Assessing landscape services as foundation for Green Infrastructure functionality: the case of the Wienerwald Biosphere Reserve

Biosphere Reserves are considered as means for the people who live and work within them to attain a balanced relationship with the natural and semi-natural environment. Moreover, they contribute to the needs of society by showing a way to a more sustainable future. The Wienerwald Biosphere Reserve partly surrounds the city of Vienna and other minor settlements, representing a well-developed example of Green Infrastructure (GI) of great cultural and natural value. Its heterogeneous landscape offers a variety of landscape services (LS). In this work, we quantified and mapped the capacity of LS offered by the open land elements of Wienerwald. Starting from a high-resolution dataset, we selected suitable indicator classes, and scored each ecological and socio-cultural service through an expert-based capacity matrix. The subsequent analyses with Geographical Information Systems (GIS) focused on the intensity and density of LS capacities by developing an index useful for mapping GI functionality. The work provides an effective monitoring tool for the Reserve’s both ecological and socio-cultural sustainability performance. It also allows detecting resilient areas, by considering both the spatial distribution and the abundance of landscape elements.


Motivation
It is nowadays globally recognized that biodiversity sustains human life by means of the so-called ecosystem services (e.g. Isbell et al. 2015). Ecosystem services are defined as 'the conditions and processes through which natural ecosystems, and the species that make them up, sustain and fulfil human life' (Daily 1997). They arise when a biophysical structure (e.g. vegetation cover) or function (e.g. slow passage of water) directly or indirectly contributes towards meeting a human need or demand. Such services (e.g. flood protection) generate benefits (e.g. serving health and safety) that contribute to overall well-being and can be valued by people (e.g. willingness to pay for protection) (Haines-Young et al. 2010;MEA 2005). In this study we address a variation of the concept of ecosystem services, i.e. landscape services. This means defining functions, services and benefits at landscape scale to integrate the concept into land management decisions (Bastian et al. 1999;de Groot et al. 2010;Willemen et al. 2010). Landscape services are the contributions of landscapes and landscape elements to human well-being (Bastian et al. 2014), and they include potentials, materials and processes of the nature (e.g. raw materials, biomass, biodiversity etc.) and services of cultural elements and constructions that come into being through human creation (e.g. buildings, settlements, infrastructure etc.) (Konkoly-Gyuró 2011;Hermann et al. 2011). Important reasons to consider landscape services include the prominent role of spatial aspects, the reference to spatial elements and to the landscape character, and the relevance of landscape services for spatial planning (Bastian et al. 2014). The pattern of multi-functional landscapes is the basis for interactions, synergies and conflicts between landscape elements (Willemen et al. 2012). Moreover, the provision of services does not always depend on the properties of an ecosystem patch, but rather on the spatial interaction among these patches (Termorshuizen et al. 2009). Last, as local people define their environment more as a "landscape" than as an "ecosystem" the term "landscape services" is preferred as a specification (rather than an alternative) of ecosystem services (Termorshuizen et al. 2009). Landscape services can be supplied by those landscape elements comprised in the so-called Green Infrastructure (GI), such as areas of high biodiversity value, land managed in a sustainable fashion, green urban and peri-urban features (parks, gardens, small woodlands, cemeteries and the like), but also artificial connectivity features, such as green bridges over road corridors, tunnels underneath transport corridors and fish passes where natural migration/ movement is hindered by development (Mazza et al. 2011). In the European Commission communication (2013) GI is defined as 'a strategically planned network of natural and semi-natural areas with other environmental features designed and managed to deliver a wide range of ecosystem services.' To provide benefits to society, GI shall be adequately planned and maintained (European Commission 2016). GI is in fact an approach that brings together both the need for strategic planning of green and open spaces and the science of landscape services (European Commission 2011). It promotes the multifunctional nature of space and the benefits that appropriate management approaches can deliver. As GI recognises and promotes the multifunctional nature of green and blue spaces and is underpinned by the science of landscape services, it has a natural affinity with the commonly accepted three pillars of sustainable development: society, economy and the environment (Purvis et al. 2018).
Biosphere Reserves are protected areas where people who live and work within them seek to attain a balanced relationship with the natural and semi-natural environment (UNESCO 2002). Moreover, they contribute to the needs of society by showing a way to a more sustainable future. They provide an example of an integrated sustainability framework, which explicitly acknowledges that complex socio-economic and ecological systems are inextricably linked (Levrel et al. 2008). For these reasons, they have a primary role in coupling nature conservation practices with sustainable socio-cultural development (Lotze-Campen et al. 2008). Being a living laboratory for the fruitful coexistence of human activities and nature protection, they may represent a suitable example of how to plan GI so to maximize the benefits to humans without depleting natural resources.
Since Biosphere Reserves are seen as models for sustainable development (UNESCO 2002), they need a rapid toolset for the assessment of their sustainability over time. The sustainability performance can be expressed by an assessment of GI multi-functionality, which implies that the whole range of landscape services, from socio-cultural to ecological, is simultaneously evaluated. Examples of benefits supported by GI through its landscape services are health and well-being, enhanced efficiency of natural resources, water management, tourism and recreation, conservation benefits, climate change mitigation, and resilience (European Commission 2013). Despite the recognition at EU level of the pivotal role of GI in meeting the EU 2020 Biodiversity Strategy's targets (European Commission 2013), the concept is yet not so well established in national, regional and local planning. This is probably due to the still ongoing difficulty in interpreting the term GI in a univocal way (John et al. 2019). Also, Slätmo et al. (2019) stressed that GI in spatial planning needs to cover many different policy sectors and that its implementation is an on-going process dependent on political willingness. Consequently, tools for implementing the assessment of the multi-functionality of landscape elements are still under progress. Examples of development of toolsets for the assessment of GI multifunctionality include the combination of spatial data with the knowledge of both experts and regional and local actors (Kopperoinen et al. 2014), the creation of performance indicators of GI (Pakzad and Osmond 2016), and the use of field questionnaire surveys to explore the perceived benefits (e.g. Qureshi et al. 2010). Nevertheless, a holistic LS point of view to address the evaluations is rarely employed.
This study aims at filling the above-mentioned gaps by providing a framework for the rapid assessment and mapping of the capacity of all the LS offered by the open land landscape elements, taking the Wienerwald Biosphere Reserve (AT) as pilot study area. The Wienerwald Biosphere Reserve is an important part of an international GI network of protected areas, but it also comprises landscape elements of natural and semi-natural areas forming a local GI offering a vast range of landscape services. Since Wienerwald partly surrounds the city of Vienna (AT), it represents an excellent training ground for developing a well-founded basis for the sustainable planning of GI in the peri-urban and rural areas around Vienna.

Goals of the study
The main objective of the study was the development and provision of technical and methodological framework for the regional assessment and mapping of landscape services provided by the open land landscape elements of the Wienerwald Biosphere Reserve, that would contribute to a replicable monitoring of the Biosphere Reserve's GI functionality performance.
In order to achieve our goal, we set out to answer the following research questions: (i) which landscape elements are suitable for representing the landscape services of the open land taking into consideration the availability of spatial data; (ii) and how to simultaneously assess and spatially represent ecological and socio-cultural landscape services.

Study area
The Wienerwald Biosphere Reserve (designated as Biosphere Reserve by UNESCO in 2005) encompasses an area of 105 645 hectares within the Austrian provinces Lower Austria and Vienna ( Figure 1). Influenced by the easternmost part of the Alps, it is characterized by a hilly terrain (sea-level from 160 m up to 893 m). Geologically the Biosphere Reserve can be divided into the northern part consisting mostly of flysch rock ("Sandstein-Wienerwald") and the southern part, primarily limestone ("Kalk-Wienerwald") that appears near the so-called "Thermenlinie". As a stepping stone within the biotope network the Wienerwald is of superregional importance (Reimoser et al. 2008;Reimoser et al. 2012 ure 1), which corresponds to an area of 27 831.5 hectares within the Biosphere Reserve. Due to its high potential for a wide spectrum of services, the forest was included by integrating the forest edges into the analysis. Furthermore, since the focus was more on the pull factors of recreation users within the open land, rather than push factors, settlements were excluded from the analysis. A difference in age and accuracy of the geodata underlying the project analyses required a quality-based selection of data within the south eastern region along the border of the Wienerwald Biosphere Reserve. On these grounds, part of the south eastern region of the Biosphere Reserve was excluded (about 3 000 hectares -2.8% of the overall Biosphere Reserve area), fact that shall be considered in the valuation of the regional assessment of the landscape services.

Definition of landscape services
We defined the landscape service classification based on the definitions developed by de Groot (2002;2006) Groot (2002;2006) includes 25 landscape services distinguished into five service categories: Regulation, Habitat, Provisioning, Carrier and Information (Table 1). We adapted the list of landscape services definitions to the properties of the study area, as follows. The Raw material service includes the provision of sand and gravel e.g. by periodic brooks. The Cultivation service refers only to the provision of substrate for the cultivation of food or fodder, not for ornamental cultivation (e.g. gardening). Furthermore, in comparison to the service Food, it refers to cultivated fields (producing crops and fodder), vineyards and orchards as well as managed meadows (fodder production). Consequently, the service Food refers to all food produced for human use (wild and cultivated) and includes all animal related farming (aquaculture, cattle etc.). Genetic resources are extended to forestry and agriculture. Waste disposal includes the landscape elements providing wastewater disposal. We discarded the service Energy conversion, due to lack of reliable data. Regarding the service Transportation, the potential transportation on waterways was not included in the evaluation as there are no suitable waterways in the project region. Tourism facilities, as a carrier service, address exclusively transformed (man-made) landscape elements, which provide touristic infrastructure such as accommodation and gastronomy. Recreation mainly refer to natural landscape elements used for recreational purpose. Furthermore, following de Groots' inclusion of eco-tourism within the Recreation service, recreation infrastructural elements such as educational trails were also included in the service ( As an element for comparing the definitions imple-mented for each service, we decided to adopt the column "Simple descriptor" available in the classification CICES V5.1, since such descriptor provides an unambiguous and clear explanation for most of the services (Table 1). For the comparison we adopted a multiple correspondence approach. The selective, multiple correspondence approach allows for more than one CICES ecosystem service to be linked to different landscape services without the need of assigning a landscape service to every CICES ecosystem service.
The comparison of our landscape services classification with the CICES ecosystem services classification indicated that the Carrier services are underrepresented in CICES, which instead emphasises the natural environment and assigns less importance to the transformed landscape elements and to the socio-cultural aspects. On the other hand, CICES proved to be more detailed in distinguishing the Regulating and Provisioning services, like for instance Disturbance prevention, to which four simple descriptors could be linked, and Food, to which seven simple descriptors could be linked (Table 1).

Data sources and landscape elements classification
The study main data source is the open land data of the Wienerwald Biosphere Reserve, originating from a detailed mapping of open land habitat types performed in the field mapping project "Offenlanderhebung Biosphärenpark Wienerwald" (Staudinger et al. 2014) for the open land of the whole Biosphere Reserve in Lower Austria and Vienna. The geometries are based both on cadastral maps with high accuracy and remote sensing with the accuracy ranging between 10-30 m (Schranz 2018, pers. comm.). We relied on the open land data as the study prime data source and included data from other data sources solely when additional information was needed to address a service more fully or if it provided more detailed information on transformed landscape elements than the natural element-focussed open land data. All additional data was customised to the spatial projection of the open land (EPSG: 31259) using ArcGIS 10.5.1 (ESRI, Redlands). Most of the additional spatial data (from now onwards Additional Spatial Indicators -ASI) was downloaded from Open-Landscape Online -supported by the International Association for Landscape Ecology and its community   (Table B). Similarly, we aggregated the ASI into 17 classes (Table A). The elements selected as ASI were classified based on their similarity and on their service provision capacity. Since settlements were excluded from the study area, the capacity of the service Habitation was expressed through the indicator "isolated buildings". Due to the "layman" and inclusive nature of the OpenStreetMap data (Open-StreetMap Austria 2019), a deletion and ranking of the point data with congruent location and different name was indispensable. Expert decisions towards a double function of the element or towards its deletion were made. Hence part of the double-counted data was dismissed due to (i) redundancy (e.g. graveyards data was adopted from the open land class; "convenience stores" or "wind mill" were not included in the dataset); and (ii) ambiguity (the vague meaning of the name "attractions" was recessive). In case of "church" and "wayside cross" we decided in favour of the smaller element. We decided a picnic site to be an inclusive element for a bench or a waste basket. Other OpenStreetMap elements provided a double function; in these cases, both points were included in the further calculations for their double capacity (e.g. "guesthouse" and "restaurant", "bench" and "viewpoint", "restaurant" and "viewpoint", "church" and "artwork", "hiking" and "cycling trail").

The reclassification of the open land habitats into 62
open land classes and the selection and classification of ASI into 17 classes led to a joint collection of 79 indicator classes (including altogether 58 202 landscape elements), which, in a next step, were inserted in a capacity matrix for the landscape service capacity assessment.

Capacity matrix: assessing service capacities
This study focuses on landscape service capacity only, similarly to many of the currently available spatial ecosystem and landscape service studies (for instance Crossman et al. 2013;Martínez-Harms et al. 2012;Egoh et al. 2012;Kopperoinen et al. 2014). Service capacity can be here defined as the hypothetical maximum yield provided by a service (Burkhard et al. 2012;Burkhard et al. 2014). Assessing the actual capability of ecosystems to provide services for human well-being needs information about their current conditions, which are induced by human activities (Burkhard et al. 2017). In this sense, nor the actual service capacity, neither the used stock of services is taken into account in this study. Other authors distinguish between ecosystem properties, potentials and services (Bastian et al. 2012), implying that ecosystems provide a certain potential to supply services based on their functioning (van Oudenhoven et al. 2012).
To define the LS capacity, we employed here the socalled capacity matrix. A capacity matrix links service providing units (definable at different spatial scales) to service supply capacities (Burkhard et al. 2009). In an assessment based on a capacity matrix, for each service providing unit a ranking proportional to the Landscape Online -supported by the International Association for Landscape Ecology and its community Drius et al.
Landscape Online 84 (2020) -Page 9 capacity for each service is assigned. Generally, expert evaluations are employed in order to gain an overview and see trends for ecosystem service assessments (e.g. Burkhard et al. 2009;Scolozzi et al. 2012). In subsequent analyses, the expert evaluation values could successively be replaced by data from monitoring, measurements, computer-based modelling, targeted interviews or statistics, although these techniques imply a much longer data processing.
Following The 25 landscape services were split and assigned to the two teams of expertise (ecological and socio-cultural) involved in the study. First, both teams of experts assigned scores for their relevant services, producing a prefilled matrix. Then, through a dedicated workshop all experts contributed to the assessment of all the scores, both by discussing the scores of the services in their field of expertise and by providing comments and suggestions regarding services of the other discipline. Helpful within the workshop was the preparation of an approachable layout of the matrix in the software Microsoft Excel (2016). To ensure transparency, the matrix was projected on a screen, clearly visible for all people involved. We approached each class individually, discussing and comparing the scores vertically (to other classes) and horizontally (to other landscape services). When a consensus was necessary, we adjusted the scores directly on the screen, again aiming for high transparency. In many cases the provision of the definitions of landscape services and detailed information on the open land types were necessary, to avoid ambiguity and misunderstanding. Agreement on the definitions to describe each landscape service also helped to avoid issues of double counting, which refer rather to the service, than to the indicator. In case of similar aspects being valued within two services, values were split between the two (e.g. the ASI recreational infrastructure is valid both for the landscape service Tourism facilities and for Recreation). The possibility to examine the spatial location of the landscape elements and an orthophoto of the study area were also helpful to come to an agreement on the scores. The capacity matrix scoring gives no absolute values since the scores often refer to the specific characteristics of the Wienerwald Biosphere Reserve. Moreover, the comparison of indicators inside the area influences the score assigned. For instance, compared to dry grassland, other grassland classes have higher water supply; this means that peaks can be developed within the matrix.
After the evaluation, we linked the scores to the spatial data in Geographical Information Systems (ArcGIS 10.5.1; ESRI, Redlands), in order to obtain estimates of the capacity of landscape service supply and map them in spatially explicit units (Burkhard et al. 2009;Burkhard et al. 2012). The steps of the mapping procedure are detailed in the following sections.

Spatial data preparation and INDEC application
Simultaneously to the scoring of landscape service capacity, we prepared the spatial data so to later integrate the capacity matrix scores in the mapping process, using ArcGIS 10.5.1 (ESRI, Redlands). Figure 2 displays the workflow of the calculations, performed in Python (Python version 2.7.13). The forest-related classes of the open land dataset were transformed into polylines, since the forest was addressed as forest edge exclusively. We applied a buffer of 10 m to the data to secure the inclusion of all point and line features within or in direct proximity of the open land. By clipping the buffered data to the Wienerwald Biosphere Reserve's outline, we defined this as the outermost border of the project study area.
Due to the heterogeneity of data sources and data sets (spatial resolution and temporal variance) and the ambiguity in size for some of the ASI (e.g. streets, playgrounds, churches), we developed a methodology that would overlook both the size and the differences in data sources. The approach was based on two concepts: location of elements and the ca-Landscape Online -supported by the International Association for Landscape Ecology and its community Drius et al.
Landscape Online 84 (2020) -Page 10 pacity scores retrieved directly from the capacity matrix. The location of the landscape elements was represented by points, meaning that line and polygon data types were transformed into point data. Then, we applied an innovative tool for mapping the landscape service capacity, which we named INDEC (analysis based on INtensity and DEnsity of service Capacity). Each passage of the workflow is described in the following paragraphs.

STEP A: Transformation of each landscape element into a point feature
Regarding the polygons, we transformed them into a point coinciding with the focal points of the polygons. Subsequently, the points generated from different geometrics (point, polyline, and polygon) and sources were merged into one "point feature class".
Based on the assumption that an ideal landscape services map would show an equally distributed maximum capacity of the service (Figure 3), we computed the equally distributed area of each point data (4 871 m²) from the total study area (278 315 000 m²) and the number of landscape elements (58 202).

STEP B: Calculation of the distance of points along line features
In order to transform the polylines into points, we had to compute the point distance. In fact, no references for setting point distance could be found. For the line features the equivalent of the sum of the length of all the lines (5 027 130 m) within the area was equally distributed into an abstract quadrat of the study area size. Within this generated grid, we doubled the value of the side lengths of one grid cell (223 m

STEP C: Calculation of the buffer sizes of landscape elements
In the theoretical optimum described in the section above, each landscape element has the same maximum capacity to provide a landscape service and the elements are equally distributed across space. The calculated areas are perceived as circles (see Figure 4). The circles represent each landscape element with an equal range in all directions from one centre point, based on the assumption that the capacity of one landscape element potentially spreads equally in each direction. Therefore, the capacity of each point is represented by an outer buffer, whose radius is linked to the scores from the capacity matrix. The buffers express the potential intensity of the capacity. We computed the radius of the average circular area and adopted it as size of the maximum capacity buffer (equal to score 5 in the capacity matrix). At the maximum service capacity, the buffers (radius = 39 m) partially overlap. In this way the connectivity among the landscape services provided by each landscape element is guaranteed. For defining the intensity related to each point, we calculated the size of the buffer of the remaining capacity matrix scores (2-4) related to the maximum area (4 871 m²). For instance, 50% capacity (value 3) of maximum capacity (value 5) is equal to 50% (2 435.5 m²) area of maximum area (4 871 m²) ( Figure  4).
The spatial data preparation provided two main outcomes: (i) the transformation of each landscape element into point data; and (ii) the definition of the buffer sizes according to the scores given the capacity matrix.
After the preparation of the spatial data and the transformation to point data for all landscape elements, we proceeded with the INDEC application ( Figure 2). First, we linked the landscape elements to the scores they provide for each landscape service, taking them from the capacity matrix. In order to balance the representation of line elements (based on their nature of being represented by many points, in comparison with the polygons, which can only be represented by one point), we included a weighting factor (wf), which is based on the biggest (e.g. cultivated field ca. 400 000 m²) and smallest (e.g. single tree, a few m²) landscape element of the open land. The relative part of the maximum area within the study area became 1, and the relative part of the minimum area became 0. All the remaining landscape elements received relative values between 0 and 1, according to their area sizes (Equation 2) and were ranked accordingly.  As exemplified in Figure 5, depending on the relative part (0-1, with rp=1 equal to ranking position 1) and on the defined maximum wf, each landscape element received a transformed wf. For example, in case of rp=0.5 and wf [def as max]=5, the resulting wf [transformed]=3. Line elements also received a weighting factor independently of their length (see also Section 2.6.2).

STEP D: INDEC -final calculations
In a geoprocessing step, we buffered the landscape elements by multiplying the buffer value with the weighting factor of the line and polygon landscape elements. We merged the overlapping buffers by "spatial dissolve", obtaining clusters (dissolved shapes) representing the cumulative intensity of service capacity. Subsequently we counted the number of landscape elements within each cluster by "spatial join", using this number as an indicator of the density of service capacity. In this way we combined both intensity (the size of the cluster) and density (the number of elements within the cluster), creating our final INDEC. The INDEC itself has no unit and originates by 50% from the relative size and 50% from the relative amount of the landscape elements. All landscape service capacities are calculated through the same INDEC procedure, although for each landscape service only the landscape elements with service capacity >1 were included in the processing.
For the index of the intensity (I IN-cl-x ) of a cluster, the cluster area (A cl-x ) is divided by the normalized cluster area (A cl-norm ), which corresponds with the optimum area per landscape element (A oALE = bs max = 4 871 m²). This value refers to the value of the equally distributed area that is used to calculate the maximum buffer size and now allows for a normalisation as reference of the area size. In case the cluster size exceeds the optimum area per landscape element, it will produce a value higher than 1. In case of smaller clusters, a value lower than 1 will be produced. For clusters equal in size to the optimum area per landscape elements, the value 1 is assigned.

Exclusion of landscape services
Data availability was a limiting aspect within the study. By using different data sources, we expanded our database to provide some indicators for all the services. However, after visualizing the preliminary capacity maps, two services had to be excluded from our list of landscape services, due to incomplete representativeness: Habitation and Waste disposal.

Validation of the plausibility of the landscape services capacity maps
The plausibility of landscape services capacity maps was validated through expert knowledge of the area during an internal workshop. The preliminary maps were displayed in Google Earth Pro (Google Earth Pro n.d.), and the two groups of experts (ecological and socio-cultural) were tasked to provide information on the areas of high and low capacity and on their location within the map of each landscape service. The discussion following the validation tasks revealed three main aspects in need of refinement. First, due to the spatially scattered nature of the study area, the weighting factor was further adjusted in order to enhance the visibility of the service capacities on the regional level. On account of a plausible and balanced display of the landscape services addressed, the final weighting factor was set to "6" for polygons and "3" for polylines. Second, since forest edges appeared spatially overrepresented in respect to the other landscape elements, creating a bias in the results, they received a final arbitrary wf = "1". Third, the scoring of the capacity matrix was revised to highlight some OL and ASI classes with respect to others. Changes affected 8% of the scores of the matrix, with 5% being upgraded to a higher score, and 3% being downgraded. The services Medicinal Resources, Disturbance prevention and Nutrient regulation faced the greatest changes with adjustments of 19% and 24% of the scores, respectively.
With these refinements we finalized our method and produced the final version of the service capacity maps.

Landscape service Water supply
Spatial distribution: the service capacity is expressed through rather small clusters, with areas of highest capacity localized along streams and pools, such as south of Purkersdorf, and west of Liesing, and in the area of Pressbaum (Wienerwald See). Areas of high capacity are also visible in the dense mosaic of landscape elements south of Königstetten. The large north-eastern region of the Biosphere Reserve is dominated by cultivated fields and shows no capacity. Another area with many low, intermediate and high INDEC values is located in the north of Alternmarkt an der Triesting ( Figure 6).
Landscape elements: The classes with the highest scores according to the capacity matrix are straight, seminatural streams, meandering, seminatural streams, channel/regulated streams, seminatural, nutrient-poor lakes and pools, nutrient-rich pools, artificial standing water connected to groundwater, springs, marshes, swamps, nutrient-poor fens, riparian woodland. A score "4" was assigned to periodic/ small brooks, reed beds, forest edges, long-rotation woodland, hedgerows dominated by trees, and extensive orchards (Tables C-H).  of Liesing; it has 60 landscape elements distributed across 11 classes. The most frequent classes are nutrient-poor extensive moist meadows, semidry managed meadows, and forest edges. Cluster 3 (INDEC 56.24) is situated in the southern area of the Biosphere Reserve, close to Obertriesting, in a mosaic of managed meadows, tree rows and hedges. It includes 40 landscape elements distributed across 10 classes, whose most frequent are forest edges, tree rows and single trees, and intermediate and intensively managed meadows (Table 2).

Landscape service Pollination
Spatial distribution: the service capacity is generally very high in the reserve, and more concentrated in the north of the Biosphere Reserve, particularly on the reserve border in the proximity of Baumgarten am Tullnerfeld, Königstetten, Klosterneuburg, and Döbling in the municipality of Vienna. Other areas for Pollination can be found in the southern part of the Biosphere Reserve, in the very large mosaic of meadows and fields just north of Altenmarkt an der Triesting (Figure 7).  Other classes present in the cluster are extensive orchards, hedgerows dominated by trees, and forest edges. This cluster is located in the southern part of the Biosphere Reserve, in the proximity of Groisbach (Table 3).

Landscape service Recreation
Spatial distribution: For the LS Recreation capacities are shown throughout the area. Many high capacity areas are located in close proximity to settlements along the Biosphere Reserve, as Höflein an der Donau, Kritzendorf, Klosterneuburg and Vienna.
In the south the hotspots are slightly more isolated with locations in Groisbach, Sankt Corona am Schöpfl and Obertriesting. In the north-west high capacities are revealed along the border of the Biosphere Reserve from Sieghartskirchen and Freundorf until Wolfpassing (Figure 8).

Discussion
In our study we aimed to address a broad range of landscape services to show the diversity of benefits the landscape provides for humans.
As faced in other studies (e.g. Wrbka et al. 2012) the availability and compatibility of spatial data was a main issue. Especially the assessment of the Carrier and Information services categories strongly depend on the availability of data additional to the mapping of the open land habitat types. Therefore, it was necessary to find a way to balance data varying in sources, age and level of accuracy. Nevertheless, the dismissal of landscape services was required for those cases where the provision of sufficient data was not accomplishable. . The attractiveness of the matrix approach results from its flexibility concerning level of detail and level of abstraction from rather simple to highly complex. Its potential to integrate all kinds of data, from expert-scores to statistics, interview data, measurements or high-end model outcomes makes it applicable in data-poor as well as data-rich environments. Results based on the flexible 0-5 rank-ing system and the linkage to geo-biophysical spatial units in ecosystem service maps provide wide application ranges in science and in decision making (Burkhard et al. 2014). The method has successfully been applied to quantify ecosystem and landscape services in several case studies (e.g. Kandziora et al. 2013;Kaiser et al. 2013;Vihervaara et al. 2010;Vihervaara et al. 2012;Hermann et al. 2014;Hainz-Renetzeder et al. 2014;Stoll et al. 2015). It has also inspired the development of ecosystem service mapping studies (e.g. Clerici et al. 2014;Baral et al. 2013). On the other hand, there are several uncertainties related to the matrix method applied for landscape analyses (Hou et al. 2013), which we could experience as well. First, in this study the expert knowledge emerged from the teams cooperating in the project. Therefore, the consensus approach provided productive input and discussion, minimizing ambiguities in the definitions and improving the development of the project. Furthermore, as also pointed out by Jacobs et al. (2015) the balancing effect of the consensus method allowed to progress on decisions where uncertainty (due to lack of data or knowledge) might have been blocking the individual. Nevertheless, an individual scoring beforehand to support the possibility of statistical analysis could be considered. Moreover, expanding the number of experts involved and thus optimizing scientific credibility shall also improve the method (Campagne et al. 2018;Jacobs et al. 2015). In fact, there is a high dependence on the observer's experience, knowledge and objectivity which services are supposed to be relevant and how to value them (Burkhard et al. 2012).
Although the issue of double counting of elements was avoided by agreement on distinct definitions or a splitting on values within strongly connected services (Wrbka et al. 2012), the distinction between the assessment of a general capacity of classes/ elements while strongly referring to local/regional conditions and peculiarities of the Wienerwald Biosphere Reserve (also due to the Wienerwald explicit data of the open land habitat types) revealed contextual inconsistencies. The possibility of a direct transferability of the scores to other biosphere reserves or nature protection areas should be object of critical revision. Another limitation of the capacity matrix and of the study was that the actual condition of the landscape element and the influence of sea- sonal aspects were not addressed. In this sense the capacity score assigned wasn't adjusted with a qualifier factor, as done in other studies (e.g. Hermann et al. 2014;Hainz-Renetzeder et al. 2014).
In line with Campagne et al. (2018) the focus on the assessment of capacities does not allow for the consideration of trade-offs or neighbouring effects (positive or negative) within the matrix, therefore the complexity and multifunctionality of ecosystems is embodied insufficiently.
Finally, although careful preparation of the materials to assure transparency and time efficiency was aimed for within the capacity matrix workshop, the goal of scoring all 79 classes within one day in consensus could not be achieved. For which reason we see an accessible size of the matrix as well as and realistic time management throughout the scoring as key issues to avoid revisions, in the achievement of a consistent result. Based on this experience we are aware that no final solution for highly complex ecosystem and landscape service assessments has been found yet and that related challenges are still manifold.
Shifting from the capacity matrix assessment to the mapping methodology, our proposed method allowed us to overcome several issues related to data quality. In fact, through the INDEC application all landscape elements were treated at the same level, being all transformed into point data. The second advantage of the method is that it conveys multiple simultaneous information of density and of cumulative intensity, based on capacity score and the size of service capacity. The approach has also the great advantage of addressing very different services ranging from ecological to socio-cultural aspects. Indeed, the INDEC was very adequate for the combination of different data sources. Combining and merging data from different sources means having to deal with different resolution, age, spatial reference, different metadata and so on. The INDEC proved very efficient and successful in overcoming the data discrepancies and therefore it could be particularly useful in projects with limited financial resources. Thanks to the INDEC, there was no modification in the original data sets of open land data and in the additional spatial indicators. Moreover, our methodology highlighted the spatial connectivity of landscape elements, thanks to the creation of clusters equally dependent on the intensity of the capacity, assigned through the capacity matrix, and on the density of the landscape elements, i.e. their proximity. Therefore, the method allowed an appropriate balance among isolated and large landscape elements (e.g. cultivated fields, meadows) and small-scaled but numerous landscape elements, such as for instance forest edges, tree rows and benches. By adopting a method that applies a weighting factor to each polygon, most of the small-scaled elements would be overlooked. This is also important regarding the connectivity needed to guarantee a functioning GI within the Wienerwald Biosphere Reserve. Liquete et al. (2015) also proposed a methodology focusing on the connectivity of selected indicators for service provision as a way to define GI functionality. The authors stress that not all green areas qualify as GI elements, a fact which we took in consideration by employing the capacity matrix.
The INDEC clusters identified the spatial connectivity of GI within the Wienerwald Biosphere Reserve not only needed for the provision of landscape services, but also required for landscape service flow (Kukkala and Moilanen, 2017). Depending on the service assessed, the service provision and its demand need to be more or less in proximity, as highlighted by Cimon-Morin et al. (2013), who argued that in regions dominated by humans (which is the case for the Wienerwald Biosphere Reserve) demand is always nearby.
The INDEC is based on a tailored landscape service assessment. Nevertheless, the methodology could be efficiently transferred to other data sets, in any other region, provided that the capacity matrix scores are revised. As such, the method can also be used for a replicable monitoring needed in biosphere reserves to measure sustainable development performance, as recommended by the Seville Strategy (UNESCO, 1996). Chapman (2012) outlines that 'adaptive monitoring based on ecosystem services provides the best means to develop necessary information for informed decision-making'. Concentrating on landscape elements as service providing units, they act as assessment endpoints in any adaptive monitoring program and can effectively identify the need for management actions. As such, the efficiency of the INDEC also makes it fruitful for applied analyses, such scenarios development. During the analyses, we carried out some exercises on scenarios development. For instance, we hypothesized the effects of management measures on the future provision of landscape services. Realistic measures might be extensive woodlets logging, change in the management of meadows, conversion of cultivated fields into vineyards or another cultivation, and so on. The INDEC might quite rapidly provide projections on the change of landscape service provision, based on each of such measures.
Last but not least, the INDEC is an approach fitting with the rationale behind the capacity matrix, since the size of landscape elements areas is not explicitly taken into account.
As regards the critical issues encountered in the development of the INDEC, we noticed that in some cases the density was a dominant factor over the intensity. This applies to the forest edges, which were so abundant that they were often included in the clusters of highest capacity. On the other hand, this method relies on the quality of the data set: in this study, we dealt with an "unbalanced" dataset, with high number of classes with few landscape elements (e.g. thermophilous dry shrubland with two landscape elements), and a low number of classes with many landscape elements: e.g. forest edges (11032 landscape elements); trails (7823); intermediate managed meadows (3283); hedgerows dominated by trees (3041).
A critical aspect in the development of the methodology was "translating" the difference in landscape services capacities into a buffer. In other words, what is the "spatial" proportion between a score "2" and a "5"? Another technical issue regarded the transformation into points: depending on the shape of the polygon, the focal points of polygons sometimes were outside the area of the polygon they belonged. This aspect remained unresolved.
The INDEC and the method employed in this work in general might be considered rather simplistic in the sense that they do not address disservices, i.e. conditions blocking the capacity of certain services, nor neighbouring aspects, i.e. influence of other landscape elements on the scoring for the service. More-over, the INDEC is static: no temporal dynamism was considered.
As a further source of criticism, it is worth mentioning that the huge data set handled needs powerful hardware and software. However, this aspect is not limited to the INDEC.

Conclusions
The definition of GI in the European Commission communication (2013) implies two main aspects which are also taken up by other authors (e.g. Kopperoinen et al., 2014): (i) conservation of valuable natural areas and (ii) the enhancement of ES provision. By developing the innovative tool INDEC, we covered both aspects -the recognition of valuable natural and semi-natural landscape elements and their capacity to deliver a whole range of ecological and socio-cultural landscape services. Although the assessment performed through the capacity matrix has a local character, the tool might be transferable to other reserves and other regions.
Several aspects might be improved and further tested in the next studies. The most interesting one would be to analyse the spatial interaction among landscape services, in order to answer stimulating research questions like: can we see trade-offs when overlapping the selected services?
Since the study focuses on service capacity only, no information on service demand is currently available. A future project might focus on this aspect, finding gaps between supply and demand (Burkhard et al. 2012).
Another stimulating analysis might be the development of the INDEC for the whole area of the Wienerwald Biosphere Reserve, by including forest and settlements. The results would be comparable to those presented in this study. In addition to this, since the INDEC method was tailored for a regional approach, it would be fascinating to tune it at the local scale, by selecting a set of municipalities within the Reserve.
In this way, communication and inclusion of stakeholders or the public might be desirable. The implementation of GI into spatial planning seems a considerable effort (Slätmo et al., 2019). As a biosphere reserve, the Wienerwald represents an excellent training ground for developing a well-founded basis for the sustainable planning of GI in the peri-urban and rural areas around Vienna bringing all necessary actors (political and administrative) together.

Acknowledgements
This study was conceived in the framework of the research project BRIMSEN (

Information services
Aesthetic Information

Recreation
Cultural and artistic info.
Spiritual and historic info.