Ecosystem Services 56 (2022) 101445 Contents lists available at ScienceDirect Ecosystem Services journal homepage: www.elsevier.com/locate/ecoser Full Length Article Using crowdsourced imagery to assess cultural ecosystem services in data-scarce urban contexts: The case of the metropolitan area of Cali, Colombia Emmanuel Zapata-Caldas a, b, Fulvia Calcagni c, Francesc Baró d, e, c, Johannes Langemeyer c, f, * a Grupo Territorios, Departamento de Geografía, Universidad del Valle, Cali, Colombia b Doctorado en Ciencias Ambientales. Universidad del Valle, Cali, Colombia c Institute of Environmental Science and Technology (ICTA), Universitat Autònoma de Barcelona (UAB), Edifici Z (ICTA-ICP), Carrer de les Columnes s/n, Campus de la UAB, 08193 Cerdanyola del Vallès, Spain d Department of Geography, Vrije Universiteit Brussel (VUB), Pleinlaan 2, B-1050 Brussels, Belgium e Department of Sociology, Vrije Universiteit Brussel (VUB), Pleinlaan 2, B-1050 Brussels, Belgium f Department of Geography, Humboldt Universität zu Berlin, Rudower Chaussee 16, 12489 Berlin, Germany A R T I C L E I N F O A B S T R A C T Keywords: Sustainable urban and metropolitan planning is increasingly benefiting from differentiated assessments of Cultural ecosystem services ecosystem services. Cultural ecosystem services (CES) are particularly relevant for urban residents’ mental and Social media data physical health, yet, quantification and mapping of CES are often challenging, especially so in the Global South. Ecosystem Service potential and realisation The use of social media data (SMD), which has recently gained importance for assessing CES at larger spatial and Metropolitan area of Cali (Colombia) temporal scales, provides a promising entry point for mitigating this informational gap in land-use planning. However, its application is mainly limited to European and North American cases and rarely applied to data- scarce urban regions in the Global South, with South America as no exception. Addressing this geographical gap, this study assesses CES of urban green spaces in the city of Cali, Colombia, based on 1,686 crowdsourced and geolocated photographs, and compares those results with a metropolitan scale CES potential assessment based on expert opinions performed in a previous study. Despite some important limitations primarily related to Flickr as a data source, we demonstrate the utility of this approach, especially for understanding the fine-scale generation of CES by small green spaces located within the urban fabric that are overlooked in the metropolitan scale expert- based assessment. These green spaces are highly relevant as inner-city pockets for CES production, especially in the form of “existence value” and “aesthetic experiences”, in contrast to green areas highlighted by experts at the metropolitan scale that serve primarily recreational purposes. Our results indicate the large potential of SMD- based CES assessment approaches for informing urban planning processes in the Global South. 1. Introduction recreation and aesthetic appreciation (Langemeyer et al., 2015; Martin et al., 2016; Pleasant et al., 2014; Willcock et al., 2016). Lack of Ecosystem service (ES) assessments are quickly gaining importance consideration of CES in landscape and urban planning can lead to for urban planning, including in the Global South (Lapointe et al., 2019; planning failures, as seen in Chan et al., (2012) and Tenerelli et al., Leh et al., 2013; Nahuelhual et al., 2013). However, spatial data avail­ (2016). ability, especially on cultural ecosystem services (CES), is often limited. In the absence of empirical data of CES, expert-based methods such CES are understood here as the non-material benefits derived from and as the ES matrix model are frequently used to determine ES “supply” or subject to relational values resulting from people’s perceptions, valua­ “capacity,” i.e., the potential to generate CES based on land-cover types tion, and interactions with and within the green infrastructure (Chan (Jacobs et al., 2015). In addition to limitations stemming from expert et al., 2016). Important examples of CES are opportunities for outdoor biases (Amorim Maia et al. 2020; Langemeyer et al. 2018), most of such * Corresponding author at: Institute of Environmental Science and Technology (ICTA), Universitat Autònoma de Barcelona (UAB), Edifici Z (ICTA-ICP), Carrer de les Columnes s/n, Campus de la UAB, 08193 Cerdanyola del Vallès, Spain. E-mail addresses: emmanuel.zapata@correounivalle.edu.co (E. Zapata-Caldas), johannes.langemeyer@uab.cat (J. Langemeyer). https://doi.org/10.1016/j.ecoser.2022.101445 Received 30 June 2021; Received in revised form 11 May 2022; Accepted 25 May 2022 Available online 21 June 2022 2212-0416/© 2022 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 assessments rely on medium resolution regional spatial land-cover data, researchers have addressed impacts of human activities on CES of wet­ which tends to overlook smaller green areas embedded within the urban lands related to ecotourism and education by using a participatory rural fabric, such as parks (Buitrago-Bermúdez et al., 2019), despite their assessment, participatory mapping methods, and interviews (Ricaurte critical importance for the local production of ES, especially CES (e.g., et al., 2014, 2017; Toledo et al., 2018). De Juan et al (2020), who Burkhard & Maes, 2017; Locke & McPhearson, 2018; Ramírez-Gómez analysed photographs from Instagram to assess the CES most valued by et al., 2015). High resolution data and approaches are required in this visitors of the Tairona National Park, conducted so far the only SMD- context. However, more accurate approaches to assess CES based on based study of which we are aware. expert interviews or user surveys are expensive and difficult to imple­ Our study in the metropolitan area of Cali—as a case study area­ ment, especially for larger geographic areas such as metropolitan re­ —based on SMD derived from the photo-sharing platform Flickr gions (Bagstad et al., 2016; Dou et al., 2017). (https://www.flickr.com/) and related to CES, is the first SMD-based Another critical issue stems from a discrepancy between the expert- study in an urban context in South America. The study objectives based mapping of CES potential and the actual realisation—i.e., the were twofold: (1) We aimed to map the realisation of CES in a high “flow”—of such benefits (Andersson et al., 2019). Unlike regulating ES, resolution, spatially explicit manner, and (2) to facilitate the uptake of which can be well determined based on biophysical characteristics, such SMD assessments by urban planning, comparing them with results of an as land-cover, the realisation of CES depends on a complex interaction expert-based estimation of CES potentials based on land-cover between institutional settings, social preferences, and abilities as well as maps—still the most common approach for incorporating CES in to the availability of green spaces (Andersson et al., 2021). For instance, urban planning at a metropolitan scale (Buitrago-Bermúdez et al., Langemeyer et al. (2018) observed a considerable mismatch between 2019)—and with fine-grain land-cover data to indicate the green spaces the landscape aesthetics potential suggested by local experts based on most important in fostering the realisation of CES. landscape characteristics and realisation estimates, particularly in urban and peri-urban areas based on a fine-grained assessment of social media 2. Material and methods data (SMD). Considering methodologies, Ilieva & Mcphearson (2018) argue for a 2.1. Case study area strong potential of sustainable urban planning approaches to benefit from the increasing availability of high resolution spatial and temporal The case study area is the metropolitan area of Cali (MAC), located in SMD. SMD, including georeferenced photographs and tracks, allow for the southwest of Colombia in the valley of the Cauca River, with an area novel and cost-efficient approaches to identify and map the realisation of about 8,500 km2 (see Fig. 1-B). The MAC is an example of a large of CES based on the observation of people’s actual behaviour (Gher­ Andean metropolis, including 30 municipalities and 30 watersheds, that mandi & Sinclair, 2019). For example, Oteros-Rozas et al. (2017), using ranges in altitude from 702 to 4,723 masl. According to the National photographs posted to the photo sharing platforms Flickr and Panor­ Administrative Department of Statistics (DANE, 2020) its population in amio, showed context-dependent preferences by looking at how land­ 2018 was about 4.2 million people (516 inhabitants/km2), mostly scape heterogeneity is related to CES diversity in five different concentrated in the city of Cali (57 %) (see Fig. 1-C). In the absence of a landscapes in Europe. Donahue et al. (2018) used SMD to characterise political-administrative metropolitan area, the MAC has been defined by park use in order to inform green space management in the Twin Cities Tabares-Mosquera et al. (2020) as a social-ecological system (sensus (Minneapolis and Saint Paul), Minnesota, United States. While SMD- Berkes & Folke, 1998) based on urban-functional and biophysical (wa­ based CES assessments are increasingly used in urban or metropolitan tersheds) criteria, and high social and ecological connections and areas in Europe, North America (e.g., Calcagni et al., in this issue; interdependencies. Donahue et al., 2018; Oteros-Rozas et al., 2017, 2018; Tammi et al., According to the land cover map developed by Buitrago-Bermúdez 2017), and parts of Asia, including Singapore and Japan (e.g., Richards et al. (2019), which applied the Corine Land Cover legend adapted to & Friess, 2015; Thiagarajah et al., 2015; Yoshimura & Hiura, 2017), Colombia (IDEAM, 2010), the MAC is characterised by 28 different land these approaches have not yet been well tested for cities in the Global cover types, 21 of which were identified as natural or semi-natural. The South. map was created at a 1:25,000 scale using a minimum mapping unit of In most developing regions, the use of SMD to map CES is still at its one hectare (Salitchev, 1979). infancy and often not well aligned with local planning needs. In Africa, SMD has so far been mainly used for assessing the influence of wildlife 2.2. Assessing the realisation of CES hotspots on the attractiveness of protected areas (Hausmann et al., 2018; Willemen et al., 2015). In South America, SMD has been used to map In this study we use geolocated photographs from within the MAC CES mostly in the southern and eastern parts of the continent, including posted on Flickr between 2004 and 2018. Flickr is a well-known photo- Argentina, Ecuador, Chile, and Brazil, addressing primarily rural areas sharing platform with about 90 million monthly users (cf. 101 Toivonen (see Catana, 2016; Kong and Sarmiento, 2022, in this issue, Martínez- et al., 2019); its content has been widely used around the world for Harms et al., 2018; Martínez-Pastur et al., 2015 respectively). Gher­ assessing the realisation of CES in a spatially explicit manner (e.g., mandi et al. (2020) studied visitation and benefits of natural and cul­ Angradi et al., 2018; Donahue et al., 2018; Dunkel, 2015; Martínez- tural heritage sites in Mexico using SMD. Yet, the potential for using Harms et al., 2018; Tenerelli et al., 2016; Yoshimura & Hiura, 2017). SMD to assess CES as a response to the stated need for spatially explicit Similarly to other studies (e. g. Gliozzo et al., 2016; Langemeyer et al., CES assessments (La Rosa et al., 2016; Nahuelhual et al., 2014) and in 2018; Amorim Maia et al., 2020), we assume that photographs shared order to support urban planning, has not been widely explored any­ online represent what people enjoy doing or seeing and what is believed where in South America. to align with shared social values (Calcagni et al., 2019). Filtering In this study, we examined the potential of applying SMD-based CES photographs for natural elements and human-nature experiences can assessments in urban planning and decision-making processes in thus be assumed to be a relevant proxy for representing the realisation of Colombia, South America. Reports on CES in Colombia include Esco­ CES—i.e., “the actual production of a service experienced by people” bedo et al. (2020), who examined perceptions and spatial literacy in (Villamagna et al., 2013) based on the appreciation of and beneficial identifying urban and peri-urban types of ES—including aesthetics value relationships with or within nature (Chan et al., 2016), including a full CES—based on interviews, quantitative machine-learning statistics, and range of human-designed to pristine ecosystems. qualitative methods. In another article, Angarita-Baéz et al. (2017) Access to public geotagged photographs is relatively easy via the studied the perceptions of CES among indigenous populations in the Flickr Application Programming Interface (API). Unlike other SMD Colombian Amazon by conducting semi-structured surveys. Other platforms, such as Facebook and Instagram, which have more restrictive 2 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 Fig. 1. Location of the MAC and the city of Cali in Colombia, South America. regulations for access and terms of use, Flickr widely permits the use of its data for scientific research (e.g., Ghermandi & Sinclair, 2019). On the Table 1 CICES classes: Simple descriptors and sample services. Based on Haines-Young & other hand, there are some known biases about Flickr data, including a Potschin-Young (2018). disproportional use of the social network by males (O’Hare and Mur­ dock, 2012). Particular biases for South America are not known to us, CICES 5.1. classes Simple Example CES Coding descriptor beyond the limitation of Flickr being less popular in that part of the world compared to Europe and North America. Recreational Using the Ecological Pictures activities environment for qualities of representing The visual content of the retrieved photographs was manually coded (activities sport and woodland that activities such as (adapting methodological approaches applied by Amorim Maia et al., promoting health, recreation; make it attractive biking, running, 2020; Hollenstein & Purves, 2010; Langemeyer et al., 2018; Tenerelli recuperation, or using nature to to hike; private swimming, et al., 2016, Tenerelli et al., 2017) and classified according to the enjoyment help stay fit gardens or hiking Common International Classification of Ecosystem Services (CICES) 5.1 through active or opportunities for immersive diving, swimming (Haines-Young & Potschin-Young, 2018). Three categories of CES were interactions) considereed: 1) activities promoting health, recuperation, or enjoyment Aesthetic The beauty of Areas of Panoramic or through active or immersive interactions—henceforth, recreational experiences nature outstanding landscape activities—; 2) aesthetic experiences; and 3) existence value (see also natural beauty; pictures panoramic views Table 1). The selection of these three CES resulted from an initial Existence value The things in Wilderness; “Close-up” screening of the content of photographs. nature that we charismatic pictures of plants All duplicate photographs and photographs with irrelevant content think should be wildlife or animals (e.g., taken indoors, selfies, etc.) were excluded from the analysis. Given conserved the relatively low number of relevant photographs, similarly to Calcagni et al. (in this issue), we did not apply the photo-user-day (PUD) Photographs were assigned to the CES “recreational activities” if the approach, which is commonly used with bigger datasets, to avoid a high features represented therein were related to enjoyment and engagement influence on the results by very active individual users (Martínez-Harms with nature when triggered by sports or other physical activities. These et al., 2018; Wood et al., 2013). activities are associated with benefits such as good health as well as 3 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 physical and mental well-being. Examples of this CES in photographs 2.3. 2.3. Comparing CES potentials and realisation from the MAC were people climbing mountains and walking, cycling, or running through natural environments. Photographs were coded as This second part of our analysis builds on a land cover-based CES “aesthetic experiences” if they depicted intellectual interactions with potential assessment performed by Buitrago-Bermúdez et al. (2019), nature—most commonly images conveying the beauty of the biophysi­ consistent with CICES 4.3 (Haines-Young & Potschin-Young, 2013) and cal characteristics or qualities of certain ecosystems. Many photographs thus comparable to the social-media based assessment of CES realisation in this CES category consisted of sightseeing places (i.e., panoramic for which our study uses the updated but back-compatible classification images, sunrises, and sunsets), which reflected benefits such as pleasure by Haines-Young & Potschin-Young (2018). and inspiration. Finally, the CES “existence value” was coded in pho­ To facilitate the comparison of the land cover-based potential tographs representing living systems that generated sentiments of assessment and the CES realisation at the metropolitan scale, we created appreciation for their existence. Most of these photographs included tables with three columns: CES realisation (i.e., number of photographs “close-ups” of flowers and/or animals (some of them in the context of that fell into each land cover category); normalised CES realisation (i.e., the city zoo). Even though the zoo can be seen as a place for utilitarian realisation divided by land cover area—number of photographs/land engagement with nature, this does not exclude non-use values, such as cover area [km2]—to determine whether the realisation of CES was high existence values, to arise together with experiential and outdoor tourism (≥3rd quartile of the normalised CES realisation range; Fig. 4), medium values. To differentiate between the two, we assume that the pictures (between 1st and 3rd quartile), or low (≤1st quartile); and expert- indicate a specific importance given to the “existence” of the depicted estimations of potentials to provide CES (i.e., land cover-based assess­ element of the natural system—based on the eco-centric perspective, ment of CES potential). which emphasizes the intrinsic value of nature (Rea & Munns, 2017; Lastly, we evaluated the city level CES realisation by overlapping the Scholte et al., 2016)—while experiential values potentially are repre­ resulting map with high-resolution spatial land cover data from the sented by the number of visitors to the zoo facility (Chan, 2012). Alcaldía de Santiago de Cali (2014). To assess the types of urban green ArcGIS 10.6 software was used to create photo count vector grid areas associated with the realisation of CES within the city of Cali, it was maps for each CES. To do so, we used the Joint Count output of the necessary to use land cover data on a finer scale than the 1:25,000 Optimised Hot Spot Analysis tool from ArcGIS (ESRI, 2018) to aggregate spatial scale of the MAC land cover map. Instead, we used spatial data photographs located within a vector grid of 1 × 1 km square size. Sum of from the Plan de Ordenamiento Territorial of Cali (Alcaldía de Santiago de photos was based on Eq. (1) after Langemeyer et al. (2018): Cali, 2014) to identify parks, ecological corridors, wetlands, and recre­ ∑ ational areas. fi,tot = Pi i 3. Results where fi,t is the CES realisation in the grid cell i given by the sum of photos geolocated in the same grid cell, Pi. The spatial distribution of the photographs within the MAC is shown in Fig. 2. The total amount of photographs uploaded to Flickr from Fig. 2. Photo count of the geotagged photographs from Flickr in a vector grid of 1 × 1 km per square. 4 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 December 2004 to June 2018 and located within the study area was photographs (287, 43.3 % of the total and medium normalised realisa­ 20,537 (Fig. 2A). Of these, 14,169 photographs had publicly-available tion) (Appendix B). URLs at the time of the assessment in August 2018 (Fig. 2B) and were Furthermore, 154 photos were classified under “recreational activ­ therefore accessible for the study. Out of these 14,169 photographs, we ities”. Once again, most photos are found within Cali and its surround­ classified 1,686 (12 %) as relevant—i.e., representing the realisation of ings (Appendix C), mainly within three land covers: recreation facilities CES—and further classified them with regard to their content into one of (3.9 % of the photographs, with a high normalised realisation), followed the three CES categories considered in this study (Table 1). The density by urban green areas (6.5 %), and continuous urban fabric (27 %), both of relevant photographs reached more than 1,000 photographs/km2 in with a high normalised realisation but with differences in the potential some areas of the city of Cali. About half of them (52 %) were classified to provide CES—4.8 and zero, respectively. Most photos are located as “existence value”, followed by “aesthetic experiences” (39 %), and within the sugar cane land cover (47.4 % of the total), however this land “recreational activities” (9 %) Fig. 3. cover has a medium normalised realisation and its potential to provide CES is 1 over 5 (Appendix D). 3.1. Comparison at the metropolitan area of Cali scale 3.2. Spatial patterns of CES realisation at the city scale Photographs representing “existence value” were widespread in the When analysing the realisation of CES in relation to the high- northern part of the MAC, but mostly present within the city of Cali. resolution land cover map from Cali’s Plan de Ordenamiento Territorial, More precisely, 12 % (104) of the photographs fell into urban green we observed that only 425 photographs—approximately a third of the areas (with a high normalised realisation; Fig. 4);, 26 % (231) were relevant pictures (i.e., 1,686 photographs)—intersected with one of the located in the continuous urban fabric, which in the expert estimation four types of land covers associated with urban green areas at the city based on land-covers was not considered to have any potential to pro­ scale (see Fig. 5). The CES “existence value” reported 366 associated vide CES, and 32.3 % (281) in pastures (Fig. 4). Another land cover that photographs at the city scale (86 % of the total number), while the CES stands out is “industrial or commercial units” which, despite having a “aesthetic experiences” reported 59 photographs (14 %). low number of photographs and reporting zero potential to provide CES, Fig. 5 shows that photographs are distributed within the north- its normalised realisation is high compared to other land covers (Fig. 4). western and southern extremes of the city (322 and 103 photos With regard to “aesthetic experiences”, the distribution pattern of the respectively). In the north-west, 289 photographs were taken in the photographs is quite similar to that of “existence value”. The city of Cali Zoológico de Cali, 6.6 % of them associated to “existence value” and 3.4 and its surroundings account for most of the photos and the highest % to “aesthetic experiences” (Fig. 5). The remaining land covers had low numbers are found in the north-western part of the city (Appendix A). A numbers of photos (under 11) and three of them only reported photo­ total of 662 photographs were assigned to this CES. Urban green areas graphs for “aesthetic experiences” (i.e., El Aguacatal and Cristo Rey have a high normalised realisation (82 photos, or 12.3 %), followed by parks, and the Río Cali ecological corridor). On the other hand, in the industrial or commercial units (27 photos, also with a high normalised southern part of the city, the Las Garzas wetland registered most of the realisation) and, finally, continuous urban fabric contains most of the photographs (64 in total) of which 92.2 % were associated to “existence value” and 7.8 % to “aesthetic experiences”. The remaining land covers reported less than eight photographs (Fig. 5). 4. Discussion 4.1. Mismatches of CES potential and realisation in the metropolitan area of Cali One important finding of our SMD-based assessment is the mismatch between the (high) potential to provide CES estimated by Buitrago- Bermúdez et al. (2019) for most of the land covers and the low nor­ malised CES realisation based on the analysis of photographs posted to Flickr. This is the case for dense forests, grasslands located in paramos, rivers, and the two different types of mosaic landscapes (i.e., mosaic of crops, pastures and natural spaces, and mosaic of pastures with natural spaces). The only exception is related to urban green areas, which re­ ported both high CES potential and high normalised realisation. The distribution of population density is likely a key driver explaining this observed pattern, which corroborates comparable studies from Europe that found peri-urban areas to be most relevant for sustaining CES (e.g., Baró et al., 2016; Langemeyer et al., 2018). This pattern could likely be attributed to barriers in terms of accessing and obtaining CES related to institutional settings, people’s perceptions, and biophysical infrastructure (Andersson et al., 2021; Wolff, 2021). Through this understanding, institutional factors, in the form of access restrictions and non-public ownership, might constitute a major obstacle for the use of peri-urban green spaces in the MAC. For instance, environmental protection (e.g., regional, and national natural parks) seems to constrain CES realisation. There is evidence that pro­ tected areas can create a barrier for people to access ecosystem services when safeguarding biodiversity from human threats results in the Fig. 3. Photo count of the “existence value” CES. Sum of photographs at the exclusion of people from these areas with high CES potential (Mcdonald vector grid level (1 × 1 km per square). et al., 2008; Palomo et al., 2011). 5 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 Fig. 4. Comparison of land cover-based potential for “existence value” CES and CES realisation at the MAC scale. Normalised realisation was calculated dividing CES realisation by land cover area (i.e., number of photographs/land cover area [km2]) to determine whether the realisation was high (≥3rd quartile [blue]), medium (between 1st and 3rd quartile [yellow]), or low (≤1st quartile [orange]). Elaborated by the authors based on data of land cover-based CES potential assessed by Buitrago-Bermúdez et al. (2019). Fig. 5. Number of photographs per land cover in the city of Cali. Elaborated by the authors based on data from Cali’s Plan de Ordenamiento Territorial (Alcaldía de Santiago de Cali, 2014). That a considerable realisation of CES was concentrated in the north- Amorim Maia et al., 2020; Langemeyer et al., 2018), high profile western part of the city of Cali cannot be related directly to a particular infrastructure seemed to influence the realisation of CES. For instance, land-cover (or green space) type but might be better explained by as­ the Zoológico de Cali, located in the north-western part of the city, was pects associated to perception—more specifically, perception of security the source of 68 % of the photographs related to the “existence value” or the lack thereof. While crime events, such as robbery and homicides, and “aesthetic experiences”, while the Las Garzas wetland in the south are more frequently associated with the city centre and the eastern part Cali suburbs was the source of 15 % of the photographs. This concen­ of the city, the north-western area is generally perceived as safer tration of pictures in these photogenic destinations could be attributed (Alcaldía de Santiago de Cali, 2011; Loaiza-Cerón, 2012). Safety per­ to the fact that people are less likely to take photographs in neigh­ ceptions might be a significant barrier or enabling factor for the real­ bourhood or pocket parks, used for daily recreation. Follow-up research, isation of CES in many South American cities, which historically have for instance through grounding surveys (e.g., Winder et al., 2021), will been characterised by having some of the highest crime rates in the have to confirm this assumption. world (Soares & Naritomi, 2013). In our study, and in line with other studies on urban CES (e.g., 6 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 4.2. CES assessments in South American cities indicating the general usefulness of SMD to assess the realisation of CES in data-scarce urban contexts like South America, it also highlights Our assessment underscores the need for finer-scale land cover data several needs for future research. in South American cities to better understand the generation of CES. The Although we applied a methodology widely used and validated analysis of the realisation of CES in urban green areas of the MAC allows elsewhere (Dunkel, 2015; Langemeyer et al., 2018; Richards & Friess, for differentiating the land covers related to such realisation, in line with 2015), in South America Flickr is used less than other SMD platforms. the relationship between land cover diversity and CES diversity (Oteros- Thus, including other such platforms in the study (e.g., Instagram, Rozas et al., 2018). Facebook, or Twitter) would probably provide more relevant data Nevertheless, our results remain partly inconclusive at the scale of (Mejia-Llano, 2020). As can be observed in Fig. 6, in the context of the the MAC due to the relatively low number of photographs available from MAC, 2013 was the only year within the period assessed—i.e., 2004 to Flickr. But the SMD-based analysis allowed us to identify spatial patterns 2018—that exceeded 200 Flickr photos for the “existence value” CES. of the realisation of CES in the city of Cali, showing higher realisation Overall, that CES was the most represented in the study area, followed levels in the northern and southern parts of the city associated with the by the “aesthetics experiences” and “recreational activities” CES. This zoo and a wetland (see Fig. 5). This was interpreted thanks to the use of situation underscores the need for other SMD sources in the given land cover data from Cali’s Plan de Ordenamiento Territorial, which context. allowed us to identify more specific urban green areas supporting CES (i. It is also likely that the low use of Flickr as a social photo-sharing e., parks, ecological corridors, wetlands, and recreational areas). network in Colombia (Flickr, 2020), in comparison with European and Therefore, the city-scale analysis clearly underlined the need for the use North American countries, has played a role in the non-identification of of finer-resolution land cover data (Locke & McPhearson, 2018), which CES in some land covers with high potential to provide CES (e.g., considers small urban green areas in Cali, such as pocket parks, as these continuous urban fabric and industrial or commercial units). Therefore, are expected to be critical providers of CES. Projects similar to the Eu­ the legal restrictions on the use of data for research as established by ropean Urban Atlas (Copernicus Land Monitoring Service, 2021) could Facebook or Instagram—which have proven to be more widely used in be developed for Colombia and other South America cities to stimulate the Colombian context (Bailey et al., 2018; Mejia-Llano, spatial data required for urban planning. 2020—unequally hamper CES research in developing countries such Compared to similar exercises developed in European studies (Lan­ as Colombia, compared to Europe and North America (Calcagni et al., gemeyer et al., 2018; Tenerelli et al., 2016, Tenerelli et al., 2017), the 2019; Wood et al., 2013). In recent years, the steady increase in the final number of relevant photographs was low compared to the total number of studies showing the usefulness of public good research has amount of SMD for the area. Clearly there are many photographs that do contributed to the revision of Twitter’s data accessibility policy. The not explicitly or implicitly include CES content. However, compared to case of Cali presented in this study further corroborates this stand and the amount of data collected through traditional subject methods, such might contribute to inspire other platforms to support open science and as surveys or interviews, studies based on SMD are cost- and time- grant wider access to their data streams. effective in terms of data collection (Ghermandi & Sinclair, 2019). In Furthermore, this study reinforces the need for (mis)match assess­ this specific case, SMD provides a starting point to understand CES ments between CES potential and the actual realisation of CES as already realised by residents and visitors to the MAC, which, according to our highlighted in previous studies (e.g., Baró et al., 2016; Schröter et al., findings, is mostly provided by urban ecosystems located close to urban 2014; Langemeyer et al., 2018) while also complementing them by areas (i.e., urban parks, green corridors, and other urban green spaces), analysing these mismatches at two different urban scales. This research in contrast with a wider range of opportunities observed in different has proven that SMD data can fill in data gaps of coarser land cover- cities of North America (Donahue et al., 2018). based approaches for mapping CES. Identifying mismatches between the realisation of CES and the potential to provide CES thus requires the 4.3. Future research needs use of even finer land cover data. Future research should consider alternative land-cover data, such as Open Street Maps (OSM, 2020), to In contexts in which SMD analyses are not yet highly developed, complement the official data provided by the city’s administration exploratory studies can serve as a first indicator of the usefulness of (Levin et al., 2017; Upton et al., 2015). SMD-based methodologies (Martínez-Harms et al., 2018; Martínez- Pastur et al., 2015). In this sense, our research serves as a pilot study for the use of SMD in South American urban and metropolitan areas. While Fig. 6. Number of photographs per year from 2004 to 2018 in the MAC. 7 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 5. Conclusions separately, guaranteeing fully anonymized data sets. The most sensitive of this data, such as names and places of residence, was removed once This study is a novel application of SMD approaches related to CES in the assessment had finished. Concerns on whether data should be the urban context of Colombia and other developing countries in the considered public or private underlie an ongoing ethical debate that, to Global South with a focus on South America, where studies have been date restricts data access to social media researchers. Even in the carried out mainly in rural and protected areas. A finding related to CES absence of a common and clear ethical framework, given the relevant in Cali is the potential for assessing CES through SMD to better inform information that it facilitates, the debate about the scientific use of social the urban planning process. This approach provides data insights that media data must continue in motion. would be difficult and expensive to collect through other commonly used means, such as expert interviews or user surveys. Declaration of Competing Interest This paper sheds light on an alternative way for data collection to inform decision-making processes related to urban green infrastructure The authors declare that they have no known competing financial planning at the metropolitan and city scales in the Global South. We interests or personal relationships that could have appeared to influence believe that this field of research can be further developed and boosted the work reported in this paper. with additional studies in the coming years, both in Colombia and in South America. To achieve that, the contribution of SMD to urban Acknowledgements planning processes must be better acknowledged. In the field of urban ecosystem service research, it is fundamental to Authors acknowledge financial support from the EU’s Horizon 2020 implement approaches that compare potential and realisation assess­ framework program for research and innovation (project NATUR­ ments of CES at different spatial scales to quantify the actual use of VATION, grant agreement ID: 730243 and INTERLACE project grant urban green areas by urban residents. In our study, high-resolution agreement 869324) and the German-Israeli Foundation (GIF grant no. I- spatial data was necessary for understanding the spatial patterns of 1533-500.15/2021). This work also contributes to the “María de urban CES realisation. Maeztu” Programme for Units of Excellence of the Spanish Ministry of Science and Innovation (CEX2019-000940-M). The authors thank the Ethics statement AGAUR Catalan governmental agency for funding the research stay of E. Z.C. at ICTA-UAB through the Laboratory for the Analysis of Social- Flickr data were collected through the Flickr API complying with Ecological Systems in a Globalised world (LASEG 2017-SGR-1588) and Flickr’s terms of service. To reinforce the anonymity, and guarantee to for supporting the work by F.C. (grant no. 2018FI_B00635). We thank Flickr users, we applied safety precautions to the only computer con­ Damiano Calcagni for his support with data collection. The manuscript taining all data. Individual data and spatial data had always been stored was proof-edited by Emilia Oscilowicz. 8 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 Appendix. A . Photo count of the “aesthetic experiences” CES. Sum of photographs at the vector grid level (1 × 1 km per square). 9 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 Appendix B . Land cover-based potential for “aesthetic experiences” CES and CES realisation comparison at the MAC scale. Normalised realisation was calculated dividing CES realisation by land cover area (i.e., number of photographs/land cover area [km2]) to determine whether the realisation was high (≥3rd quartile [blue]), medium (between 1st and 3rd quartile [yellow]), or low (≤1st quartile [orange]). Elaborated by the authors based on data of land cover-based CES potential assessed by Buitrago-Bermúdez et al. (2019). 10 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 Appendix C . Photo count of the “Recreational activities” CES. Sum of photographs at the vector grid level (1 × 1 km per square). Appendix D 11 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 . Land cover-based potential for “recreational activities” CES and CES realisation comparison at the MAC scale. Normalised realisation was calculated dividing CES realisation by land cover area (i.e., number of photographs/land cover area [km2]) to determine whether the realisation was high (≥3rd quartile [blue]), medium (between 1st and 3rd quartile [yellow]), or low (≤1st quartile [orange]). Elaborated by the authors based on data of land cover-based CES potential assessed by Buitrago-Bermúdez et al. (2019). References Dou, Y., Zhen, L., De Groot, R., Du, B., Yu, X., 2017. Assessing the importance of cultural ecosystem services in urban areas of Beijing municipality. Ecosyst. Serv. 24, 79–90. https://doi.org/10.1016/j.ecoser.2017.02.011. Alcaldía de Santiago de Cali, 2011. Violencia, convivencia y dinámica social en Cali: Dunkel, A., 2015. Visualising the perceived environment using crowdsourced photo lectura desde el observatorio social. Observatorio Social de Cali. geodata. Landscape Urban Plann. 142, 173–186. https://doi.org/10.1016/j. Alcaldía de Santiago de Cali. (2014). Plan de Ordenamiento Territorial (POT), año 2014. landurbplan.2015.02.022. https://www.cali.gov.co/planeacion/publicaciones/106497/pot_2014_idesc/. Escobedo, F.J., Bottin, M., Cala, D., Sandoval Montoya, D.L., 2020. Spatial literacy Amorim Maia, A.T., Calcagni, F., Connolly, J.J.T., Anguelovski, I., Langemeyer, J., 2020. influences stakeholder’s recognition and mapping of peri-urban and urban Hidden drivers of social injustice: uncovering unequal cultural ecosystem services ecosystem services. Urban Ecosyst. 23 (5), 1039–1049. https://doi.org/10.1007/ behind green gentrification. Environ. Sci. Policy 112, 254–263. s11252-020-00962-y. Andersson, E., Borgstrom, S., Haase, D., Langemeyer, J., Mascarenhas, A., ESRI. (2018). Optimised Hot Spot Analysis. https://desktop.arcgis.com/en/arcmap/latest/ McPhearson, T., Wolff, M., Laszkiewicz, E., Kronenberg, J., Barton, D.N., Herreros- tools/spatialstatistics-toolbox/optimized-hot-spot-analysis.htm. Cantis, P., 2021. A context-sensitive systems approach for understanding and Flickr. (2020). Travel Photography in Colombia. https://www.flickr.com/groups/ enabling ecosystem service realisation in cities. Ecol. Soc. 26 (2) https://doi.org/ colombia/members/. 10.5751/ES-12411-260235. Ghermandi, A., Camacho-Valdez, V., Trejo-Espinosa, H., 2020. Social media-based Andersson, E., Langemeyer, J., Borgström, S., McPhearson, T., Haase, D., Kronenberg, J., analysis of cultural ecosystem services and heritage tourism in a coastal region of Barton, D., Davis, M., Naumann, S., Röschel, L., Baró, F., 2019. Enabling Green and Mexico. Tourism Manage. 77 https://doi.org/10.1016/j.tourman.2019.104002. Blue Infrastructure to Improve Contributions to Human Well-Being and Equity in Ghermandi, A., Sinclair, M., 2019. Passive crowdsourcing of social media in Urban Systems. BioScience 69 (7), 556–574. https://doi.org/10.1093/biosci/biz058. environmental research: A systematic map. Global Environ. Change 55, 36–47. Angarita-Baéz, J.A., Pérez-Miñana, E., Beltrán Vargas, J.E., Ruiz Agudelo, C.A., Paez Gliozzo, G., Pettorelli, N., Muki Haklay, M., 2016. Using crowdsourced imagery to detect Ortiz, A., Palacios, E., Willcock, S., 2017. Assessing and mapping cultural ecosystem cultural ecosystem services: A case study in South Wales, UK. Ecology and Society 21 services at community level in the Colombian Amazon. Internat. J. Biodiver. Sci. (3). https://doi.org/10.5751/ES-08436-210306. Ecosyst. Services Manage. 13 (1), 280–296. https://doi.org/10.1080/ Haines-Young, R., Potschin-Young, M., 2013. Common International Classification of 21513732.2017.1345981. Ecosystem Services (CICES, Version 4.3). In: Report to the European Environment Angradi, T., Launspach, J., Debbout, R., 2018. Determining preferences for ecosystem Agency, pp. 1–17. benefits in Great Lakes Areas of Concern from photographs posted to social media. Haines-Young, R., Potschin-Young, M., 2018. Revision of the common international J. Great Lakes Res. 44 (2), 340–351. https://doi.org/10.1016/j.jglr.2017.12.007. classification for ecosystem services (CICES V5.1): A Policy Brief. One Ecosystem 3, Bagstad, K.J., Reed, J.M., Semmens, D.J., Sherrouse, B.C., Troy, A., 2016. Linking e27108. biophysical models and public preferences for ecosystem service assessments: a case Hausmann, A., Toivonen, T., Slotow, R., Tenkanen, H., Moilanen, A., Heikinheimo, V., & study for the Southern Rocky Mountains. Reg. Environ. Change 16 (7), 2005–2018. Minin, E. Di. (2018). Social Media Data Can Be Used to Understand Tourists’ https://doi.org/10.1007/s10113015-0756-7. Preferences for Nature-Based Experiences in Protected Areas. 11(February), 1–10. Bailey, A.A., Bonifield, C.M., Arias, A., 2018. Social media use by young Latin American https://doi.org/10.1111/conl.12343. consumers: An exploration. J. Retail. Consumer Serv. 43 (January), 10–19. https:// Hollenstein, L., & Purves, R. S. (2010). Exploring place through user-generated content: doi.org/10.1016/j.jretconser.2018.02.003. Using Flickr tags to describe city cores. 1(1), 21–48. https://doi.org/10.5311/ Baró, F., Palomo, I., Zulian, G., Vizcaino, P., Haase, D., Gómez-Baggethun, E., 2016. JOSIS.2010.1.3. Mapping ecosystem service capacity, flow and demand for landscape and urban Ideam, 2010. Leyenda nacional de coberturas de la tierra. Metodología CORINE Land planning: A case study in the Barcelona metropolitan region. Land Use Policy 57, Cover adaptada para Colombia Escala 1:100.000. Instituto de Hidrología, 405–417. https://doi.org/10.1016/j.landusepol.2016.06.006. Meteorología y Estudios Ambientales. Berkes, F., Folke, C., 1998. Linking social and ecological systems for resilience and Ilieva, R.T., McPhearson, T., 2018. Social-media data for urban sustainability. Nat. sustainability. Linking Soc. Ecol. Syst. 1, 13–20. Sustainability 1 (10), 553–565. Buitrago-Bermúdez, O., Zapata-Caldas, E., Gaitán-Idárraga, L. F., Tabares-Mosquera, M., Jacobs, S., Burkhard, B., Van Daele, T., Staes, J., Schneiders, A., 2015. “The Matrix MorenoDelgado, D., Gómez, A., Otero, Y., Burbano, Z., & Bedoya, A. (2019). Reloaded”: A review of expert knowledge use for mapping ecosystem services. Ecol. Lineamientos para la gestión ambiental de la estructura ecológica principal del área de Model. 295, 21–30. https://doi.org/10.1016/j.ecolmodel.2014.08.024. influencia metropolitana inmediata de Cali, Colombia. Kong, I., Sarmiento, F.O., 2022. Utilizing a crowdsourced phrasal lexicon to identify Burkhard, B., Maes, J. (Eds.), 2017. Mapping Ecosystem Services. Pensoft Publishers. cultural ecosystem services in El Cajas National Park, Ecuador. Ecosystem Services Calcagni, F., Amorim-Maia, A.T., Connolly, J.J.T., Langemeyer, J., 2019. Digital co- 56, 101441. https://doi.org/10.1016/j.ecoser.2022.101441. construction of relational values: understanding the role of social media for La Rosa, D., Spyra, M., Inostroza, L., 2016. Indicators of Cultural Ecosystem Services for sustainability. Sustain. Sci. 14 (5), 1309–1321. https://doi.org/10.1007/s11625- urban planning: A review. Ecological Indicators 61, 74–89. https://doi.org/ 019-00672-1. 10.1016/j.ecolind.2015.04.028. Calcagni, F., Nogué-Batallé, J., Baró, F., Connolly, J. J. T., Langemeyer, J., n.d.. A tag is Langemeyer, J., Baró, F., Roebeling, P., Gómez-Baggethun, E., 2015. Contrasting values worth a thousand pictures: a framework for an empirically grounded typology of of cultural ecosystem services in urban areas: The case of park Montjuïc in relational values through social media. Ecosystem Services. In this issue. Barcelona. Ecosyst. Serv. 12, 178–186. https://doi.org/10.1016/j. Catana, A.V., 2016. Using social media to assess cultural ecosystem services generated in ecoser.2014.11.016. protected areas in Patagonia. Norwegian University of Science and Technology. Langemeyer, J., Calcagni, F., Baró, F., 2018. Mapping the intangible: Using geolocated Chan, K.M.A., Balvanera, P., Benessaiah, K., Chapman, M., Díaz, S., Gómez- social media data to examine landscape aesthetics. Land Use Policy 77, 542–552. Baggethun, E., Gould, R., Hannahs, N., Jax, K., Klain, S., Luck, G.W., Martín- Lapointe, M., Cumming, G. S., & Gurney, G. G. (2019). Comparing Ecosystem Service López, B., Muraca, B., Norton, B., Ott, K., Pascual, U., Satterfield, T., Tadaki, M., Preferences between Urban and Rural Dwellers. 69(2), 108–116. https://doi.org/ Taggart, J., Turner, N., 2016. Opinion: Why protect nature? Rethinking values and 10.1093/biosci/biy151. the environment. Proc. Natl. Acad. Sci. 113 (6), 1462–1465. https://doi.org/ Leh, M.D.K., Matlock, M.D., Cummings, E.C., Nalley, L.L., 2013. Quantifying and 10.1073/pnas.1525002113. mapping multiple ecosystem services change in West Africa. Agric. Ecosyst. Environ. Chan, K.M.A., Satterfield, T., Goldstein, J., 2012. Rethinking ecosystem services to better 165, 6–18. https://doi.org/10.1016/j.agee.2012.12.001. address and navigate cultural values. Ecol. Econ. 74, 8–18. https://doi.org/10.1016/ Levin, N., Lechner, A.M., Brown, G., 2017. An evaluation of crowdsourced information j.ecolecon.2011.11.011. for assessing the visitation and perceived importance of protected areas. Appl. Copernicus Land Monitoring Service. (2021). European Urban Atlas. https://land. Geogr. 79, 115–126. https://doi.org/10.1016/j.apgeog.2016.12.009. copernicus.eu/local/urban-atlas DANE. (2020). Proyecciones y reproyecciones de Loaiza-Cerón, W., 2012. Distribución espacial del índice de propensión al homicidio población. https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia- (IPH) en las comunas de Cali. Colombia. Perspectiva Geográfica 17 (17), 169–192. y-poblacion/proyeccionesde-poblacion. https://doi.org/10.19053/01233769.2267. de Juan, S., Ospina-Alvarez, A., Villasante, S., & Ruiz-Frau, A. (2020). Wide-scale Locke, D.H., McPhearson, T., 2018. Urban areas do provide ecosystem services. Front. assessment of cultural ecosystem services in coastal areas using graph theory on social Ecol. Environ. 16 (4), 203–205. https://doi.org/10.1002/fee.1796. media data. http://arxiv.org/abs/2007.14308. Martin, C.L., Momtaz, S., Gaston, T., Moltschaniwskyj, N.A., 2016. A systematic DANE, 2020. Proyecciones y reproyecciones de población. Departamento Administrativo quantitative review of coastal and marine cultural ecosystem services: Current status Nacional de Estadística. and future research. Marine Policy 74, 25–32. https://doi.org/10.1016/j. Donahue, M.L., Keeler, B.L., Wood, S.A., Fisher, D.M., Hamstead, Z.A., McPhearson, T., marpol.2016.09.004. 2018. Using social media to understand drivers of urban park visitation in the Twin Martínez-Harms, M.J., Bryan, B.A., Wood, S.A., Fisher, D.M., Law, E., Rhodes, J.R., Cities, MN. Landscape Urban Plann. 175, 1–10. Dobbs, C., Biggs, D., Wilson, K.A., 2018. Inequality in access to cultural ecosystem services from protected areas in the Chilean biodiversity hotspot. Sci. Total Environ. 636, 1128–1138. https://doi.org/10.1016/j.scitotenv.2018.04.353. 12 E. Zapata-Caldas et al. Ecosystem Services 56 (2022) 101445 Martínez-Pastur, G., Peri, P.L., Lencinas, M.V., García-Llorente, M., Martín-López, B., Salitchev, K.A., 1979. Cartografía. Pueblo y Educación. 2015. Spatial patterns of cultural ecosystem services provision in Southern Scholte, S.S., Todorova, M., Van Teeffelen, A.J., Verburg, P.H., 2016. Public support for Patagonia. Landscape Ecol. 31 (2), 383–399. https://doi.org/10.1007/s10980-015- wetland restoration: what is the link with ecosystem service values? Wetlands 36 (3), 0254-9. 467–481. Mcdonald, R.I., Kareiva, P., Forman, R.T.T., 2008. The implications of current and future Schröter, M., Barton, D.N., Remme, R.P., Hein, L., 2014. Accounting for capacity and urbanization for global protected areas and biodiversity conservation. Biol. Conserv. flow of ecosystem services: A conceptual model and a case study for Telemark, 141 (6), 1695–1703. https://doi.org/10.1016/j.biocon.2008.04.025. Norway. Ecol. Ind. 36, 539–551. https://doi.org/10.1016/j.ecolind.2013.09.018. Mejia-Llano, J. C. (2020). Estadísticas de redes sociales 2020: usuarios de Facebook, Soares, R.R., Naritomi, J., 2013. Understanding high crime rates in Latin America. Econ. Instagram, YouTube, LinkedIn, Twitter, Tiktok y otros. https://www.juancmejia.com/ Crime (Issue July). https://doi.org/10.7208/chicago/9780226153766.003.0002. marketing-digital/estadisticas-deredes-sociales-usuarios-de-facebook-instagram- Tabares-Mosquera, M., Zapata-Caldas, E., Buitrago-Bermúdez, O., 2020. Valoración de linkedin-twitter-whatsapp-y-otros-infografia/. servicios ecosistémicos para la identificación de estructuras ecológicas Nahuelhual, L., Carmona, A., Laterra, P., Barrena, J., Aguayo, M., 2014. A mapping metropolitanas: el caso de Cali, Colombia. Cuadernos de Investigación Geográfica 46 approach to assess intangible cultural ecosystem services: The case of agriculture (2), 603–631. https://doi.org/10.18172/cig.3952. heritage in Southern Chile. Ecological Indicators 40, 90–101. https://doi.org/ Tammi, I., Mustajärvi, K., Rasinmäki, J., 2017. Integrating spatial valuation of ecosystem 10.1016/j.ecolind.2014.01.005. services into regional planning and development. Ecosyst. Serv. 26, 329–344. Nahuelhual, L., Carmona, A., Lozada, P., Jaramillo, A., Aguayo, M., 2013. Mapping Tenerelli, P., Demšar, U., Luque, S., 2016. Crowdsourcing indicators for cultural recreation and ecotourism as a cultural ecosystem service: An application at the local ecosystem services: A geographically weighted approach for mountain landscapes. level in Southern Chile. Appl. Geogr. 40, 71–82. https://doi.org/10.1016/j. Ecol. Ind. 64, 237–248. https://doi.org/10.1016/j.ecolind.2015.12.042. apgeog.2012.12.004. Tenerelli, P., Püffel, C., Luque, S., 2017. Spatial assessment of aesthetic services in a OSM. (2020). Open Street Map, about. https://www.openstreetmap.org/about. complex mountain region: combining visual landscape properties with Oteros-Rozas, E., Martín-López, B., Fagerholm, N., Bieling, C., Plieninger, T., 2017. crowdsourced geographic information. Landscape Ecol. 32 (5), 1097–1115. https:// Ecosystem services and landscape features across five European sites. Ecol. Ind. doi.org/10.1007/s10980-017-0498-7. https://doi.org/10.1016/j.ecolind.2017.02.009. Thiagarajah, J., Wong, S.K.M., Richards, D.R., Friess, D.A., 2015. Historical and Oteros-Rozas, E., Martín-López, B., Fagerholm, N., Bieling, C., Plieninger, T., 2018. Using contemporary cultural ecosystem service values in the rapidly urbanising city state social media photos to explore the relation between cultural ecosystem services and of Singapore. Ambio 44 (7), 666–677. https://doi.org/10.1007/s13280-015-0647-7. landscape features across five European sites. Ecol. Ind. 94, 74–86. https://doi.org/ Toivonen, T., Heikinheimo, V., Fink, C., Hausmann, A., Hiippala, T., Järv, O., 10.1016/j.ecolind.2017.02.009. Tenkanen, H., Di Minin, E., 2019. Social media data for conservation science: A O’Hare, N., & Murdock, V. (2012). Gender-based models of location from Flickr. GeoMM methodological overview. Biol. Conserv. 233, 298–315. 2012 - Proceedings of the 2012 ACM International Workshop on Geotagging and Its Toledo, D., Briceño, T., Ospina, G., 2018. Ecosystem service valuation framework applied Applications in Multimedia, Co-Located with ACM Multimedia 2012, 33–38. https:// to a legal case in the Anchicaya region of Colombia. Ecosyst. Serv. 29, 352–359. doi.org/10.1145/2390790.2390802. Upton, V., Ryan, M., O’Donoghue, C., Dhubhain, A.N., 2015. Combining conventional Palomo, I., Martín-López, B., López-Santiago, C., Montes, C., 2011. Participatory scenario and volunteered geographic information to identify and model forest recreational planning for protected areas management under the ecosystem services framework: resources. Appl. Geogr. 60, 69–76. https://doi.org/10.1016/j.apgeog.2015.03.007. The Doñana socialecological system in Southwestern Spain. Ecol. Soc. 16 (1) https:// Villamagna, A.M., Angermeier, P.L., Bennett, E.M., 2013. Capacity, pressure, demand, doi.org/10.5751/ES03862-160123. and flow: A conceptual framework for analysing ecosystem service provision and Pleasant, M.M., Gray, S.A., Lepczyk, C., Fernandes, A., Hunter, N., Ford, D., 2014. delivery. Ecol. Complexity 15, 114–121. https://doi.org/10.1016/j. Managing cultural ecosystem services. Ecosystem Services 8 (November), 141–147. ecocom.2013.07.004. https://doi.org/10.1016/j.ecoser.2014.03.006. Willcock, S., Hooftman, D., Sitas, N., O’Farrell, P., Hudson, M.D., Reyers, B., Ramírez-Gómez, S.O.I., Torres-Vitolas, C.A., Schreckenberg, K., Honzák, M., Cruz- Eigenbrod, F., Bullock, J.M., 2016. Do ecosystem service maps and models meet Garcia, G.S., Willcock, S., Palacios, E., Pérez-Miñana, E., Verweij, P.A., Poppy, G.M., stakeholders’ needs? A preliminary survey across sub-Saharan Africa. Ecosyst. Serv. 2015. Analysis of ecosystem services provision in the Colombian Amazon using 18, 110–117. https://doi.org/10.1016/j.ecoser.2016.02.038. participatory research and mapping techniques. Ecosyst. Serv. 13, 93–107. https:// Willemen, L., Cottam, A.J., Drakou, E.G., Burgess, N.D., 2015. Using social media to doi.org/10.1016/j.ecoser.2014.12.009. measure the contribution of red list species to the nature-based tourism potential of Rea, A.W., Munns, W.R., 2017. The value of nature: Economic, intrinsic, or both? african protected areas. PLoS ONE 10 (6), 1–14. https://doi.org/10.1371/journal. Integrated Environmental Assessment and Management 13 (5), 953–955. https:// pone.0129785. doi.org/10.1002/ieam.1924. Winder, S., Lee, H., Seo, B., Lia, E., Wood, S., 2021. Landscape scale insights into Ricaurte, L.F., Olaya-Rodríguez, M.H., Cepeda-Valencia, J., Lara, D., Arroyave-Suárez, J., recreational activities derived from image content and open source classifiers. Max Finlayson, C., Palomo, I., 2017. Future impacts of drivers of change on wetland Ecosystem Service Partnership Conference Europe. ecosystem services in Colombia. Global Environ. Change 44 (April), 158–169. Wolff, M., 2021. Pushing the perception button - Unlocking barriers to recreational https://doi.org/10.1016/j.gloenvcha.2017.04.001. benefits of urban green spaces (Accepted). Ecol. Soc. Ricaurte, L.F., Wantzen, K.M., Agudelo, E., Betancourt, B., Jokela, J., 2014. Participatory Wood, S.A., Guerry, A.D., Silver, J.M., Lacayo, M., 2013. Using social media to quantify rural appraisal of ecosystem services of wetlands in the Amazonian Piedmont of nature-based tourism and recreation. Sci. Rep. 3 https://doi.org/10.1038/ Colombia: Elements for a sustainable management concept. Wetlands Ecol. Manage. srep02976. 22 (4), 343–361. https://doi.org/10.1007/s11273-013-9333-3. Yoshimura, N., Hiura, T., 2017. Demand and supply of cultural ecosystem services: Use Richards, D.R., Friess, D.A., 2015. A rapid indicator of cultural ecosystem service usage of geotagged photos to map the aesthetic value of landscapes in Hokkaido. Ecosyst. at a fine spatial scale: Content analysis of social media photographs. Ecol. Ind. 53, Serv. 24, 68–78. https://doi.org/10.1016/j.ecoser.2017.02.009. 187–195. https://doi.org/10.1016/j.ecolind.2015.01.034. 13