based on the CICES classification (Chapter 2.4), provisioning services can be grouped into the classes reported below. Different methods and data sets are available for mapping these classes. A short overview of selected mapping methods and data sets is provided in the following sections.

Cultivated crops & reared animals and their outputs (e.g. cereals, vegetables, meat, milk)

These provisioning services are mainly commercially valued and traded as the direct output of agriculture from arable land and pastures. They are amongst the best monitored ES and their level of production is documented in agricultural statistics or accounting in many areas. Therefore, they can be easily mapped using land cover/land use maps in combination with indicators of crop or animal production (e.g. t/ha/year crop yield, number of animals/ha, l/ha/year milk production) from national or other statistics. This corresponds to a tier 2 mapping approach (see Chapter 5.6.1). This method has minimal data requirements and is therefore easy and quick when the corresponding data are at hand. With such data sets, maps for these provisioning services can be generated for local up to global scales. Use of a single indicator, however, neglects the effects of the management regime and the environmental characteristics of the agricultural ecosystem (e.g. soil texture, climatic and hydrological conditions), all of which influence the level and quality of ES generated. It is also possible to include anthropogenic system inputs and environmental effects as indicators instead of only crop yield or animal numbers. Due to the commercial character, there is a large additional input (e.g. fertilisers, pesticides) in most agro-ecosystems. Furthermore, it is important to notice that in the case of reared animals, land cover/land use does not automatically correspond to the area of supply. Livestock is often kept in buildings, resulting in point accumulation within the respective map.

There are also crop or animal production models (e.g. Common Agricultural Policy Regionalised Impact (CAPRI) model or Agricultural Production Planning and Allocation (APPA) model) accounting for the ecosystem's capacity, environmental effects and human inputs to obtain more accurate results. Nevertheless, these models are time and data intensive. They are suitable if the aim is to better understand a certain production system or create a crop and animal production map for a certain location under a specific socio-economic scenario or environmental constraint.

For general purposes and the mapping of multiple provisioning services, look-up tables are in common use (Chapter 5.6.4). For some outputs from reared animals (meat, milk), only aggregated or average data exist (slaughtering for a defined reference date).

On the local scale (e.g. farm), detailed analyses can be included in maps, such as variations over a season. Considering the growing season of cultivated crops, the supply does not always match the continuous demand. Over the entire growing season, up until the moment of harvest, crops can be considered as only potential provisioning services. The real use (flow) is connected to harvest, processing and consumption.

Wild plants and wild animals and their outputs (e.g. wild berries, mushrooms, water cress, game, fish, honey from wild bees)

This class includes both commercial and subsistence berry and mushroom collection, fishing and hunting for food. These less marketed ES can provide subsistence especially in less developed countries, while they are considered by some stakeholders and researchers as hobby or recreational activities in other regions. Only few examples exist for the mapping of these provisioning services because of their individual character. Statistical data are available for hunting and recreational fishing in some regions or countries. Data and information are often available for commercial fishing as regulations (e.g. exclusive economic zone and catch quota) have to be respected in many regions. However, mapping the exact area of where fish and seafood are extracted includes uncertainties due to the mobility of most species. Mapping fishing grounds requires GPS data or interview data. Usually, statistical data are grouped into specific areas (e.g. Baltic Sea or North Sea) or on an administrational level (e.g. states and countries) making the mapping less spatially explicit. For some individual or subsistence activities, such as berry/mushroom picking, recreational fishing or hunting, licences are needed. This can be used as a proxy to quantify the amount of potential users. Most important though, is the exact area of supply and the respective amount of provisioning service. This requires laborious and possibly expensive, field studies and interviews. Data from random sampling is often used for extrapolation. Methodological studies need to reveal which natural and socio-economic settings, extrapolation or value-transfer are accurate enough for reliable results.

Mapping of berry and mushroom collection and game hunting is possible by combining land cover/land use and species habitat maps with other biophysical layers (e.g. management intensity, climatic factors) and with accessibility or travel time from settlement areas. Another approach is participatory mapping (Chapter 5.6.2) of indicators such as kg/ha berries collected or the location of honey collection, relying on the knowledge and wild food collection habits of local inhabitants and stakeholders. Wild food collection is also closely related to cultural services (such as cultural diversity and traditions) and is affected by the human-environmental relationship and societal conditions (e.g. laws, regulations, property rights). This information can be included in the mapping processes through, for example, overlaying protected areas or area with restricted access. Figure 2 presents an example for medicinal plants.

Figure 2.

Participatory mapping of medicinal plants in the Bereg region, Hungary: Local stakeholders were questioned if, where and which medicinal plants grow and if they are collected. The growth and collection of medicinal plants were related to different land cover types. Although, the study's objective was an assessment of ES, the results could be translated into a map showing the location of this provisioning service in a further step (Source: Petz et al. 2012).

Few studies are available on regional or national scales. Mapping of these services (except commercial fishing) is most suitable for the local or regional scale, making it possible to include high resolution data and information needed for sustainable management.

Fibres and other materials from plants and animals (e.g. timber, cotton, grass as fodder)

This class includes both consumptive and ornamental uses and both commercial (e.g. industrial timber production) and subsistence (e.g. local wood collection) uses. Timber, grass and fodder production have been widely mapped, whereas other materials, such as cotton and silk are rarely mapped. Mapping methods range from the use of a single indicator (e.g. m³/ha wood harvested, kg/ha grass collected) to complicated forest or vegetation production models e.g., the European Forest Information SCENario Model EFISCEN (see Figure 3 for an application), the Global Forest Model (G4M). Subsistence use can be mapped using participatory techniques as well, especially at local scales. An application example for mapping

Figure 3.

Mapping timber harvest (m?/ha/yr) in 2050 under the VOLANTE A2 business-as-usual scenario modelled with the EFISCEN (European Forest Information SCENario) model on the European scale. Yellow indicates no harvest and grey indicates non-forest areas on the map (Source: Schelhaas & Hengeveld pers. comm).

local fuel wood supply is shown in South Africa in Figure 4. For timber, statistical data are available. However, separation from fuel wood is difficult as sometimes several products are manufactured from the same source (e.g. timber, woodchips). In contrast to annual or seasonal supply from some fibres and fodder (e.g. cotton, hay), wood products are usually only harvested over longer time periods (> 40 years) due to growing phases in the temperate zones. Fast-growing species are harvested in the (sub-)tropical regions at shorter intervals.

Figure 4.

Mapping fuel wood supply in the Baviaanskloof Catchment in South Africa using local data consultations. Combining multiple indicators (left) results in a fuel wood yield map (right). Several data sets were combined to show the spatially diverse supply of fuel wood (including topography, accessibility and also conservation areas), (Source: Petz et al. 2014).

Similarly with cultivated crops and reared animals, there is a mismatch between the supply and demand for (some of) these services, as timber and wood products are marketed across the globe. Some fibres (e.g. cotton) belong to the most important provisioning services and are heavily traded globally. Very few studies in the context of the ecosystem service concept exist, although some (global) statistics on production are available. Recycling and multiple uses or purposes of materials result in possible uncertainty of these assessments.

The provisioning services of cultivated crops, reared animals and their outputs and materials from plants and animals are often produced on the same farm. In this case, double-counting is possible. Nowadays, fodder is however imported to many intensive farming regions, making the assessment of provisioning services more difficult or uncertain. Local mismatches of supply and demand result from this, which make the mapping of these services incomplete.

Genetic materials from wild plants and animals (e.g. medicines and wild species used in breeding programmes)

Genetic material from wild plants can be used for biochemical industrial and pharmaceutical processes (e.g. medicines, fermentation), as well as bio-prospecting activities (e.g. wild species used in breeding programmes). Genetic materials have been mapped infrequently. On a similar basis, wild food and medicinal plants have also a close link with cultural traditions and societal conditions. The occurrence or supply of medicinal plants could be mapped similarly to wild food by combining species richness and land cover/land use data or applying participatory mapping studies. Biodiversity models could provide useful information about the occurrence of different wild species. Suitable habitats for and spatial dynamics of mobile species, such as insects or mammals, can be explored with agent-based models (Chapter 4.4).

These provisioning services have been rarely mapped, although there is ongoing research (considering different species and ecosystems ranging from the tropical rainforest to marine environments). Usually, this covers only limited areas.

Animals and plants from aquaculture

Mapping provisioning services from aquatic ecosystems is usually more difficult. Information on water bodies is often not as detailed from land cover/land use maps as for terrestrial ecosystems (see Chapter 7.4). More detailed information about protected areas, different habitats, or spawning areas is needed to map animals and plants from aquatic ecosystems. An application example for mapping fishery areas in the Baltic Sea is shown in Figure 5.

Figure 5.

Mapping fishery areas in coastal ecosystems of the German Baltic Sea. The example shows that a land cover based approach results in an over-estimation of supply area. Here additional information is included to depict spatially explicit areas of restriction due to protection within a National Park. Boat traffic is partially prohibited. Temporal fishing restrictions exist in the spawning areas. Recreational fishing (at land) is only allowed in the designated spots and areas where boat traffic is allowed.

Wild caught fish in marine and freshwater ecosystems is an important food resource globally. Due to declining stocks and regulations (e.g. EU fisheries regulations) aquaculture is employed more and more to meet the demand for seafood and algae.

These data are available on different spatial scales and, in most cases, over a very long time period as they are important for the economy. Here also single indicators (e.g. fishing statistics in t/year) are available. Infrastructure from aquaculture, such as cages, basins, ropes, is visible in the field and can be used to identify the extent of the provisioning area.

Water for drinking and non-drinking purposes

Water extraction is usually undertaken in single spots where the conditions are suitable (i.e. infrastructure and water quantity, quality and intensity). Groundwater is recharged over a larger area and depends on ecosystem conditions, such as substrate and vegetation cover. Surface water is used in many regions where ground water extraction conditions are not suitable.

Maps can show groundwater yield and the amount of water (m³) that can be extracted without declining the yield. Hydrological models can be applied to simulate the effects of changes in consumption and hydrological and climatic conditions (Chapter 4.4).

Some statistical data are available for average water consumption (drinking water, non-drinking water) and the water incorporated or locked in products (e.g. food, clothes). Large regional mismatches occur on the global scale due to trading of products. The price for water could be used as an indicator for mapping as well.

Temporal changes in water demand is an important aspect in management, especially in areas with high usage (e.g. from tourism). In some cases, water needs to be pumped from other areas to fulfil the demand. These regional and temporal mismatches make mapping of water as an ES challenging.

Plant and animal-based energy resources (e.g. fuel wood, crops and labour provided by animals)

It is more common and easier to map plant-based energy resources than animal-based ones. The mapping of energy crops, such as oilseed rape, is similar to food crops. Sometimes they are also competing in cultivated areas.

The supply of fuel wood can be mapped with a single indicator (e.g. forest areas, occurrence of certain tree species), forest production models (e.g. EFISCEN, G4M) for commercial use and participatory approaches for subsistence use. Large regional differences exist. In some regions, fuel wood is the only energy source for cooking and heating whereas, in other regions, it is only a supplementary source or even unnecessary (e.g. urban areas with good energy infrastructure).

Labour provided by animals as an ES has not been mapped yet. It could be mapped, for example, using statistics involving the quantity of animals. In some areas, labour provided by animals is important in agriculture, but also for transportation. However, due to mechanisation in many sectors, this provisioning service is of less and less importance.

Leena Kopperoinen, Sandra Luque, Patrizia Tenerelli, Grazia Zulian & Arto Viinikka

Introduction

Cultural ecosystem services (CES) bind elements between social and ecological concepts. They are seen as nature's intangible benefits related to human perceptions, attitudes and beliefs. People obtain spiritual enrichment, cognitive development, reflection, recreation and aesthetic experiences from ecosystems (Table 1). People's perceptions can differ significantly, not only person by person, but also from one area and culture to another. Therefore, CES are not readily transferrable from one place to other environments.

Table 1.Download as CSV 

CES according to Common International Classification of Ecosystem Services v. 4.3.

Division	Group	Class	Examples
Physical and intellectual interactions with biota, ecosystems, and land-/seascapes [environmental settings]	Physical and experiential interactions	Experiential use of plants and animals. Experiential and physical use of land- / seascapes in different environmental settings	In-situ whale and bird watching, snorkelling, diving and other experiential enjoyment of nature. Walking, hiking, climbing, boating, leisure fishing (angling), leisure hunting and other physical activities in nature.
	Intellectual and representative interactions	Scientific	Subject matter for research both on location and via other media.
		Educational	Subject matter of education both on location and via other media. Nature as a location for education.
		Cultural heritage in connection to nature	Historic records, cultural heritage e.g. preserved in water bodies and soils, interplay of nature and culture, traditional uses of nature, cultural identity.
		Entertainment	Ex-situ viewing / experience of natural world through different media, such as photographs, films, literature.
		Aesthetics	Beauty of nature and land- / seascapes, artistic representations of nature.
Spiritual, symbolic and other interactions with biota, ecosystems, and land-/seascapes [environmental settings]	Spiritual and/or emblematic	Symbolic	Emblematic plants and animals e.g. national symbols such as American eagle or Welsh daffodil, sense of place, place identity.
		Sacred and / or religious	Spiritual values, ritual identity e.g. 'dream paths' of native Australians, holy places, sacred plants and animals and their parts.
	Other cultural outputs	Existence	Enjoyment provided by the pure existence of wild species, wilderness, ecosystems and land- / seascapes.
		Bequest	Willingness to preserve plants, animals, ecosystems, land- / seascapes for the experience and use of future generations; moral / ethical perspective or belief.

CES have both use and non-use values including existence, bequest, option and intrinsic values. Relational values referring to cultural identity and well-being derived from people's relationships with both other people and nature and mediated by particular places are also typical of CES. The focus of CES can be on individual needs and values or those fulfilled and possessed at a collective level. At both levels, CES concretely contribute to human well-being, public health and psychological experiences. As a result, CES are greatly appreciated by people and, in many instances, they are even better acknowledged than other ES. In more traditional communities, CES are often essential for cultural identity, livelihoods and even survival. The problem is, however, that many CES are difficult to quantify or their value too complex to assess and map. That has led to an over-emphasis on recreation and ecotourism which are empirically and conceptually easier to identify and measure while, at the same time, neglecting other important CES that matter to people but which are not as easy to measure (e.g. spiritual services).

This chapter aims to present how to survive the challenge of mapping non-material services, what examples of methods exist to map the potential provision of CES at different spatial levels, how to involve stakeholders in the mapping activity and what are the options that social media provide for CES mapping. Many methods useful for mapping CES are also presented elsewhere in this book.

What is specific about mapping cultural ES

As CES are considered non-material benefits, their quantification can be rather challenging: how to get hold of values linked to human perceptions compared to, for example, provisioning services where the actual stock of material can be quantified using different units of measure? Rapid quantitative mapping might not be easy for complex CES but it is possible to map them by combining knowledge and (also qualitative) methods from different disciplines, including not only natural and environmental sciences but also psychology, anthropology and other social sciences.

In order to map CES, methods to capture cultural norms and to express plurality of values in a spatially-explicit way are needed. Some researchers consider CES and their value measurable since they are expressed in human actions. Values ascribed to CES can be identified, for example, using the presence of certain products of an area, visible manifestations of CES in the physical landscape, or the number of studies, artistic representations etc. about an ecosystem as proxies. Spatial datasets giving location to certain socially or culturally normative values of the environment (e.g. inventoried cultural heritage or valuable landscapes; green areas of sufficient size and location) can also be used as indicators of areas providing CES.

However, if a more detailed and precise picture of CES is to be gained in a specific area, local people must be involved in mapping. Thus, mapping CES is inherently participatory if it is to be done properly.

As a result, mapping CES capacity and demand are interwoven. What is considered as potential capacity of an area is dependent on what brings people well-being and what people perceive they need and value in terms of CES. This needs to be understood first. When there is knowledge of this, different datasets can be used to identify where valued environments, features, species, or opportunities for specific experiences having the capacity to provide CES are located (Table 2). Mapping actual demand for CES is frequently done by using participatory mapping methods or indirectly utilising contents of social media. Participatory mapping means involving stakeholders, locals, etc. to identify, assess or otherwise value and point out on a map, areas or spots where they enjoy or feel CES (see more about participatory mapping in Chapter 5.6.2). Social media based methods include, for example, asking people to take photos of perceived CES in an area, involving people in scoring photos of different landscape types or, for example, analysing geo-tagged photos uploaded on social media. The latter is an indirect method to reveal people's preferences and locate their activities. Other geo-referenced contents of social media can also be analysed and used for the same purpose. CES capacity and demand, as well as the flow, can also be mapped using deliberative mapping methods where a group of people discuss, compile knowledge and finally build a consensus on these in a certain area on a map.

Table 2.Download as CSV 

A non-exhaustive list of methods suitable for mapping CES. Level of needed expertise refers to the degree of needed skills in GIS and / or statistics.

Method type	Method name	For mapping: Capacity = C, Flow = F, Demand = D	Which CES can be mapped with the method	Level of needed expertise: Low = L, Medium = M, High = H	Characteristics of the method in regard to CES mapping
Models, mapping methods	ESTIMAP	C, D	Potentially all CES	M	The GIS processes are relatively easy to implement, requiring only a medium level of GIS expertise. The model allows simulation of different scenarios and evaluation of different policy options; it is flexible and can be downscaled and modified in order to fit local needs and conditions. Expert opinion is needed for inputs variables selections and scoring. Scientific evidence for the used thresholds is scarce and they thus mainly rely on expert opinion, too.
	InVEST - recreation module	C	Recreation, nature tourism	H	Predicts the spread of person-days of recreation and tourism, based on the locations of natural habitats, accessibility and built features that factor into people's decisions about areas for recreation. Regression mapping that uses photos as a dependent variable. http://data.naturalcapitalproject.org/nightly-build/invest-users-guide/html/recreation.html
	MIMES		Recreation, nature tourism	H	Suits ideally the examination of trade-offs under various economic, policy and climate scenarios in space and over time. Allows for testing management scenarios that would be socially unacceptable.
	ARIES / k.LAB	C, D	Aesthetics, potentially all CES	H	Is based on probabilistic modelling using Bayesian framework. Requires expert-level modelling skills. http://www.integratedmodelling.org/
	GreenFrame	C, D	All CES	M	The GIS processes are relatively easy to implement, requiring only a medium level of GIS expertise. Makes use of a multitude of GIS datasets combined with both scientific and local expert scorings. Data on harmful phenomena that diminish the CES potential can be included in the analysis. Both quantitative and qualitative data can be used. Gives an overall picture of the relative spatial variation of CES provision potential.
	Land cover / land use based mapping	C, D	Recreation, aesthetics, education	L	Gives only a very rough proxy with high uncertainty level. Suits best for quick mapping of specific recreational or experiential activities.
Social media based mapping	Photo-series analysis	C, (F), D	Physical and intellectual interactions with biota, ecosystems, and land- / seascapes (including recreation, nature tourism, landscape aesthetics, cultural heritage and education)	M-H	It represents a cost-effective way of gathering space-and time-referenced data on observed people's preferences. It does not directly allow for obtaining information related to the user characteristics (socio- and psycho-cultural). Inherent bias is related with the interpretation of pictures. The photo-sharing community may not be representative of all social groups (the represented population will be dependent on the level of access to information technology, education and age and the user's ability / willingness to correctly geo-tag the photos).
	Other analyses of social media content	C, (F), D	Recreation, nature tourism, cultural heritage, potentially all CES	M-H	Specialised social media communities can produce data on, for example, sites suitable for specific activities but which are not commonly known and do not exist in databases. Communities may not be representative of all social groups (the represented population will be dependent on the level of access to information technology, education and age, and the user's ability / willingness to correctly geo-tag locations.
Participatory mapping (On-site and off-site mapping)	On-site map surveys using paper maps	F, D	All CES	L	Easy to implement anywhere and anytime. Time-consuming and laborious. Only restricted amount of information can be collected unless plenty of workforce is available. Collected information may be better in quality as any problems in mapping can be solved immediately. Good social skills are needed.
	On-site map surveys using electronic device	F, D	All CES	L	Easy to implement anywhere and anytime. Time-consuming and laborious. Only restricted amount of information can be collected unless plenty of workforce is available. Collected information may be better in quality as any problems in mapping can be solved immediately. Good social skills are needed. Malfunction of electronic device can happen any time.
	Interviews for the elicitation of values	F, D	All CES	L	Laborious and time-consuming and thus a limited number of people can be reached. Gained knowledge is much more detailed and much deeper understanding of the local CES can be derived in addition to maps.
	On-line map surveys	F, D	All CES	M	Several companies providing opportunity to implement on-line map surveys exist (paid service). Service includes usually basic reporting tools. Planning a workable survey can be demanding. All population groups can be difficult to achieve (access to computer, skills of using it), digitising is not always easy for laypeople for several reasons (ability to locate places on maps, etc.). Usually low response rate. With simple point mapping, lots of data can be derived. Background information of the respondent and additional information can be collected together with the map markings. Used for spatial planning purposes to gather knowledge and feedback.
	Deliberative mapping in a group on paper maps or using device, e.g. computer, visual table or landscape theatre	C, F, D	All CES	L-M	Demands good facilitation skills as the data on CES is mapped in a face-to-face setting and can involve participants of varying map reading skills and with opposing views. Sensitive to malfunction of electronic device if those are used.
	Mobile phone applications	F, D	All CES	L	Suitable for mapping CES related activities, values and perceptions of a target group at local scale. Works also for environmental awareness-raising simultaneously with mapping.
Landscape analysis	Viewshed analysis	C. F, D	Landscape aesthetics	H	Combines social media and physical landscape analysis. It represents a cost-effective way of gathering space-and time-referenced data on observed people's preferences. The viewshed is an approximation of the real visible surface. Quality of assessments depends on the resolution of the digital elevation model. Analysis includes computational complexity.
	GIScame	C	Landscape aesthetics	M	Landscape aesthetics; aesthetical aspects can be characterised by analysing landscape structure or the distribution of land use types with the help of landscape metrics. http://www.giscame.com/giscame/english_home.html

In a tiered approach for mapping, Tier 1 does not easily fit for the spatial representation of CES (see Chapter 5.6.1 for explanation of different tiers). For some CES, for example 'ecosystems as sites for activities', land cover can be used as a proxy for the landscape's suitability for different use types from the perspective of potential capacity. The demand side is easier to map in a Tier 1 approach as people can be asked to score or value different land cover types with regard to their appreciation in terms of CES (for example, 'forests bring me feelings of sanctity' or 'meadows are aesthetically important for me'). Still, the outcome remains quite vague as it is usually not purely the land cover that adds the cultural meaning but a combination of different attributes.

Tier 2 suits better for CES mapping as more detailed and specific data can be used to give variation to the characteristics of an area. Types of data beneficial for mapping potential CES capacity include data on cultural heritage sites, sites with events combining culture and nature, spiritual or religious sites, habitats of symbolic species (caution in visualising sites of threatened species on a map must be applied), or recreational facilities, such as trails or campsites. The selection of data depends naturally on the CES in question. In the demand side, statistics of recreational visitors or the number of fishing licences in an area, for example, can be used for recreational ES demand, number of visitors of a religious event linked to a specific natural site for spiritual ES, or number of photos taken of beautiful scenery as a measure of aesthetic appreciation of a place.

Tier 3 mapping, based on process-based models, could be understood as modelling the availability of, the accessibility to and the demand for CES that are needed in a certain place. The exceptions are intrinsic and bequest values of nature as CES. They can be understood as services which people need not necessarily be able to experience or to see by themselves but which they want to preserve because of their value for current and future generations and for which they feel joy and thus receive a non-material benefit. People can identify these kinds of places on the map or they can name specific species or habitats that they value after which they can be placed on a map based on other data.

In the following, we give examples of some available CES mapping methods that represent Tier 1 (mapping CES demand using a matrix; Chapter 5.6.4), Tier 2 (photo series analysis and ESTIMAP-recreation model) and Tier 3 (viewshed analysis).

Mapping CES using a matrix-based approach

A matrix-based approach can be used as a quick and relatively easy way to map supply of or demand for CES. In its most simple form, only land cover data or similar one dataset is sufficient for this purpose. If supply is mapped, experts can be asked to score each land cover type based on its capacity to provide different CES. On the other hand, residents of the study area can be asked to do the same based on how important they personally perceive the different land cover types in terms of CES, i.e. for which land cover types they have demand. As a result, a number of scored matrices are produced in both cases. After some basic statistical operations, such as calculating variance and medians of the given scores, a result matrix is produced. This can be easily transferred to a GIS and combined with the land cover data to produce a map (see Chapter 5.6.4).

An example of a result matrix and a map produced from it is presented in Table 3 and Figure 1. The example stems from a real-life planning process in the city of Jarvenpaa, Finland, where an open participatory workshop was arranged for the residents to map the demand for CES. Participants of the workshop were given clear instructions for the matrix scoring task both orally and on paper. In addition, written explanations of different CES classes were also given as guidance. The CICES classification was used as a basis but the CES classes were simplified and broken down to sub-classes in a way that was easily understandable for laymen. The previously created green infrastructure (GI) typology for the city was used as land cover data (see the GI typology map in Chapter 7.3.1). Participants scored individually each GI type (= environment type) based on how important it was to them personally in terms of different CES. In the Jarvenpaa case, the matrix task was followed by a spatially-explicit map exercise in another room. This allowed for both a general overview of the demand for different CES in different environments, as well as spatially-explicit knowledge of locations that people value.

Figure 1.

Mapped demand for CES based on scored matrices by individual residents in the City of Jarvenpaa, Finland.

Table 3.Download as CSV 

Demand for different types of green and blue environments as a source of CES based on scored matrices by individual residents in the City of Jarvenpaa, Finland.

CES sub-class / GI type	Forests	Croplands	Grasslands	Allotments	Allotments with huts	Urban parks	House green	Green buffer zones	Mires and wetlands	Lakes	Rivers	Creeks
Recreation in nature	2.0	0.8	1.4	0.8	0.6	1.9	1.9	1.3	1.3	2.0	1.8	1.1
Nature as a subject matter and site for education	1.9	1.3	1.9	1.0	1.0	1.5	1.5	1.4	1.6	1.8	1.8	1.5
Natural aesthetics	1.9	1.3	1.9	1.1	1.3	1.8	1.8	1.5	1.6	2.0	1.9	1.8
Artistic inspiration from nature	1.9	1.4	1.6	0.6	0.7	1.7	1.4	1.1	1.4	1.7	1.7	1.6
Identity value of nature	1.7	1.0	1.4	0.7	0.4	1.7	1.6	0.7	0.8	1.7	1.4	1.3
Place for obtaining empowerment from nature	2.0	0.9	1.3	0.7	0.4	1.6	1.6	0.7	1.0	1.9	1.6	1.3
Feeling of holiness in nature	1.7	0.7	1.1	0.1	0.3	1.0	1.0	0.6	0.9	1.3	1.3	0.9
Intrinsic and bequest values of nature	1.7	1.0	1.1	0.8	0.7	1.6	1.3	0.7	1.7	2.0	2.0	1.6
All CES together	1.8	1.0	1.4	0.8	0.7	1.5	1.6	1.0	1.3	1.7	1.6	1.3

When simple matrix-based maps are used, the restrictions of the method must be kept in mind. The demand map, such as in the given example, reflects the perceptions of people in a given location and they are seldom transferrable to other locations. They are also coarse generalisations and, in reality, there can be several factors that either improve or diminish the demand for certain locations even if the type of environment is important in general. For example, a forest may be located next to an industry with problematic emissions or the quality of water in a certain lake is poor and even aesthetically unpleasing.

Photo series analysis

There is limited access to spatially explicit data in relation to cultural activities. Yet, there is a growing need for territorial planning to incorporate the perception of numbers of visitors who could be attracted by landscape aesthetic or cultural heritage amongst other key cultural values. As representative field data are expensive and time consuming to gather, gaining understanding on how CES can be spatially defined and visualised is still challenging. A novel way to overcome this is to use crowdsourcing information.

Until recently, user generated contents are providing volunteered geographic information in different place-based applications. The very fast rate of image uploading on popular social media platforms offers potential for a new mapping paradigm based on a crowd of observers. Recent studies have used geo-located photographs retrieved from online platforms to explore place perception. Public image storage analysis has already been applied in studies for assessing CES. These studies suggest an empirical approach based on the location of visitors, assuming that visitors are attracted by the location where they take photographs. This approach opens up opportunities to directly analyse the presence of beneficiaries on the provision site which provides a proxy for actual service provision.

The main limitation of using public image storage analysis to retrieve geo-located information is given by the representativeness of the social media platforms or in relation to specific groups. However, the taken photographs can be considered as observed people's preferences which are less vague than declared preferences. The spatial distribution of visitors' preferences provides an indicator of CES, allowing a local analysis of service providing areas and addressing the lack of quantitative indicators of CES. Application Programming Interfaces (APIs) are publicly available for Web 2.0 applications, such as Flickr and Panoramio, allowing accessing the data, including the photograph's metadata, tags and geographic position.

Different spatial analysis methods can be applied to analyse the specific patterns and identify the landscape settings which shape the actual service provision. A systematic visual analytic process, based on expert knowledge, also allows the identification of different CES categories and their relative importance (Figure 2). Photographs of animal and plant species can, for instance, be classified as "experiential use and enjoyment of wildlife", while photographs of sport and recreational activities, such as skiing, climbing, hiking and camping, represent "physical use of landscape". Other categories such as "landscape aesthetics" and "cultural heritage" can also be identified by photo-content analysis.

Figure 2.

Photo location and count by CES category in Quatre Montagnes Region as case study demonstration (French Alpine Mountain Range).

Moreover, the temporal attributes of the photo-series (date), available on most public photo-archives, can be used to analyse the seasonality of CES (Figure 3). Specific time and location may show over supply, therefore conflict and trade-offs between different ES can also be mapped.

Figure 3.

Example: Seasonality of CES categories in the study site of Quatre Montagnes.

The photo-series analysis can be applied at different spatial scales, ranging from municipality to national, according to the context, photo-density and positional accuracy of the photographs. The final service provision map can therefore inform stakeholders and policy makers at different institutional levels on priority areas (Figure 4). Finally, the analysis of community-contributed photographs can be used to design location-based interviews, questionnaires or focus groups in order to take into account socio- and psycho-cultural aspects which are related to CES values.

Figure 4

Methodological framework for exploratory analysis of CES through Geographically Weighted Regression (GWR). Given the high diversity of habitats and ecosystems in the study site of Quatre Montagnes, we assume that CES delivery is context-specific and we expect a significant geographical variation in the relations between the photo-count and explanatory variables. The variables correspond to the physical (environmental settings) and infrastructure (opportunity settings) characteristics of the landscape whose spatial variation may affect the CES provision. The spatially weighted regression showed that specific variables correspond to prominent drivers of CES at the local scale. Dominant habitat, accessibility, diversity of habitat and proximity to view points were identified as the variables having a major impact on CES.

Mapping landscape aesthetic service through viewshed analysis

The aesthetic value of landscapes, such as scenic beauty, represents a specific category of CES which has received growing attention in the socio-ecological research. Although the visual aesthetic quality of landscapes has been researched for centuries, standardised and quantitative assessment approaches are so far scarce.

Geo-tagged photographs uploaded on online photo-sharing platforms can be used to locate aesthetically attractive areas and derive the frequentation rate. Together with the biophysical and built-up characteristics of the landscape, the photo-series allows the analysis of complex visual landscapes which are associated with scenic beauty as it is perceived by beneficiaries at specific locations. In open areas, as scenic beauty is especially related to panoramic view, photographs capturing panoramas can therefore be used as spatially explicit data of actual service provision. These data can be related with biophysical factors of the landscapes seen from the respective viewpoints. The visible area and the respective visual indicators can be calculated for each theoretic viewshed, derived from a Digital Elevation Model, corresponding to the photograph location. The viewshed is thus considered as a Service Providing Area from the perspective of the beneficiary (Chapter 5.2). The landscape aesthetic theory allows the linking of landscape visual indicators to the landscape's visual characteristics. Those indicators represent the landscape structures related to the information functions of the landscape which contribute to enjoyment of scenery as a final service. A quantitative framework can thus be applied to identify the landscape variables contributing to the visual landscape attraction.

The procedure for capturing and mapping the visual character of the landscape has been applied in the same study region as the photo series analysis (Figure 5). The analysis allowed the evaluation of the visual landscape preferences by considering the information from the users' source and assuming the relationship between the mental landscape perceptions and the visual scale. Different visual indicators were considered which refer to six different components of the landscape: depth, relief, land cover, landform, geology and habitat. Each indicator was linked to nine visual concepts, describing different landscape characters and landscape aesthetic theories. The visual indicators were finally used to run a cluster analysis in order to identify spatial patterns and geographical regions (Figure 6).

Figure 5.

Methodological framework for visual landscape assessment. The visible area (viewshed) was calculated for each location of photo representing panoramic views. The four distance zones were set to respect the degrading visual properties with increasing distance from the viewpoint.

Figure 6.

Landscape clusters based on the visual indicators. Four different typologies of landscapes - corresponding to groups of viewsheds - emerged as distinct clusters. The spatial distribution of the landscape groups showed a clustered pattern, allowing a regionalisation of the landscape characters.

This approach provides a framework for performing a systematic analysis of scenic beauty aspects and facilitates interpretation of the landscape information function. By expressing the actual service provision in a spatially explicit way, we can learn about the beneficiaries' perception and the landscape's visual character providing integrated information which can support landscape monitoring and regional planning.

Modelling CES supply using the opportunity spectrum approach: ESTIMAP recreation

Public, nature-based, outdoor recreational activities include a wide variety of practices ranging from walking, jogging or running in the closest green urban area or at the river/lake/sea shore, bike riding in nature after work, picnicking, observing flora and fauna, organising daily trips to enjoy the surrounding beauty of the landscape, amongst a myriad of other possibilities. These activities have important roles in human well-being and health. While tourism is an occasional activity, local outdoor recreation affects the daily life of people. The ESTIMAP recreation model (see also chapter 4.4) assesses the capacity of ecosystems to provide nature-based outdoor recreational opportunities which can be enjoyed on a daily basis. The model consists of three parts: (1) Recreation Potential (RP); (2) Recreation Opportunity Spectrum (ROS); (3) The number of potential trips.

The Recreation Potential (RP) (Figure 7) represents a composite dimensionless indicator that estimates the potential capacity of a group of identified landscapes and features to provide opportunities for local outdoor recreation. The provision varies according to four main components: (1) the suitability of land to support recreational activities; (2) the blue-green infrastructure in urban areas; (3) the presence and typology of natural protected areas and natural features; (4) the presence and quality of water bodies and coastal areas (inland and sea).

Figure 7.

Potential nature-based opportunities for recreation in Europe.

The Recreation Opportunity Spectrum (ROS). People can benefit from the opportunities provided by nature for recreational activities if they are able to reach them. The ROS was chosen as a method to map different degrees of service available according to their proximity to the people. First, a proximity map is computed by combining Euclidean distance from urban and Euclidean distance from roads. A final map of recreation opportunities is then computed by a cross tabulation between the RP and the Proximity using a second set of parameters with thresholds for the degree of recreation opportunities provided by nature and the degree of proximity and remoteness. Parameters can be based on national standards or law (normative) or on observed data.

Number of potential trips

The potential flow of the service to visitors can be estimated by computing the share of potential trips that can theoretically be undertaken in order to reach the different ROS zones. As mentioned above, the present model addresses daily recreation therefore, according to literature, two reference distances were identified for close-to-home and daily maximum travelled distance: 8 and 80 km.

A moving window with a kernel file is applied to a raster grid of population density to compute an estimate of potential trips per each pixel in the grid per day. Figure 8 shows a map of potential close-to-home trips.

Figure 8.

Potential close-to-home trips in Europe. The graph represents the shape of a distance decay function which can be used to model the close-to-home trips. Y axis represents the decay function, X axis the distance.

The percentage of potential trips per ROS zone can be calculated by dividing the sum of potential trips per ROS zone by the total of all possible trips, see graphs in Figure 9.

Figure 9.

The Recreation Opportunity Spectrum in Europe. More details are provided for two cities (Naples and Helsinki); for both cities, an estimate of close-to-home potential trips was computed. Pie charts represent the percentage of potential trips to all ROS categories.

Burkhard B, Kroll F, Nedkov S, Muller F (2012) Mapping supply, demand and budgets of ecosystem services. Ecological Indicators 21: 17-29.
Maes J, Crossman ND, Burkhard B (2016) Mapping ecosystem services. In: Potschin M, Haines-Young R, Fish R, Turner RK (Eds.) Routledge Handbook of Ecosystem Services. Routledge, London, 188-204.

Adrienne Gret-Regamey, Bettina Weibel, Sven-Erik Rabe & Benjamin Burkhard

A tiered approach for ecosystem services mapping

Similar to the approaches mentioned above, a tiered approach for ES mapping is proposed in this chapter: it is most useful to define the tiers according to the goal of the mapping exercise (see Chapter 5.4) to make sure the information relevant for the related decision-making process is provided. This supports the efficiency of the mapping process avoiding far too complex approaches where rough estimates would be sufficient.

In a first step, the different components of the analysed human-environment system should be described which include the ecosystems and ES as well as the beneficiaries and institutions involved and their interactions. For example, for microclimate regulation in urban areas, the considered ecosystems are usually green urban areas, the service they provide is microclimate regulation, beneficiaries are residents and institutions are city planning agencies. These system components can be described at different levels of detail, for example, the ecosystem can be described in terms of its condition and structure (see Chapter 3.5), the service provided can be quantified in different units (see Chapter 2.4), the ES demand can be structured according to different beneficiary groups (Chapter 5.1) and different instruments of institutions including NGOs or businesses (see Chapter 7), for example, can be identified. This description of components should make the boundary of the considered system and the spatial and temporal scale explicit. ES beneficiaries and institutions represent relevant stakeholders who could be considered in the decision-making process.

Once these components have been described, the appropriate tier and associated ES mapping method can be selected. To guide this selection, we present a decision tree in Figure 1. The first question addresses the process-understanding of the human-environment system. If interactions between the system components are relevant and a deeper understanding of processes is needed (e.g. to understand how management of ecosystem components can influence the provision of ES), a tier 3 approach would be required. Otherwise, if the purpose of the map is mainly to provide a rough overview of ES values in a certain area, their abundance, presence and absence, a tier 1 approach can be selected. If information about different ES is required at a certain level of detail but not linked to an explicit management question tackling the human-environment system components and processes, a tier 2 approach may be suitable. However, if the ES map is to be used to explicitly evaluate management measures, again a tier 3 approach should be considered. After the most suitable tier has been identified, the availability of resources for the ES mapping should be evaluated. In case resources are severely limited, a method involving a lower tier can be applied. Yet, efforts should be made to identify the most suitable tier to provide information that is useful for decision-makers.

Figure 1.

Decision tree guiding the selection of tiers for ES mapping.

We associated the five different categories of ES mapping methods (see above) with the different tier levels: while most methods are applicable at all tier levels, they usually have a focus at a certain level as indicated in Figure 1 with the shading. ES quantification and mapping methods are described in more detail in Chapters 4 and 5.

Box 1. Illustrating the tiered approach: Microclimate regulation

In this example, we illustrate the tiered approach for mapping microclimate regulation within urban areas with ES mainly provided by green space and important in the context of heat island effects. The components of the human-environment system include green urban spaces as ecosystems, microclimate regulation as provided ES, residents as the main user group and city planning agencies as main institutions. If the purpose is to provide a rough overview, i.e. to compare cities or city districts, no detailed process-understanding is required and a tier 1 approach would be most suitable. Using a lookup table approach, the microclimate regulation can be estimated based on the amount of green space as illustrated in Figure 2. Alternatively, experts could also rank the different land use/land cover (LU/LC) classes according to their suitability for providing microclimate regulation.

If the map is to be used to analyse microclimate regulation in more detail without providing information for an explicit management measure targeting system components or processes, a tier 2 approach can be applied. Here, we present a causal relationship approach, where the green volume is estimated by combining high resolution remote sensing data with LU/LC information: Green areas are estimated from the remote sensing information based on the normalised-difference-vegetation-index (NDVI), which allows, for example, identifying single trees. Additionally, the remote sensing data provides information about the height of these identified green areas to estimate the volume. As reducing the urban heat islands by increasing microclimate regulation requires an understanding of how certain measures such as changes in the amount and/or structure of green area quantitatively affect the cooling potential, a process-understanding is needed guiding us to a tier 3 approach.

In a tier 3 approach, the cooling effect is estimated based on a model combining ecological information, i.e. the cooling potential of various vegetation types with the given green infrastructure and their green volumes: the volume of green infrastructure can be derived from a detailed land use typology at the cadastral level based on field surveys with classes such as private yards, sport facilities and infrastructural green. Each class of the typology is related to the amount of trees, grasses, shrubs and settlement or infrastructure present. For the categories tree, grass and shrubs, the volume is estimated based on well-known geometric relations and combined with remote sensing information. The potential cooling effect for high, middle and low green infrastructure can then be modelled considering climate information such as precipitation, temperature and solar radiation. Finally, the effect of infrastructure such as roads or buildings on the cooling potential is considered for estimating the resulting cooling effect.

Figure 2.

Illustrating the tiered approach for microclimate regulation.

Nora Fagerholm & Ignacio Palomo

Collecting data through participatory mapping approaches

Data collection with PGIS approaches represents pluralism. Common data collection methods include self-administered surveys, either web- or paper-based, face-to face surveys and workshops (see Table 1 for a comparison of these methods). Mapping activity typically engages the lay-public such as residents or visitors to an area but also various stakeholders including land holders, environmental professionals, planning practitioners and other experts. Random and meaningful sampling, on site recruitment and volunteered open participation or different methods for stakeholder prioritisation can be applied to select participants for the mapping process.

Table 1.Download as CSV 

Three common data collection approaches in PGIS (web-based surveys, face-to-face surveys and workshops) for ES assessment and their characteristics.

Characteristics	Web-based surveys	Face-to-face surveys	Workshops
Participants type	Often lay public	Lay public and experts	Often experts (e.g. local inhabitants with ecological knowledge, environmental specialists, planning practitioners)
Time, cost and facilitation requirements	Time efficient but resources needed for inviting participants, no facilitation needed and participation not restricted by specific time and place	Time consuming as each person needs to be met individually, resources also needed for inviting participants and training interviewers	Time efficient as it allows all data gathering during the workshop, but demanding for preparation and training facilitators
Sampling method	Random sampling, volunteered open participation	Random sampling, on site recruitment, meaningful sampling, stakeholder prioritisation	Meaningful sampling, stakeholder prioritisation
Sample size and representativeness	Easier to reach a larger and more representative sample, although survey respondent rate often remains low (under 15%)	Depends on available resources, possibility to control representativeness	Remains often low, statistical representativeness not targeted
Type of participation and its effect on data quality	Instrumental and self-administered, difficult to analyse the level of understanding of the participant and data quality	Self-administered but allows facilitation and a detailed exploration of the issue analysed, contributes positively to data quality	Allows communication among participants and detailed exploration of the issue analysed (deliberative mapping), contributes positively to data quality

PGIS mapping of ES involves either digital mapping interfaces, often web-based, with zoomable background maps or printed map layouts commonly presenting one given scale. Information, given on the background maps typically includes aerial/satellite image overlaid with basic map elements, or topographic maps showing, for example, relief and basic natural and man-made features. The most often applied method for marking has been point placement (e.g. movable plastic discs or stickers) followed by drawing polygons presenting areas or using predefined land units as a basis for assigning values.

Most commonly applied typologies for mapping include ES classifications (MA, TEEB, CICES, see Chapter 2.4), adaptations of these to specific contexts or landscape values and landscape services typologies developed in case studies and based on research on social values. Direct ES identification and valuation through an inductive approach, not deriving from a given typology, have been rarely applied.

Analytical process

A wide variety of analytical approaches can be applied to PGIS data on ES (Figure 1). Typically the analytical process starts by describing the characteristics of informants who participated in ES mapping. Spatial analysis often begins with description of the spatial patterns and characteristics of ES through testing the level of clustering or dispersion, intensity/density estimation, diversity of ES, identification of hotspots and by calculating distances, for example, to the respondent's home. Between pairs of ES, the spatial overlap has been studied through correlation analysis to look at the co-existence of ES. In spatial analysis of mapped ES, areas have a predefined precise boundary but points are treated as representing the centroid of the spatial occurrence of a specific ES extending outwards to an unknown distance.

Figure 1.

Example of an analytical process for PGIS data from basic descriptive steps to more advanced spatial and statistical analysis where PGIS data is integrated with other spatial data sets.

Spatial concurrence is commonly studied through overlay analysis to explain the relationship to physical land features (such as land cover, land use, management units, land change) or ecological data. In addition, spatial indices such as landscape metrics derived indices are common to quantify the distribution across different land use or management units within the study area. Clustering techniques have been found useful for exploring the potential relation between the mapped ES and, for example, land use and socio-demographic characteristics of informants. Interest has also been paid to extrapolate and model the distribution of the participatory mapped ES to locations where data were not collected through value transfer methods. Analysis of ES bundles has not been frequent but is gaining more attention as is the analysis of ES flows and trade-offs.

Joerg A. Priess & Leena Kopperoinen

Which methods and tools are available for participatory mapping

In this section, we briefly present four of the participatory mapping methods available for CS in the context of ES mapping and assessment, covering work with conventional paper maps or tables, or digital tools such as geographic information systems (GIS) or smartphone apps. All methods can be used with different levels of citizen and scientist involvement. The higher the level of involvement of citizens, the higher the level of knowledge needed for citizens, for instance, about different ES, to handle spatial data in a geographic information system, to prepare paper or digital maps of the study area, or to evaluate information generated during the CS project.

In every ES mapping approach, citizens and scientists need to define the mapping perspective, i.e. whether they want to address the (potential) supply of, the demand for, or the actual use (= flow) of ES (see Chapter 5.1). Additionally, in most approaches, spatial and temporal units / coverage also need to be clarified (see Chapter 5.7.5).

ES use has been mapped by citizens and scientists using the MapNat smartphone application. Colours of flags indicate different types of ES use. The selected ES use also indicates the frequency of use and the importance, both reflecting the view of the person who mapped the ES use (Figure 1).

Figure 1.

ES use in Europe.

Four CS-compatible mapping examples

Mapping ES with paper maps: Identifying and locating ES on topographic or thematic maps can be carried out as an indoor or an outdoor approach, the latter enabling participants (at least partly) to view and observe the area of interest. Mapping units may be predefined, for example, using units of land cover, or may be identified during the project as, for example, spatial units are assumed to differ between ES. For a quick qualitative ES mapping/assessment (tier 1; see Chapter 5.6.1), one workshop or one field visit may be sufficient, while the generation of more detailed information (tiers 2-3) can be expected to require more time and/or additional sources of information. ES identification and mapping can be carried out individually, either resulting in calculated ES means or ranges or both. Alternatively, ES can be mapped based on a group consensus (called deliberative mapping). The approach can be used to map (potential) supply of and demand for ES, trade-offs, mismatches etc. Data collected with paper maps can be digitised afterwards in GIS (Figures 2-3: Example from Sipoo, Finland).

Figure 2.

Using paper maps in a local master plan exhibition to collect cultural ES related values, in Sipoo, Finland. Although a digital map of the planning area (see computer in front) was available for citizens, paper maps were preferred by them.

Figure 3.

A map presenting the opportunity spectrum of the CES group 'Aesthetics and cultural heritage' in the background and residents' point and polygon markings of the same CES group. Examples of open-ended explanations of the markings have been added on the map. The border of the local master plan area is shown as well.

Mapping ES with GIS: This approach is comparable to the first, the difference being the replacement of paper maps by PCs, laptops or tablet computers (see also PPGIS Chapter 5.6.2). The great advantage of digital mapping is that different types of spatial information can be linked or combined, for example, to derive appropriate mapping units, as well as evaluating ES mapping results. However, at least one citizen or scientist with GIS software experience is needed. The approach can be used to map (potential) supply of and demand for ES. The technical threshold might, however, invoke a selection bias dependent on the knowledge of the involved participant.

The Matrix method to map ES: This method is presented in Chapter 5.6.4. In CS projects, both paper and digital versions may be used. As explained in the previous examples, the matrix method can also be applied individually or as a group exercise and the approach is also suitable for mapping (potential) supply of and demand for ES and, additionally, actual ES use.

Use of a smartphone app such as MapNat for ES mapping: Similar to mapping ES with GIS, MapNat is a participatory GIS approach (see Chapter 5.6.2). Mapping ES uses the GPS (geo-positioning) unit of the smartphone or a tablet to locate ES at the current position of the user. Alternatively, app users can just use their fingers to map ES directly on the device's screen. In this tool, the mapping perspective is focusing on recording the actual use of ES, either directly during ES use, or afterwards identifying the location in the app's map view. This method can also be applied individually or as a group exercise. Contrasting with the previous examples, this tool also provides access to the ES records and valuations of all other app users worldwide (Figure 1), because all records are sent to and redistributed by an internet server.

Benjamin Burkhard

Introduction

Ecosystem services (ES) are spatio-temporal explicit phenomena. Thus, ES supply, flow and demand (see Chapter 5.1) can be linked to units in space and time. One mapping method is the ES 'matrix' approach, which links ES to appropriate geo-biophysical spatial units. Thereafter, their supply, flow and/or demand are ranked using a relative scale ranging from 0 to 5 (not relevant to very high, see Figure 1). Based on this normalisation of ES rankings, various ES are made comparable and different points in time (including scenarios) can be assessed. Therefore, the approach has the potential to integrate all kinds of ES-related data based on diverse scientific disciplines or ES quantification methods (see Chapter 4) and of varying quality and quantity in illustrative matrix tables and maps. It can be applied in data-poor as well as in data-rich study areas, fulfilling mapping purposes from first ES screening studies and awareness-raising to very comprehensive integrated trans-disciplinary ES assessments. In this Chapter, the ES matrix approach is described and related uncertainties are discussed.

Figure 1.

Overview of the ES matrix approach, based on geospatial map data, the actual matrix and resulting ES maps.

Data sources and quantification methods

In its simplest form of application, the ES matrix assessment uses spatial LULC data as proxies for ES supply. The advantage of LULC data is, besides its availability in many regions of the world, that many provisioning ES (see Chapters 2.4 and 5.5.2) can be specifically and uniquely linked to single LULC types. Timber, for example, is harvested from forests, crops grow on agricultural fields and fish and seafood occur only in water bodies, rivers and the ocean. Regulating ES (see Chapter 5.5.1) and cultural ES (see Chapter 5.5.3) are usually supplied in well-functioning and not too far degraded ecosystems which can be related to more natural LULC types. ES maps, based on LULC information, provide important spatial landscape information which already can be helpful for the identification and awareness-raising of ES and their supply and demand patterns.

In case the ES mapping purpose goes beyond providing a rough overview of ES supply or demand in space, further data and ES quantification approaches can be integrated. The tiered ES mapping approach (see Chapter 5.6.1) helps select the appropriate method based on the mapping purpose, the necessary process-understanding and needed explicit measures and, last but not least, the data and resources availability.

ES data from all three mapping tiers can be integrated into the ES matrix. The use of expert knowledge for ES quantification and qualification has, for example, become very popular and increasingly accepted within the scientific community. More comprehensive ES assessments would otherwise demand large resources in terms of time and personnel. Data from statistics, for instance about agricultural or forestry production or existing studies with relevant information - if available and appropriate - should further be integrated in the ES matrix assessment. Model outcomes provide further useful data applicable for ES mapping (see Chapter 4.4).

In an optimum case, data from all tiers can be acquired for the same area, time and spatial scale and in comparable resolution. These data can then be triangulated in order to be cross-checked and to find the most suitable, reliable and useful (for the specific mapping purpose) ES quantification and mapping method (see also Chapter 4.6). Figure 2 shows how such triangulation could take place based on data from the three tiers, normalised to the relative 0-5 scale on which the ES matrix approach is based.

Figure 2.

Data triangulation across ES mapping tiers illustrating the outcomes of different methods on the relative 0-5 scale. The ES map-maker needs to decide which value is most reliable, realistic and useful for the actual ES mapping purpose.

Reid WV, Berkes F, Wilbanks T, Capistrano D (Eds.) (2006) Bridging scales and knowledge systems: concepts and applications in ecosystem assessment / Millennium Ecosystem Assessment. Island Press/World Resources Institute, Washington, DC.

Marion Kruse

Sharolyn Anderson, Alberto Giordano, Robert Costanza, Ida Kubiszewski, Paul Sutton, Joachim Maes & Anne Neale

Data Sources

There are myriad sources of data that can potentially inform and contribute to the production of maps for ecosystems and ES (see Section 4). A non-exhaustive list might include various types of satellite imagery, human population census data, agricultural productivity statistics, soil maps, vegetation maps, air quality measurements, biological census data, transportation and other infrastructure maps and climate station data and maps (http://biodiversity.europa.eu/maes). These data can be applied to the production of different kinds of ecosystems and ES mapping.

A key question to answer is how to structure and organise the representation of ES? This question applies to all cartographic representations ranging from the local to the regional, to the national and to the international. One approach is to create a separate layer for every ecosystem service (e.g. one layer for carbon sequestration, one for erosion control, one for spiritual values etc.). This approach is convenient from a taxonomic perspective but can be problematic, as variations in most of these services are driven by land cover proxy measurements (e.g. boreal forests sequester X kg/ha/year whilst deserts sequester Y kg/ha/year), but, in others, they vary as a function of spatial interactions with other spatially variable information (e.g. spiritual value will likely vary as a function of proximate population density, the income of that population and the spiritual values of the proximate population). Carbon sequestration provides a salient example of the relevance of these issues. It is increasingly regarded as a policy-relevant ecosystem service as a result of climate change. At a national level, authoritative, verifiable and valid ground-based measures of carbon sequestration which include direct measurements of vegetation and soil would likely be needed to produce a comprehensive, country-wide map of carbon sequestration.

Scientific accuracy, transparent methods of measurements and reliable and independent interpretation and dissemination of results would be needed to ensure the legitimacy of the process, both internally at the country level and in the international arena. Here, again, we run into the problem of economic costs, in the sense that valid and authoritative maps representing real and dynamic phenomena may be expensive to produce, maintain and update at the required levels of cartographic detail, accuracy and currency. For example, the 2010 United States Census of the Population cost approximately $13 billion to conduct, or over $40 per person counted and mapped. The degree to which large investments can be made by individual countries in order to map ecosystems and ES remain to be seen. Perhaps the solution is partnerships between countries - examples include the European Union's Joint Research Centre (JRC) and the United Nations Environmental Programme (UNEP) - as well as efforts by individual countries to create, maintain and share primary environmental data, including initiatives by US government agencies (for example the National Aeronautic and Space Administration (NASA) and the National Oceanic and Atmospheric Organisation (NOAA)).

A map of recreation potential offered by the traditional cultural landscape and nature. This map is based on the recreation opportunity spectrum approach. The red dots are places of archaeological interest.

Map of the water storage capacity of soil (expressed in mm) in the Netherlands is derived from the Atlas of Natural capital which collects spatially explicit data of ES at national scale.

Box 2. Mapping ecosystem services at the national extent for the conterminous United States

In the US, the Environment Protection Agency leads a multi-organisation effort to develop and host a suite of nationwide maps of ecosystem services (ES) indicators and indices in EnviroAtlas (https://epa.gov/enviroatlas). This open access tool allows users to view, analyse and download a wealth of geospatial data and other resources related to ecosystem goods and services. More than 160 national indicators of ecosystem service supply, demand and drivers of change provide a framework to form decisions and policies at multiple spatial scales, educate a range of audiences and supply data for research. A higher resolution component is also available, providing data for finer-scale analyses for selected communities across the US. The ecosystem goods and services data are organised into seven general ecosystem benefit categories: clean and plentiful water; natural hazard mitigation; food, fuel and materials; climate stabilisation; clean air; biodiversity conservation; and recreation, culture and aesthetics. EnviroAtlas incorporates many data sources with multi-resolution (i.e., 1 m and 30 m) land cover data providing fundamental information. The data are updated at 5 year increments, subsequent to US National Land Cover Dataset updates.

This map shows the kind of data layers that are available in EnviroAtlas. For one of the indicators in the climate stabilisation category, this map shows the amount of carbon stored in the above-ground tree biomass. Like most of the national maps in EnviroAtlas, the data are summarised by medium sized watershed drainage basins known as 12-digit hydrological unit codes (HUCS). There are approximately 85,000 of these HUCS in the conterminous US, with each being approximately 104 km?. Users of EnviroAtlas can also overlay demographic maps to gain the perspective of proximity and population dynamics of beneficiaries.

Katalin Petz, Clara J. Veerkamp & Rob Alkemade

Box 1. Example soil erosion prevention on global rangelands

Figure 1 provides an example map for the current state of soil erosion prevention on global rangelands. A further developed version of the Universal Soil Loss Equation (USLE) applied in the IMAGE model was used for mapping this service. Erosion prevention was mapped with an index (0-100) based on soil erodibility, rainfall erosivity, both derived from IMAGE and a refined land use/cover index derived from the vegetation cover fractions of the Global Land Cover map. Low erosion prevention (i.e. high erosion risk) is the result of steep slopes, sensitive soil and scarce vegetation cover (e.g. in the Mediterranean, Central Australia and Chile).

Figure 1.

Soil erosion prevention ES on global rangelands (Petz et al. 2014).

Box 2. Global crop production under two extreme scenarios

With the help of scenarios, the trends of ES delivery can be projected over time. In this example, the global crop production is simulated with the IMAGE and GLOBIO-ES models for two future scenarios. The production of cereals, rice, maize, pulses, root and tubers is taken as an indicator of crop production. The demand for crops is driven by changing lifestyle and population, whereas technology, environmental factors and management determine the production efficiency hence the crop yield. The two scenarios are adjusted SSP scenarios (i.e. new IPCC scenarios) used in the OpenNESS EU project. The 'Wealth-Being' (WB) scenario stands for economic growth, while the 'Eco-Centre' (EC) scenario promotes sustainable management around the globe. Figure 2 illustrates the potential change in crop yield in 2050 in comparison to the base year of 2010. Crop yield increases in developing countries (e.g. Africa, India) in the EC scenario, while the WB scenario projects lower crop yield in these countries, but higher yield increase in US and Brazil.

Figure 2.

Change of crop production under two extreme scenarios (PBL 2016).

Evangelia G Drakou, Camino Liquete, Nicola Beaumont, Arjen Boon, Markku Viitasalo & Vera Agostini

ES provided by marine and coastal habitat types

Each marine or coastal habitat type can generate different ecological functions which can then generate ES for the benefit of human beings. In Table 1, we list the major marine and coastal habitats and the MCES they provide according to what has been documented in the literature. The missing links between habitats and ES highlight the areas with the largest knowledge gaps, but not the lack of a link. It is worth mentioning here that very few of these ES have been actually mapped.

Table 1.Download as CSV 

Major marine and coastal habitat types and their links with ES as documented in the literature. The (u) symbol represents the relationships between habitat types and ES that have been assessed and documented in the literature. The (?) is there to represent the lack of sufficient knowledge to assess and hence quantify and map this relationship.

* These habitats and ES are still very poorly analysed.

Provisioning

Regulating and maintenance

Cultural

Food provision

Water storage / provision

Biotic materials/Biofuels

Water purification

Air quality regulation*

Coastal protection

Climate regulation

Ocean nourishment

Life cycle maintenance

Biological regulation*

Recreation Tourism

Symbolic/Aesthetic values

Cognitive effects

Beach and dunes

?

?

?

?

?

?

?

?

?

?

?

?

?

Coastal wetland

?

?

?

?

?

?

?

?

?

?

?

?

?

Estuary

?

?

?

?

?

?

?

?

?

?

?

Mangrove

?

?

?

?

?

?

?

?

?

?

?

?

?

Coral Reef

?

?

?

?

?

?

?

?

?

?

?

?

?

Maerl bed*

?

?

?

?

?

?

?

?

?

?

?

?

?

Oyster reef

?

?

?

?

?

?

?

?

?

?

?

?

?

Macroalgal bed

?

?

?

?

?

?

?

?

?

?

?

?

?

Seagrass meadow

?

?

?

?

?

?

?

?

?

?

?

?

?

Unconsolidated sediments

?

?

?

?

?

?

?

?

?

?

?

?

?

Open ocean pelagic

?

?

?

?

?

?

?

?

?

?

?

?

?

Mapping marine and coastal ecosystem services

To map ES provided by marine and coastal ecosystems similarly to the terrestrial ecosystems, one has to understand the process of ES provision, from the ecosystem components, functions and processes to the actual ES. For each component of the ES provision chain, data need to be acquired and quantification methods applied throughout. This information can be used to spatially represent the ES distribution. In Figure 1 we illustrate the process of generating a map of MCES with a hypothetical example.

Figure 1.

The figure depicts the way data and ecological models contribute to the different components of a basic ecosystem service generation framework (ES cascade at the bottom of the figure) in order to generate ES maps. In an example of whale watching tourism as an ES provided by whales, species and habitat distribution models are used to describe the basic ES components. Then models are used to describe the ecosystem functions. The outputs of all these models are then combined along with socio-economic parameters (in the example we refer to the number of whale watchers, but it could also be revenues from whale watching) in order to generate a final map of the benefit or value from whale-watching tourism. The arrows show the flow of information within the elements of the ES cascade.

In the oceans and coastal seas, many ecosystem functions occur within the water column which adds a third spatial dimension to the system. These functions change with depth, water temperature, solar irradiance, salinity and other factors and are extremely variable in space and time. This makes it difficult to capture this information in two-dimensional maps.

MCES maps are delivered by:

Analysis of primary data, for example, high resolution remote sensing of the coastal and pelagic zone, field sampling and socio-economic surveys. It can be very accurate, but it is also time and resource consuming.

Habitat maps can be used to translate seabed habitat maps into capacity to deliver ES based on scoring factors. This method can be feasible and quick if the seabed habitat maps of the study area are already available. However, the scoring system can be subjective and the results reflect only the services provided by benthic habitats.

Modelling

Models such as those below can be used:

Ecosystem models optimally integrated with socio-economic data, or bio-economic models. They can be relatively accurate with quantifiable uncertainty and capture three-dimensional (3D) processes across spatial scales. Still they require a lot of data, time and expertise. Model outputs may not be usable as such; composites or proxies often need to be generated for MCES mapping.

Already available MCES mapping tools (see the following section).

Most MCES maps depict the ES capacity and very few address the actual flow of, or the demand (Chapter 5.1) for MCES. The analysis of all these ES aspects is essential, especially for MCES whose use is often distant from the source of ES provision (e.g. the nutritional value of globally consumed tuna or climate regulation by mangroves in South-East Asia).

Required data for MCES mapping

The possibility of creating MCES maps is often limited due to scarcity of spatial data. For proper ES mapping, data should ideally be available for:

Habitats' spatial distribution (or their model-derived proxies);

ecological state of the habitats;

water quality affecting ES provision (e.g. eutrophication or amount of harmful substances);

species distribution of dominant, habitat forming and keystone species that either provide or support ES;

biomass of fish and other seafood;

human activities affecting the production of ES or those which could be used as indicators for ES use (e.g. fishing activity, tourism etc.).

Collecting such data is laborious and expensive, mostly because of the methodological challenges. Some examples are given in the following text.

Data on benthic habitats need to be collected with echo-sounding methods and tedious geological analysis of the sonar data. Species data need to be collected with a suite of methods that vary in spatial coverage and taxonomic accuracy. Data on sea bottom substrate and larger species can be collected with underwater cameras, while information on smaller species can be derived with underwater surveys (e.g. through scuba diving) and benthic sampling. Species identification often requires microscopic analysis.

Some proxies for ES can be created for more cost-effective methods. The new satellite instruments provide high resolution data (e.g. WorldView3 images have a resolution of 30 cm) that can be used to create proxies for some ES, like habitats essential for fish production. Semi-automatic in situ mapping devices, such as robot gliders, have been developed for collecting sea bottom data instead of cruises on research vessels. Such methods can complement, but never entirely replace, the traditional methods.

Spatial data on certain human activities can easily be derived from public databases, but in most cases data are scarce. Proxies need to be calculated although these create uncertainties in the mapping.

MCES mapping tools

Different online tools, models and methodological frameworks allow practitioners to assess and map different components of the MCES generation chain (Figure 1). Amongst the most popular and well-established ones, are the models from the InVEST (http://www.naturalcapitalproject.org/invest/) toolkit that use ecological production functions to assess the supply and demand of MCES. These can assess wave energy, coastal protection, marine fish aquaculture, marine aesthetic quality, fisheries and recreation and marine habitat provision. ARIES (http://ariesonline.org/) has also been applied for MCES assessment to generate maps mostly in coastal areas, using artificial intelligence networks and expert opinion. In most of these models, data availability and quality are the major issues that make their application difficult.

Several initiatives focus on publishing spatially explicit information regarding or potentially supporting MCES mapping. The SeaAroundUs (http://www.seaaroundus.org/data/#/spatial-catch) project has released a map server showing time series of the spatial distribution of fisheries around the globe. The EU has recently released a new tool for mapping fishing activities (MFA) for the European seas which is based on AIS (Automatic Identification System) data acquired by fishing vessels. AquaMaps also provide maps of marine species distribution globally. The Baltic Sea data and map service, by the Helsinki Commission, provides spatial data on biodiversity and human activities on sea. The Ocean Health Index Project provides a global map of ES provided by the sea and how sustainably the countries are using them.

Challenges of MCES mapping

There is a high level knowledge pool on the functioning of the marine ecosystems and high expertise on ES mapping methods. Yet these two only recently started converging in an interdisciplinary manner. Hence the number of MCES assessments that actually provide maps is still very limited. Challenges to MCES mapping include:

The dynamic three-dimensional (3D) nature of the marine environment, especially in the pelagic zone, makes it difficult to produce two-dimensional maps. Averaging over time and space is necessary and hence the level of spatial accuracy is low.

Information on the distribution of habitat is scarce or entirely lacking making it difficult to map MCES based on these habitats.

As the ecological functions and processes behind many ES, such as biological regulation, are not known or not easily quantified, their mapping is difficult.

Cultural ES, such as recreation, aesthetic information or inspiration, are based on human experiences which may be very variable. Linkage of such experiences to a specific habitat is difficult.

Data on ES demand or use is sensitive thus hard to obtain for some ES with high commercial value (e.g. food provision from fisheries).

Uncertainty in data and maps is too high to be useful in a policy context, therefore having often a negative feedback effect on momentum to create these maps.

Susanne Frank & Christine Furst

The role of spatial and temporal scales in ES mapping; scale dependencies of different (groups of) ES

Ecosystem services (ES) are scale-dependent. While a single tree can have a positive impact on the micro-climate (local scale), it does not necessarily impact the climate regulation at global scale. Interactions of spatial and temporal scales make the mapping of ES more complex. Therefore, without taking into account the age of a tree and its relations with other trees or other land uses (landscape scale/regional scale), no precise statement on its contribution to climate regulation can be derived across scales. Additionally, scale dependence is related to different perspectives, including the ES provider (supply) and the ES beneficiary (demand), as well as ES assessment (expert) and ES management (stakeholder). In this chapter, we assess and clarify the various aspects of scale interactions and perspectives in the context of ES mapping.

The difficulty of scale interactions lies in many aspects. The mapping exercise as such can be conducted in a straightforward manner at many scales; the greater challenge is the data availability. Regarding scales, the assessment of ES at the local scale is sometimes easier, because systems are smaller and thus better investigated, understood and supported by data. At the regional scale, spatial interactions between various ecosystems make the ES investigation more difficult, as boundary phenomena between ecosystems or land use types have been less investigated. Knowledge and data about the influence of composition and configuration of land use types remains limited (see Chapter 3.6). For assessing ES at continental or global scales (see Chapter 5.7.3), ES data from local, regional and national assessments (see Chapters 5.7.1 and 5.7.2) need to be up-scaled. Therefore, mapping and assessment of ES at global scale might involve high uncertainty. Additionally, indicators which are used to assess ES, are in many cases scale-sensitive.

Furthermore, the beneficiaries as well as the perception of ES benefits change across scales: supply and demand are largely scale-dependent. At local scale, individuals might be directly dependent from provisioning services such as food or water ("private ES") so that their activities (land management, purchase) are directed towards harmonising the spatio-temporal variation between supply and demand. Global services, such as global climate regulation, are relevant for humanity ("public services"). Their spatial and temporal dynamics, also called "spatio-temporal lag", are huge (see Chapter 5.2). Consequently, their perception and appreciation have the character of a shared value which complicates their assessment and the application of financial instruments such as Payments for Ecosystem Services (PES) to boost them (Table 1).

Table 1.Download as CSV 

Generalised scheme of the antagonism of ES awareness across spatial scales. *and further local actors and stakeholders

Scale	Measurement	Perception of benefits	Beneficiary	Payment for ES
Local	Easy	Good	Land owner*	Feasible
Regional
National	¦	¦	¦	¦
Continental
Global	Difficult	Poor	Human kind	Difficult

In land management and land use planning, the ES concept contributes to the assessment of sustainability from a highly integrative perspective that covers regulating, provisioning and cultural ES and allows the assessment of the value of biodiversity as a supportive backbone to enable ES supply (see Chapter 2.2). A multitude of indicators (e.g. in the context of CICES 4.3 (http://cices.eu/), Chapter 2.4) has been introduced for the different service groups. However, many of them address a specific scale so that the subsequent assessments require intense data collection, analysis and aggregation. Taking regulating services as an example, "mediation of smell/noise/visual impacts" relates to local or regional scale, while "dilution by atmosphere, freshwater and marine ecosystems" refers to regional, national or even global scale.

In this chapter, we explore how to integrate data from different scales in a comprehensive manner. Using the results from the project RegioPower (www.eli-web.com/RegioPower/) as an example, in Boxes 1-3, we show how local data can be up-scaled for supporting decision-making at the regional level.

Box 1. Bridging spatial scales

In RegioPower, we focussed on exploring regional biomass provisioning capacities and focus here on the service "timber production". Measured or modelled data, as well as stakeholder experience or expert opinion can serve as the basis for the assessment of this service. We made use of forest inventory data and regional statistics (harvesting, trade) and included empirical data when no specific information could be obtained. Through normalisation, this quantitative information basis can be adjusted for trade-off analyses with other services, such as "Aesthetics" or "Carbon sequestration". Subsequently, with the help of the software GISCAME (www.giscame.com), the effective capacity of providing services bundles and their balance can be assessed in a spatially explicit manner or as summary information at regional scale. This approach of local data collection and subsequent normalisation for up-scaling to larger scales (Figure 1) can be applied for many ES and for various spatial scales (regional, national or larger).

Figure 1.

Upscaling of local stand data (growing stock in m? per ha, right map) to the regional scale (relative scale from 0-100) using a normalization apporach (left map).

Scale interactions

To move from local data to regional decision support, various data need to be collected, harmonised and integrated. Data might encompass measured data from field studies, empirical data from surveys, modelled data, or expert judgements if quantitative data is not available. Hence, the first challenge is the identification of adequate indicators. Regarding spatial reference, a cross-scale approach might be necessary, for example, the collection of local data, in order to regionalise them for an ES assessment at the regional or national scale (Box 1).

Once the status quo of ES is assessed and mapped, the next challenge is the consideration of the temporal scale (Box 2). Provision of and demand for ES change during time. If available, historic data should be used as a basis for the development of future land use and management alternatives which should support decision-makers in finding the most sustainable planning strategies.

In addition to space and time, thematic interactions need to be taken into account to avoid unexpected trade-offs (Box 3). With the term thematic, we refer to thematically heterogeneous ES, for example, provisioning, regulating and cultural services. Various ES, which are relevant for a specific study in terms of spatial scale and management challenges, should be mapped and assessed. At least, some ES from each category (provisioning, regulating and cultural services) need to be considered for a reliable analysis of ES synergies and trade-offs. Depending on the case study framework, ES that are relevant for decision-making, should especially be considered. However, neglecting one thematic ES group might lead to unforeseen trade-offs.

Figure 2.

Temporal variation of the provisioning service "timber provision" considering three timber assortments.

Figure 3.

Mapping of various ES capacities at the regional scale and trade-off visualisation.

Chapters 4 and 5 describe many different methods and approaches for mapping ecosystem services (ES) across time and space. However, as with any mapping exercise, the usefulness of the map is only as good as the input data ("garbage in, garbage out"). It is important to be aware of the common data and quantification challenges when making ecosystem service maps to prevent production of poor quality maps.

The aim of this chapter is to discuss some of the common challenges quantifying ES for use in maps. The chapter principally focuses on challenges when data is scarce and/or system understanding is poor. Challenges relating to scale are also considered, such as the Modifiable Area Unit Problem (MAUP; see also Chapter 3.2) and the ecological fallacy, the importance of metadata (data about data) and the need to avoid double-counting of ecosystem service values in maps. This chapter will offer solutions to these problems, including a list of online spatial data resources to fill data gaps.

It is a general principle that wealthier countries and regions with advanced economies will have higher resolution and more accurate spatial data that can be used to map ES. There will often be readily accessible high resolution climate, topography, soil, biodiversity, land use and land cover data. Data acquisition will be at relatively low cost and data may even be available free of charge when governments commit to open data policies.

In locations where there is a lack of ecosystem service data, it will be necessary to fill data gaps with alternative approaches such as remote sensing, participatory mapping, land use proxies and/or use of lower resolution global-scale datasets. Creative ways to fill data gaps are needed when ecosystem service mapping projects have limited resources to collect new data and build new models.

The rapid growth in ES research and implementation risks being undermined by poor data management and mapping practices. There is a recognised inconsistency in ecosystem service modelling and mapping methods which limits the use of ecosystem service information in national accounts and policy decision-making related to the environment. A basic set of metadata should be recorded during every ecosystem service quantification and mapping study. For example, information about the mapping study, such as purpose, location, duration, administrative unit mapped, citations and project investigators should be recorded and published with the maps. For each ecosystem service modelled/mapped, attributes such as ecosystem service indicator, data source, quantification unit and method, scale, extent, resolution, time period and beneficiary definition should be recorded on a blueprint. Completing metadata and blueprints for ES quantification/mapping will provide users of the data and maps with a confidence in the pedigree and usefulness of the information.

Producing metadata and blueprints as part of a mapping exercise provides a level of standardisation of the data for easy inclusion in catalogues such as the ESP Visualisation Tool (ESP-VT). The ESP-VT is an online web portal and catalogue for uploading, downloading and querying spatial information on ES (Chapter 7.9). Another important cataloguing tool is the Intergovernmental Platform on Biodiversity and Ecosystem Services (IPBES) Catalogue of Assessments. The IPBES Catalogue aims to build a global database of studies of ecosystem service quantification and valuation.

Double-counting is an economics term that refers to the erroneous practice of counting the value of goods or services more than once. Double-counting of ecosystem service values arises, for example, when supporting or intermediate ES such as soil formation, nutrient cycling and photosynthesis are valued and mapped in conjunction with the valuation and mapping of final ES. The problem of double-counting also occurs when there is overlap between ES because of vague service definitions and categorisations and/or limited understanding of ecosystem functions and processes. Recent classification systems such as CICES and the US EPA's FEGS-CS have taken considerable care to ensure final ES are clearly categorised to minimise the likelihood of double-counting in valuation and mapping.

Brown G, Fagerholm N (2015) Empirical PPGIS/PGIS mapping of Ecosystem Services: A review and evaluation. Ecosystem Services 13: 119-133.
Crossman ND, Burkhard B, Nedkov S, Wil- lemen L, Petz K, Palomo I, Drakou EG, Martin-Lopez B, McPhearson T, Boyanova K, Alkemade R, Egoh B, Dunbar M, Maes J (2013) A blueprint for mapping and modelling Ecosystem Services. Ecosystem Services 4: 4-14.
Drakou EG, Crossman ND, Willemen L, Burkhard B, Palomo I, Maes J, Peedell S (2015) A visualisation and data-sharing tool for ecosystem service maps: Lessons learned, challenges and the way forward. Ecosystem Services 13: 134-140.
Fu B-J, Su C-H, Wei Y-P, Willett IR, Lu Y-H, Liu G-H (2011) Double counting in Ecosystem Services valuation: causes and countermeasures. Ecological Research 26: 1-14.
Maes J, Crossman ND, Burkhard B (2016) Mapping Ecosystem Services. In: Potschin M, Haines-Young R, Fish R, Turner K (Eds) Routledge Handbook of Ecosystem Services. Earthscan, UK, 188-204.
Salmivaara et al. (2015) Exploring the Modifiable Areal Unit Problem in Spatial Water Assessments: A Case of Water Shortage in Monsoon Asia. Water 7: 898-917.
Willemen et al. (2015) Mapping and valuing ecosystem services in South Africa, Tanzania and Zambia. Final report to UNCCD, consultancy CCD/15/GM/03.
Wong D (2009) The modifiable areal unit problem (MAUP). In Fotheringham, Stewart A, Rogerson P. The SAGE handbook of spatial analysis. Los Angeles: Sage 105-124.

Different ecosystem services (ES), as well as different mapping purposes, require different quantification and mapping approaches. Although there is increasing knowledge (Chapters 2 and 3) and a high diversity of methods and tools ready to be applied (Chapter 5), several ES remain problematic to map. This can be repeatedly found in related study reports, ES mapping reviews or other publications. However, integrative trans-disciplinary ES assessments will provide maps that are applicable for diverse purposes (see Chapter 7).

The aim of this chapter is to present and to discuss those ES which have been shown to be problematic to map. Related issues can be simply grouped into lack of knowledge and inherent uncertainties, conceptual questions, unclear ES spatial or temporal identification and localisation and specific technical mapping questions (see also Chapter 6.1).

Within this chapter, we want to share knowledge of ES that are problematic to map, to contribute to a better understanding of the reasons behind the problems and to show different options which can demonstrate how to deal with these problems.

Climate regulation, provision of water, food and timber, regulation of water flows and recreation are among the most mapped ES. This is indeed confirmed by several review studies which collected information on indicators for mapping ES. Less or even no mapping is observed for genetic, ornamental and medicinal resources, biological control, life cycle maintenance and gene pool protection and for cultural ES other than nature-based recreation and tourism and aesthetic beauty.

Cultural ecosystem services

As for cultural ES, it is fairly evident that virtually all focus has gone to mapping recreation in nature and to aesthetic beauty of the landscape. In addition, mapping of emblematic habitats and species can profit from spatial data with different sources (species occurrence and citizen science; see Chapter 5.6.3). Intellectual, spiritual or symbolic interactions with nature are much harder to map, though not impossible. Key issues with the intellectual and representative human-environmental interactions (including scientific interactions, heritage, cultural entertainment, aesthetic, symbolic, sacred and/or religious, existence and bequest values) are related to their high subjectivity and dependence on socio-cultural system settings. Therefore they are difficult to indicate, quantify and map.

In this section, we illustrate a generic approach for mapping cultural ES, based on a methodology which is used for mapping nature-based recreation. Figure 1 maps two cultural ES, based on a mapping of the recreational opportunity spectrum (ROS). The ROS approach brings together two sources of information: the recreation potential of ecosystems (measured using, for example, data on nature reserves, bathing water quality, ecosystem degradation) and the accessibility of this potential for people (e.g. roads, infrastructure, distance to populated areas). In a similar manner, other cultural ES can be mapped.

Figure 1.

Mapping cultural ecosystem services (nature-based recreation activities and cultural heritage) based on mapping service provision versus access to the service. Service provision is expressed in three ordinal classes (low, medium and high) while access to the service is approximated by proximity (expressed in three ordinal classes: far, proximal, near). The case study relates to Campo dei Fiori, a regional park in Lombardy (Italy).

By using information on other values or by participatory mapping approaches (Chapter 5.6.2), the potential for ecosystems to provide a suite of cultural ES including education, inspiration or spiritual experiences can be mapped. In Figure 1, a similar approach was used to map cultural heritage in a regional nature reserve: different levels of service provision (low, medium, high) are cross-tabulated with different levels of proximity (far, proximal, near) resulting in nine different classes which are mapped. Mapping different cultural services, instead of focussing on recreation and tourism, is relevant for planning. Such mapping exercises may be eye-openers for decision-makers and increase ownership and legitimacy of an ecosystem-based approach to solving problems related to spatial planning.

Mapping demand for ecosystem services

Demands for many regulating ES are also not easy to define or to map (see example in Box 1). Demands and preferences for micro and regional climate regulation and related benefits can, for example, be highly individual and specific. Respective indicators often quantify temperature amplitudes or deviations of precipitation, wind or evapo-transpiration compared to surrounding areas or standard values. We are aware that regulating ES demands and related perceived human benefits may differ considerably.

For global climate regulation, ES benefits refer to non-desired temperature changes, storm events or coastal hazards. The service providing areas SPA (for example, the large forest belts) can be mapped at specific locations, whereas the service benefitting areas SBA (Chapter 5.2) are of global extension (see Figure 2).

Figure 2.

Example of global climate regulation with regional SPAs (service providing areas) and global SBAs (service benefiting areas).

Box 1. Pollination

Pollination and seed dispersal are very prominent examples of a regulating ES that is mapped frequently. The final goods would be the fruit or flower to be consumed or enjoyed by humans, whereas the pollination process itself (the pollen-transfer) could be seen as an ecosystem function or intermediate ES (Chapter 2.3), providing the base for the actual ES supply.

In order to map pollination, often potential habitats for pollinators, species numbers or the amount of pollinators are used as proxies to indicate the actual pollination function (see also Chapter 6.3). However, we need to be aware that habitats or the occurrence of certain species belong to biophysical structures and processes within the ES cascade (Chapter 2.3). Demand for pollination services can be mapped based on the amount and location of agricultural, garden or wild plants demanding pollination.

Figure 3.

Pollination by wild animals is very important for the delivery of several ES.

For provisioning ES demand-mapping, the problem of often highly complex and globalised supply-demand patterns occurs for many goods and services. For goods that are based on multiple ES (e.g. a chair made of wood, fabric, metal), the question arises where to localise the demand for the particular services and where to map their supply. A solution can be to consider the place of the last contribution of an ecosystem as the place of supply (e.g. the forest) and the place of final use by the consumer (sitting on a chair) as the demand area or SBA. Complex trading schemes of goods, which involve several retailers, resellers and distributors along the path from the ecosystem to the final consumer are adding to the issues of mapping provisioning ES supply-demand patterns.

For many cultural ES, the question about whether the benefits contributing to individual well-being should be located i) directly at the place where the service is provided (e.g. a SPA in the form of a good beach used for recreational activities), ii) at the home of the beneficiary (i.e. the place where she/he spends most time of the year), or iii) at both sites. All three options make sense but are related to uncertainties and may lead to misinterpretations.

Spatial or temporal questions

A single map has two spatial dimensions and is static, so it is not very useful to show temporal changes. Yet, the environment and ecosystems exist in three dimensions in space and often undergo highly dynamic changes (fourth dimension). This contrast brings about particular challenges for mapping which we illustrate here for certain ecosystems and their services.

Most ecosystems can be relatively well mapped and spatially separated. Forests, grasslands or wetlands obviously occur in three dimensions but it is relatively straightforward to map them and assign specific ES to them. Often ecological processes in terrestrial ecosystems follow seasonal cycles related to primary production so that annual averages can be calculated and attributed to these ecosystems and, hence, to ES maps (Chapter 5.3).

Marine areas are more complex to map due to their three-dimensionality, water current dynamics (especially in tide-influenced waters) and the significantly different components they include. One solution could be to produce ES maps per ecosystem type and per service: one for the water surface (relevant, for example, for cultural ES, transport, energy), another for the water column (e.g. for nutrition, energy, mediation of flows, maintenance of physical, chemical, biological conditions) and one for the benthic habitats and the sea bottom (e.g. materials, nutrition, mediation of flows).

Groundwater represents a special case as it challenges both the representation of ecosystems in maps (often based on land cover data) and the typical classification systems for ES. Groundwater ecosystems are vital providers of water for drinking and non-drinking purposes. In two dimensions, they spatially overlap with all other ecosystems and processes in groundwater layers sometimes take place in decades, if not longer. Several questions emerge with respect to groundwater as an ES: which ecosystems are the providers of groundwater ES, where to localise the supply and to what extent is the provision of groundwater for drinking or non-drinking purposes an "ecosystem" service (see Box 2 for further details)?

Box 2. Groundwater

The easiest and most pragmatic approach is to assign the groundwater ES to the ecosystems or land cover types lying above the ground.

There is also an ecological rationale for doing so. The generation of groundwater ES is mediated by ecological processes which take place in forests or agricultural ecosystems where vegetation influences the re-charge of groundwater layers beneath them. In addition, part of the groundwater is in the root zone of these ecosystems whereas other, deeper, groundwater layers are part of the abiotic crust of the earth. These aquifers also often receive groundwater from above but are mainly abiotic depositories of groundwater generated elsewhere (in spite of the presence of some biotic organisms). Above-ground, water supply based on groundwater only often occurs as discrete local phenomena such as springs, dwells or water taps. Transportation installations such as water pipes or canals lead to a spatial decoupling of SPAs and SBAs (Chapter 5.2).

A second approach considers groundwater as a separate ecosystem and accounts for the specific ES delivered by groundwater. Groundwater abstraction for different user purposes is assigned to this ecosystem type and not to the above-ground ecosystem where the abstraction takes place but this involves working with multiple maps to avoid overlap.

A third approach considers groundwater as a subsoil asset or as system which delivers abiotic flows.

It is difficult indeed to always identify a clear boundary between the abiotic and ecosystem components of natural capital. Water which also comes through in the treatment of groundwater in different related pilot studies is a key example in this regard. A guiding question for classifying natural capital components into abiotic or ecosystem elements needs to address whether or not a given component is primarily shaped or maintained by biological organisms and their interaction with the abiotic environment.

Figure 4.

Example provisioning ES water supply based on groundwater or surface water (based on http://groundwater.sdsu.edu/).

A similar question can be addressed when considering soil and soil-related ES. Soil is an important part of our natural capital and soil science is a well-developed discipline with a great deal of information available in soil maps. Prominent ES delivered by the soil are erosion control and, obviously, soil formation and composition. The first two approaches also apply here when accounting for soil ES: either they are assigned to the ecosystem they support (e.g. forest, cropland, or grassland) or they are considered as a separate soil system overlapping with other ecosystems. Both approaches are possible depending on the context and the purpose of the study.

Summary tables

The following tables give an overview of selected ES, common mapping problems with them and suggested solutions.

Table 1.Download as CSV 

Selected regulating ecosystem services (ES), mapping problems and suggested solutions.

Regulating ES examples	Possible problems	Solutions
Global climate regulation	Focus on carbon dioxide and climate change only	Integrative mapping approaches (Chapter 5.6)
	Focus on carbon sequestration only	Combination of carbon sequestration and storage (i.e. in soils)
	Large difference between SPA and SBA	Overlay or intersect SBA and SPA maps (Chapter 5.2)
Regional climate regulation	ES demand relates to certain human preferences of ecosystem states	Relate to standard, critical or legitimate values
Ventilation and transpiration	Human preferences or defined (critical) levels
Hydrological cycle and water flow maintenance	Human preferences of "constant water flow" (flood and drought prevention)	Combine with respective ES demand maps
Soil formation and composition	Soil formation is a very slow process	Consider long-term effects (Chapter 5.3)
	Strong overlaps with ecosystem functions	Distinguish between natural processes, external inputs and avoid double-counting
Mass flow regulation (e.g. erosion control)	Avoided events need be assessed	Model-based approach with and without ES supplied (Chapter 4.4)
Flood or storm protection
Pollination	Strong overlaps with ecosystem functions, high potential of double-counting	Separate between potential use and actual use (see Chapter 5.1) Little concern about double-counting when mapping single ES
Pest and disease control
Mediation of waste, toxics and other nuisances
Maintaining water conditions

Table 2.Download as CSV 

Selected provisioning ecosystem services (ES), mapping problems and suggested solutions.

Provisioning ES examples	Problem	Solution
Cultivated crops, reared animals, aquaculture	High impact of external anthropogenic system inputs	If possible, include indicators of human contributions and external environmental effects
	Complex ES supply chains	In absence of better data, use spatially explicit crop and stock statistics for mapping
	Animal stables and aquafarms = locally discrete point SPA
Biomass-based energy sources	E.g. maize, rape: final use not always clear	Combination with respective ES demand
Timber	Temporal: long growth phases	Distinction between ES potential and flow (Chapter 5.1)
Groundwater for drinking/ non-drinking purposes	Subsurface SPAs; SBAs delocalised and point sources (wells, pipes)	See Box 2 in this chapter

Table 3.Download as CSV 

Selected cultural ecosystem services (ES), mapping problems and suggested solutions.

Cultural ES examples	Problem	Solution
Physical and experiential interactions	Different and subjective preferences	Participatory GIS (Chapter 5.6.2), Citizen science (Chapter 5.6.3)
Intellectual and representative interactions	ES supply and demand difficult to indicate, quantify and localise	Applying the ROS approach (see this Chapter) can help raising awareness Participatory GIS (Chapter 5.6.2)
Spiritual and/or emblematic interactions

Although not all ES can readily be mapped, we still argue that an inclusive approach is to be preferred. Too often, ecosystem assessments focus on selected ES only such as food production, climate regulation, or water quality regulation using lack of knowledge, data gaps and conceptual problems as an argument for justifying a limited perspective.

However, the strength of ES as a concept is that it offers scientists, planners and decision-makers a frame that encompasses all benefits we receive from nature. We should therefore act accordingly. The ES matrix approach (see Chapter 5.6.4) is a first valid alternative to map and assess problematic ES and put them in the whole picture. The combination of using land use and land cover data and expert- or evidence-based ranking of potential supply and demand as proxies has been successfully applied at various spatial scales and for different purposes. Several chapters of this book provide more detail or use matrix-mapping for different applications.

For cultural ES, we advance here the approach based on mapping nature-based recreation. Mapping cultural ES inevitably depends on quantifying an information flow from nature to people. Mapping the recreational opportunity spectrum involves combining ecological data with socio-economic information such as demographic statistics and the location of infrastructure and can serve as an example for mapping other cultural ES.

More efforts could be applied to the use of biodiversity data for mapping ES. A recent study used species occurrence data from GBIF, the global biodiversity information facility, to map wild food in Europe. Species are the basis of ecosystems and thus the main service providing units for several ES (see also Chapter 2.2). Linking occurrence data with trait information will be key to mapping those ES with a strong connection to biodiversity such as pollination.

Importantly, high resolution mapping is a solution to several conceptual and technical problems. This is particularly evident in heterogeneous landscapes with a mixture of cropland, semi-natural vegetation and forest.

Quantifying ES such as food production, pollination, or maintenance of soil quality often leads to questions about double-counting. This arises as a result of the latter two regulating ES contributing to the former service. A detailed mapping of cropland with spatial delineation of the semi-natural vegetation such as hedges, forest patches and grass strips also allows spatially segregating areas that provide regulating ES from areas that are dedicated to production. Moreover, double-counting mainly occurs when different ecosystem functions, services and benefits are finally aggregated to one single number (such as the total economic value TEV). Compiling maps of individual ES helps to avoid double-counting and highlights the vital role of regulating services and ecosystem structures.

Burkhard B, Kandziora M, Hou Y, Muller F (2014) Ecosystem Service Potentials, Flows and Demands - Concepts for Spa- tial Localisation, Indication and Quanti- fication. Landscape Online 34: 1-32.
Crossman ND, Burkhard B, Nedkov S, Wil- lemen L, Petz K, Palomo I, Drakou EG, Martin-Lopez B, McPhearson T, Boyano- va K, Alkemade R, Egoh B, Dunbar M, Maes J (2013) A blueprint for mapping and modelling ecosystem services. Eco- system Services 4: 4-14.
Grizzetti B, Lanzanova D, Liquete C, Reynaud A, Rankinen K, Hellsten S, Forsius M, Cardoso AC (2015) Cook-book for water ecosystem service assessment and valuation. EUR 27141. Publications office of the European Union, Luxembourg.
Haines-Young R, Potschin M (2010) The links between biodiversity ecosystem services and human well-being. In: Raf- faelli D, Frid C (Eds) Ecosystem Ecology: A New Synthesis. Cambridge University Press, Cambridge, 110-139.
Martinez-Harms MJ, Balvanera P (2012) Methods for mapping ecosystem service supply: a review. International Journal of Biodiversity Science, Ecosystem Services & Management 8: 17-25.
Villamagna AM, Angermeier PL, Bennet EM (2013) Capacity, pressure, demand, and flow: A conceptual framework for analysing ecosystem service provision and delivery. Ecological Complexity 15: 114-121.
Paracchini ML, Zulian G, Kopperoinen L, Maes J, Schagner JP, Termansen M, Zandersen M, Perez-Soba M, Scholefield PA, Bidoglio G (2014) Mapping cultural ecosystem services: A framework to assess the potential for outdoor recreation across the EU. Ecological Indicators 45: 371-385.
Schulp CJE, Thuiller W, Verburg PH (2014) Wild food in Europe: A synthesis of knowledge and data of terrestrial wild food as an ecosystem service. Ecological Economics 105: 292-305.

Many ecosystem services (ES) are very difficult or even impossible to measure. While the amount of crops produced is monitored at a detailed level in the EU, it is less straightforward to quantify how many potential floods have been avoided and to what extent functioning of the ecosystem renders this flood risk mitigation. Consequently, ES are commonly mapped based on a combination of a limited number of measurements, expert-based or empirically derived proxies and model-based mapping procedures. For example, the capacity of the landscape to sequester CO₂ has been mapped through first measuring CO₂ sequestration rates in different ecosystems, which are then upscaled by combining average sequestration rates per ecosystem with a map delineating these ecosystems.

In these modelling or upscaling approaches, as well as in the underlying measurements, uncertainties arise. First, measurement equipment is not 100 % accurate and people who measure environmental variables can make mistakes. A common method to measure soil organic carbon stocks clearly illustrates this situation. In the "loss on ignition" method, a soil sample is first dried in an oven to remove all soil moisture and then weighed. Next, the sample is placed in an oven at over 400 °C for 24 hours to burn all organic matter. The sample is then weighed again and the weight difference represents the amount of organic matter. This is translated into organic carbon using a conversion factor of (1/1.72) g carbon per g organic matter. Both weighing procedures have a small error, resulting in an uncertainty in the amount of organic matter. The conversion factor from organic matter to organic carbon is well established, but differs depending on the origin of the organic matter, meaning that the factor is uncertain as well. Altogether, the loss on ignition method has an uncertainty of approximately 2 %, meaning that, when a soil organic carbon stock of 5 % is reported, the value can actually vary between 3 % and 7 %.

In many approaches for mapping ES, such measurements are then coupled to maps of land cover or environmental variables such as soil maps or elevation maps to create an ecosystem service map. Land cover maps are commonly derived from remote sensing imageries. These maps are uncertain with regard to the location or shape of objects and with regard to the characteristics of mapped objects. Uncertainty on the shape or location of objects is called geometric uncertainty and is a result of the spatial resolution of the data. While high-resolution remote sensing imagery such as the 20 m SENTINEL products are able to capture small land cover patches and land cover types with a limited cover, upon a coarser resolution, such features get lost. Neither linear landscape elements like hedgerows, ditches and tree lines nor individual trees can be captured even with a 20 m resolution. For several ES, such landscape elements are essential for the supply, meaning that the inability to capture them limits the possibility of satellite-derived land cover data in mapping ES.

Uncertainty on the attribute values is called thematic uncertainty and arises when classifying the reflectance signature into a land cover classification. Thematic uncertainty implies that when a land cover map displays grassland, there is, for example, a 90 % probability that there is actually grassland on that specific location while there is a 10 % probability that, in reality, another land cover type is present.

Mapping of other biophysical variables which are used as input to ecosystem service maps exhibit additional uncertainties due to upscaling of measurements. This includes simplifying the continuous variation of soil characteristics into soil types, or inaccuracies in measuring elevation. An additional source of uncertainty for mapping ES is that, due to data availability limitations, inputs from a range of different years are often used.

Additional to these "technical" uncertainties, conceptual uncertainties may arise as well. Most ES can be defined in different ways. This is related to the understanding of the processes that ensure the service supply and to the aims of the mapping study.

When mapping an ecosystem service, many different choices can be made on the indicator used, the data and methods used to quantify the indicator and the final presentation of the indicator. When a set of maps and measurements, each with an uncertainty range as a result of these technical and conceptual limitations, is combined into an ecosystem service map, the uncertainties in the input propagate into uncertainty of output in ecosystem service maps.

On a global scale, the production of 35 % of the food crops depends on pollinators. Both managed honeybees as well as wild bees are important for pollination. Several crops are exclusively pollinated by wild pollinators while, for many other crops, wild pollinators significantly contribute to the yield quantity and quality. This is a frequently mapped ecosystem service and the approaches available for mapping clearly demonstrate the source and impact of conceptual and technical uncertainties that arise when mapping ES in general.

Mapping an ecosystem service basically involves, firstly, selecting an indicator to quantify the service; next gathering spatial and non-spatial input data in an iterative manner along with defining the model to quantify the service; and finally, applying the model to the data.

Each step introduces specific uncertainties.

The ecosystem service "pollination" can be quantified using different indicators (Table 1).

Table 1.Download as CSV 

Overview of indicators for the ecosystem service "pollination".

Category	Definition
Landscape-based indicators	Capacity of the landscape to support pollinator communities
	Percentage area of potential pollinator habitat
	Distance to pollinator habitat
	Probability that a location is visited by pollinators
Species-based indicators	Abundance of pollinators
	Abundance of specific pollinator species
	Species richness of pollinators
Crop-based indicators	Yield quantity
	Financial benefits of yield of pollinator-dependent crops
	Percentage yield loss upon absence of pollinators

Which indicator to choose depends on the scale and aim of the study. For a detailed, small-scale study that aims to assist with the planning of green infrastructure in a specific agricultural landscape, it may be relevant to consider which crops are typically grown in the landscape, which pollinators are important to those crops and what are the additional habitat requirements for those species. Here, indicators based on species composition or on abundance might be the most appropriate. ES are often mapped on a national or continental scale to support national or EU policies. At such a large scale, abundance is often not feasible due to lack of data to calibrate or validate the required models and the variation in space and time of crops grown makes it less relevant to distinguish specific pollinator guilds. Therefore, more generic measures such as landscape composition are used. Several large-scale pollination maps are based on the presence of suitable habitat for pollinators and the distance to these habitats in croplands.

A key parameter for pollination services is the distance between a pollinator habitat (nesting site) and the crop which needs pollination. For calculating the distance to pollinator habitat using a land cover map, one should decide whether each land cover type provides habitat or not. This introduces new uncertainties. While for many individual pollinator species, habitat requirements are known, these often do not match the level of detail displayed in land cover maps. One can only assume that the specific vegetation type and structure or host plant which a pollinator community requires, is present in a land cover type that is only described with a level of detail like "natural grassland" or "deciduous forest". Secondly, a single land cover type can include both suitable and unsuitable areas, dependent on the types of vegetation included in the land cover or dependent on the management. Based on the knowledge of pollinator behaviour and of the land cover map, the map maker has to distinguish between presence or absence of a habitat. Depending on background knowledge and the exact nature of the uncertainties described above, different map makers can decide differently in similar situations. This introduces a new uncertainty.

More uncertainty arises when calculating the distance to habitat, depending on the resolution of the data. Each pixel gets assigned one distance value which, especially when using coarse resolution data, can deviate. In addition, land cover data often fail to properly represent small landscape elements such as tree lines, hedgerows or individual trees. An alternative approach that can help overcome this is, for example, using a map of the density of such elements. But deriving distance to habitat from a density map also introduces a new uncertainty.

Relations between distance to habitat and the effectiveness of animal pollination have been previously established. Close to natural habitat, bee abundance and species richness tend to be high while richness and abundance decrease upon increasing dis