Mapping ecosystem service capacity, flow and demand for landscape and urban planning: a case study in the Barcelona metropolitan region

Ecosystem services (ES) mapping is attracting growing interest from landscape and urban planning, but its operationalization in actual decision-making is still limited. A clear distinction between ES capacity, flow and demand can improve the usefulness of ES mapping as a decision-support tool by informing planners and policy-makers where ES are used unsustainably and where ES flow is failing to meet societal demand. This paper advances a framework for mapping and assessing the relationships between ES capacity, flow and demand with a focus on the identification of unsatisfied demand. The framework was tested in the Barcelona metropolitan region, Spain, considering two ES of critical relevance for the urban population: air purification and outdoor recreation. For both ES, spatial indicators of capacity, flow, demand and unsatisfied demand were developed using proxy- and process-based models. The results show a consistent spatial pattern of all these components along the urban-rural gradient for the two ES assessed. The flow of both ES mainly takes place in the periurban green areas whereas the highest capacity values are mostly found in the protected areas located on the outskirts of the metropolitan region. As expected, ES demand and particularly unsatisfied demand are mostly situated in the main urban core (i.e., Barcelona and adjacent cities). Our assessment also reveals that the current landscape planning instrument for the metropolitan region mostly protects areas with high capacity to provide ES, but might lead to declining ES flows in periurban areas due to future urban developments. We contend that the mapping of ES capacity, flow and demand can contribute to the successful integration of the ES approach in landscape and urban planning because it provides a comprehensive picture of the ES delivery process, considering both ecological and social underlying factors. However, we identify three main issues that should be better addressed in future research: (1) improvement of ES demand indicators using participatory methods; (2) integration of ecological thresholds into the analysis; and (3) use of a multi-scale approach that covers both the local and regional planning levels and cross-scale interactions between them.

Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: . Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 6 148 The conceptualization of ES demand (and unsatisfied demand) used here is inherently 149 challenging at the operational level because it requires information about desired or required 150 end conditions which can vary among different stakeholder groups, especially for cultural ES. 151 For the sake of our analysis, in this paper we used environmental quality standards and 152 recommendations as prescribed in policy as a proxy threshold to determine expected desired 153 or required end conditions related to ES demand from a societal perspective (see Paetzold et 154 al., 2010; Baró et al., 2015). A risk perspective is commonly used to quantify demand for 155 regulating ES (Wolff et al., 2015). Under this approach, demand for air purification can be 156 indirectly indicated considering the magnitude of pressures needing regulation (i.e., air Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: . Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 8 'open areas ' and 'urban' planning systems. However,given the relationship between 191 transportation and certain components of the selected ES (e.g., demand for air purification),   Barcelona (PTMB, 2010). See also Table 1. Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: . Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 9 Table 1. Description of the planning systems of the Territorial Metropolitan Plan of Barcelona (PTMB, 2010 The cultural ES outdoor recreation is probably one of the most valued ES in cities, decisively 225 contributing to enhance physical and mental health of the urban population (Chiesura 2004;226 Gómez- Baggethun et al., 2013;Triguero-Mas et al., 2015). The city of Barcelona and many of 227 its surrounding middle-size cities are characterized by a high degree of compactness and high 228 population density, involving a scarcity of inner green areas (Baró et al., 2014). Periurban million visitors according to a recent study (IERMB, 2008).

234
The regulating ES air purification is also the subject of growing attention in the policy 235 agenda. Abatement of air pollution is still a pressing challenge in most major urban areas 236 worldwide, especially in regard to dioxide nitrogen (NO2) and particulate matter (WHO,  , 2013; 2014). ESTIMAP is a collection of spatial models for ES assessment originally Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version:  continental scale, therefore it was adapted to the regional scope of this paper to make it usable 259 for urban and landscape planning. In the following subsections we describe this adaptation and downscaling process. presence of water. Each component was composed of one to four internal factors considered Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version:  Mapping outdoor recreation flow is challenging because data on the actual recreational use or  medium to very high capacity recreation areas (i.e., recreation capacity equal or higher than Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F Baró et al., 2015). species composition of urban vegetation. Since this information was not available for the 345 entire case study area, the capacity indicator for air purification was estimated following the Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F Where α and β are, respectively, the intercept and slope coefficients corresponding to each 352 broad land cover type j, namely forest, bare soil, water or any combination thereof. Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F 15 Considering the risk perspective described above (subsection 2.1), air purification demand 378 was mapped based on NO2 concentration levels and population density. A spatial cross-379 tabulation was carried out between both variables following the same approach as for 380 recreation, i.e., the higher NO2 concentration and population density the higher demand Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F Urban-rural gradients have been used to analyze ecological patterns and processes in urban  Table 1).

412
Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F   Unlike recreation capacity, the largest amount of high recreation flow values is to be found in 442 the forest areas located in the surroundings of urban settlements (Fig. 4b).
In general, riverine 443 and coastal (e.g., beaches) ecosystems also show very high recreation flow values. Obviously, 444 these results were expected as the flow assessment was restricted to close-to-home outdoor Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F air purification also shows higher flow values in periurban forest areas than other natural sites 447 located in the hinterland, although the spatial transition is smoother as compared to recreation 448 (Fig. 6b). Again, the natural parks of Collserola and Montseny illustrate these patterns demand values in the BMR for both analyzed ES ( Fig. 4c and Fig. 6c). This urban Finally, results show that unsatisfied demand is circumscribed to the urban core of Barcelona 468 and several middle-size cities ( Fig. 4d and Fig. 6d). Unsatisfied demand for recreation in total) where NO2 concentration is highest.

474
The urban-rural gradients of recreation and air purification for the BMR illustrate graphically 475 the spatial patterns shown on the maps and described in the above paragraphs (see Fig. 5 and 476 Fig. 7). The gradient for ES capacity is similar for both ES. The lowest values are in the first Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F 20 5 km, in the Barcelona core city, and they present a gradual rise as we move away from the 478 city centre. In both cases, capacity shows a substantial increase after km 5 followed by a 479 slight decrease after km 10-11. The periurban natural areas surrounding Barcelona (e.g., Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F  Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F . Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F area is only 31.18 km 2 (4.0%). As observed on the maps, the highest flow values are generally 541 located nearby or within suburban and urban land; hence possible impacts in terms of 542 urbanization processes can be anticipated in these areas. As expected, demand and unsatisfied 543 demand areas are mostly classified in the urban planning system. Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F  Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version:   The assessment of ES mismatches between flow and demand shows that unsatisfied demand 572 is mostly located in the urban core of the BMR and in several middle-size cities. We consider 573 that planning and policy strategies intended to reconcile flow and demand at the local level 574 should focus on different components of the framework depending on each ES.

576
For air purification, urban policies should focus on drivers of demand (i.e., air pollution 577 concentrations). Previous studies (Baró et al., 2014;2015) show that average air quality 578 improvements due to air pollution removal by vegetation is relatively low at the urban core, 579 suggesting a limited effectiveness to address ES mismatches by increasing ES flow through 580 strategies such as implementing tree-planting programs or selecting trees with high air 581 pollution removal capacity. Moreover, factors such as vegetation configuration and climate 582 conditions can limit the ability of vegetation to remove air pollutants, especially at the patch Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: . Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 26 scale such as in street canyons (Vos et al., 2013). Therefore, policy interventions should focus 584 on reducing and limiting traffic in certain areas, increasing public transport, incentivizing the 585 use of non or low-emitting vehicles (e.g., bicycles and electric vehicles), and enhancing developed here for the two selected ES are sufficiently credible and salient for landscape and Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 27 urban planning purposes in the case study area. However, several limitations and challenges 614 for future research can be highlighted from our assessment.

616
One of the main limitations of this approach is that the mapping of ES demand and flow 617 mostly relies on proxies (e.g., population density, air quality and distance) to indicate 618 expected demand and use. Therefore, there is potential for error if the assumed causal  unfortunately we are not aware of any recreation survey at the regional or national level which 639 covers the case study area. The demand approach for air purification, considering the 640 exposure of population to air pollution levels, is consistent with most assessments of demand 641 for regulating ES based on risk reduction (Wolff et al., 2015). However, a further refinement 642 could be achieved by identifying and mapping specific risk groups such as children and Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 28 elders, or by considering the areas where inhabitants practice outdoor activities and, therefore, 644 where they can be exposed to air pollution (Sunyer et al., 2015).

646
Another issue not considered in the spatial models used here relates to ecological thresholds 647 or tipping points (Andersen et al., 2009). An ecological threshold can be defined as a "point at 648 which an (ecological) system experiences a qualitative change, mostly in an abrupt and Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: . Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06

675
The issue of the spatial scale of ES capacity, flow and demand maps (Geijzendorffer and 676 Roche, 2014) also arises from this research. Our spatial results reflect that the actual use or 677 experience of the two ES analyzed highly depends on the proximity between ES providing 678 areas and benefiting areas (Syrbe and Walz, 2012), leading to relevant unsatisfied demands 679 which are mainly located at the urban core of the BMR. Therefore, we argue that both the 680 regional and local scales should be considered in these assessments in order to 681 comprehensively support planning and policy (Scholes et al., 2013). For instance, a more 682 detailed resolution could take into account small ES providing areas which are often 683 overlooked in regional assessments. These areas might have a relevant impact in terms of ES 684 flow and unsatisfied demand in the urban core. Moreover, the proposed interventions for both 685 ES could be much more accurately designed in local scale studies. However, the lack of fine 686 resolution spatial data for the appropriate quantification of ES capacity, flow and demand 687 indicators is probably a major challenge for this type of analyses (Derkzen et al., 2015). This 688 issue also calls for a strong institutional coordination between local and regional authorities 689 dealing with urban and environmental policy and for the harmonization of planning 690 instruments at different scales. Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: . Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 30 5. Conclusions 693 We advanced a spatial application of the ES capacity, flow and demand framework and tested 694 its usefulness for landscape and urban planning in a case study. Our results suggest that the approach that covers both the local and regional planning levels and cross-scale interactions 719 between them. Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version:  We thank Mingyuan Zhao (former master student in Wageningen University) and Carles Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10.1016/j.landusepol.2016.06.006 Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version:  As shown in Figure A1, the model follows a composite mapping procedure based on the 919 aggregation of the three components (i.e., degree of naturalness, nature protection and water).

920
Each component was developed through one or several factors considered relevant in the case 921 study of the BMR and for which spatial input data was available ( Table A1). All the scores 922 assigned to the water component factors and to the 'remarkable trees' factor of the nature 923 protection component were subject to a distance decay modelling, assuming that the 924 recreation potential decreases as the distance from the specific feature (e.g., a beach) 925 increases. The following inverse logistic function (1) was applied to these factors: Where: d is the distance from the specific feature, α and K are the size and shape parameters 930 of the function adjusted according to a distance threshold assessment, and w is the assigned 931 score. The parameters α and K were respectively set at 0.0035 and 30 for the factor 'beaches', 932 corresponding to a distance thresholds of 1000 m at which the score is decreased by 50% and 933 2000 m at which the score is zero, and 0.008 and 30 for the rest of factors, corresponding to 934 distance thresholds of 500 m and 1000 m (see also Table A1). The distance thresholds were 935 defined based on the expert consultation process.  (2) 944 945 Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. Mapping ecosystem service capacity, flow and demand for landscape and urban planning: A case study in the Barcelona metropolitan region. Land use policy 57, 405-417. doi:10. 1016/j.landusepol.2016.06.006 37 Table A1 947 Recreation potential components and data sources. Factors, scores and distance function thresholds defined in the expert consultation process. Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F

954
Potential trips for mapping the expected outdoor recreation flow were estimated using a 955 neighbor operator with a custom matrix. The custom matrix was based on the distance decay 956 function (1) considering α = 0.008 and K = 30 (see also Fig. A2). Important note: This is the author's post-print version of a research paper that was accepted for publication in the journal Land Use Policy (Elsevier). Therefore, it underwent full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the published version: Baró, F

963
Cross-tabulation matrix between a reclassified raster of Euclidian distances to recreation sites and the population 964 density grid used to obtain the outdoor recreation demand values (legend on the right). It assumes that all 965 inhabitants in the case study area have similar desires in terms of (everyday life) outdoor recreational 966 opportunities, but their level of fulfillment depends on proximity to recreation sites. Distance breaks consider the 967 recommended standards by regulatory agencies (Stanners and Bourdeau, 1995; Barbosa et al., 2007).  996 Input data for the processing of the air purification model. All the input variables were computed at 100 m of 997 resolution (pixel size). Table A4 1002 Cross-tabulation matrix between NO2 concentration levels and population density used to obtain the air 1003 purification demand values (legend on the right). It assumes that the higher NO2 concentration and population 1004 density the higher demand values. NO2 concentration break consider the current NO2 concentration limit in