FACULTAD DE CIENCIAS AMBIENTALES CARRERA PROFESIONAL DE INGENIERÍA AMBIENTAL “ON THE USE OF HEDONIC PRICE INDICES TO UNDERSTAND ECOSYSTEM SERVICE PROVISION FROM URBAN GREEN SPACE IN FIVE LATIN AMERICAN MEGACITIES” “Sobre el Uso de Índices de Precios Hedónicos para Comprender la Provisión de Servicios Ecosistémicos de las Áreas Verdes Urbanas en Cinco Megaciudades de América Latina” Tesis para optar el Título profesional: INGENIERA AMBIENTAL Presentado por: URSULA LORET DE MOLA LEÓN Asesor: PhD. BRENTON LADD LIMA – PERÚ 2018 FACULTAD DE CIENCIAS AMBIENTALES CARRERA PROFESIONAL DE INGENIERÍA AMBIENTAL “ON THE USE OF HEDONIC PRICE INDICES TO UNDERSTAND ECOSYSTEM SERVICE PROVISION FROM URBAN GREEN SPACE IN FIVE LATIN AMERICAN MEGACITIES” “Sobre el Uso de Índices de Precios Hedónicos para Comprender la Provisión de Servicios Ecosistémicos de las Áreas Verdes Urbanas en Cinco Megaciudades de América Latina” Tesis para optar el Título profesional: INGENIERA AMBIENTAL Presentado por: URSULA LORET DE MOLA LEÓN Asesor: PhD. BRENTON LADD LIMA – PERÚ 2018 DEDICATORIA Deo omnis gloria AGRADECIMIENTO Agradezco a Brenton Ladd, por la gran oportunidad de aprender e investigar con él; por el apoyo, la confianza y todo el esfuerzo sobrehumano para que la titulación sea una realidad. A todos los que permitieron la publicación de esta investigación con sus aportes, comentarios y revisiones: Sandra Duarte, Nils Borchard, Ruy Anaya La Rosa y Brian Zutta. Así también a la Dirección General de Investigación de la Universidad, Percy Mayta, por el pago de la publicación. A mi familia, amigos, y en especial a María Ignacia. ÍNDICE RESUMEN ABSTRACT I. ARTICULO CIENTIFICO 1 On the Use of Hedonic Price Indices to Understand Ecosystem Service Provision from Urban Green Space in Five Latin American Megacities 2 1. Introduction 3 2. Methods 4 2.1. Characteristics of the Cities Used in the Comparison of Hedonic Price Indices Across Cities 4 2.2. Green Space Quantification across the 5 Megacities 5 2.3. Socio-Economic Variable Estimation across the Five Megcities 6 2.4. Case Study: The Effect of Landscape Ecology Variables on Hedonic Price Indices 6 2.5. Statistical Analysis 7 3. Results 7 3.1. Comparison of Hedonic Price Indices across the Megacities 7 3.2. Case Study: The Effect of Landscape Ecology Variables on Hedonic Price Indices 8 4. Discussion and Conclusions 9 4.1. General Considerations 9 4.2. Methodological Reflections 9 4.3. Relevance of the Results for Urban Planning Strategies 10 Appendix A 11 Appendix B 12 References 12 II. VERSION INICIAL 18 Quantifying the impact of urban green space on real estate prices in five mega cities of Latin America 19 1. Introduction 19 2. Materials and Methods 20 2.1 Study area 20 2.2. Data 20 2.3. Urban green space 21 2.4. Statistical analysis 22 3. Results 22 4. Discussion 23 References 24 III. CARTAS DE PRESENTACIÓN Y RESPUESTAS A LOS REVISORES 26 IV. VERSION EN ESPAÑOL 45 Sobre el Uso de Índices de Precios Hedónicos para Comprender la Provisión de Servicios Ecosistémicos de las Áreas Verdes Urbanas en Cinco Megaciudades de América Latina 46 1. Introducción 46 2. Métodos 47 2.1. Características de las ciudades empleadas en la comparación de índices de precios hedónicos a través de las ciudades 47 2.2. Cuantificación de las áreas verdes a través de las 5 megaciudades 49 2.3. Estimación de variables socio-económicas a través de las cinco megaciudades 50 2.4. Caso de estudio: Efecto de las variables de ecología del paisaje en los índices de precios hedónicos 50 2.5. Análisis estadístico 50 3. Resultados 51 3.1. Comparación de los índices de precios hedónicos a través de las megaciudades 51 3.2. Caso de Estudio: Efecto de las variables de ecología del paisaje en los índices de precios hedónicos 52 4. Discusión y Conclusiones 53 4.1. Consideraciones generales 53 4.2. Reflexiones metodológicas 53 4.3. Relevancia de los resultados para las estrategias de planificación urbana 54 VI. ABREVIATURAS 55 VII. ANEXOS 56 ANEXO 1. Certificado de publicación por revista Forests 57 ANEXO 2. Correo electrónico – notificación de aceptación de publicación 58 ANEXO 3. Constancia de pago de publicación 59 RESUMEN Las megaciudades de América Latina (AL) se enfrentan a enormes desafíos para brindar bienestar a millones de personas que viven en ellas. Las altas tasas de urbanización y la limitada capacidad administrativa de las ciudades de AL para planificar y controlar el crecimiento urbano han conducido a un déficit crítico de áreas verdes y, por ende, a resultados no óptimos en términos de sostenibilidad urbana. Este estudio busca evaluar la posibilidad de emplear los precios inmobiliarios para proveer una estimación del valor monetario de los servicios ecosistémicos proporcionados por las áreas verdes urbanas, en cinco megaciudades de América Latina: Bogotá, Buenos Aires, Lima, Ciudad de México y Santiago de Chile. Utilizando imágenes de Google Earth para cuantificar las áreas verdes y el análisis de regresión múltiple, evaluamos el impacto de las áreas verdes urbanas, las tasas de criminalidad, la densidad de negocios y la densidad poblacional en los precios inmobiliarios en las cinco megaciudades mencionadas. Además, para un subconjunto de datos (Lima y Buenos Aires) analizamos los efectos de las variables de ecología del paisaje (tamaño del parche de área verde, conectividad, etc.) en los precios de bienes raíces. Ello, para facilitar una primera idea de cómo los atributos ecológicos de las áreas verdes urbanas pueden determinar el nivel de provisión de servicios ecosistémicos en diferentes contextos urbanos en América Latina. Los resultados muestran una fuerte relación positiva entre la presencia de áreas verdes urbanas y los precios inmobiliarios. Las áreas verdes explican el 52% de la variabilidad en los precios inmobiliarios en las cinco megaciudades estudiadas. La densidad poblacional, la densidad de negocios y la delincuencia solo representaron un impacto menor en los precios inmobiliarios. Nuestro análisis de las variables de ecología del paisaje en Lima y Buenos Aires muestra que la relación entre las áreas verdes y el precio inmobiliario es específica del contexto; requiriendo mayor investigación para comprender cuándo y dónde los atributos ecológicos de las áreas verdes afectan los precios inmobiliarios. Todo ello, necesario para que los administradores de las áreas verdes urbanas en las ciudades de AL puedan optimizar su configuración ecológica y así maximizar la provisión de servicios ecosistémicos de las áreas verdes, a menudo limitadas. Palabras clave: Áreas verdes urbanas; precios inmobiliarios; megaciudades de América Latina; Google Earth; Índices de Precios Hedónicos ABSTRACT Latin American (LA) megacities are facing enormous challenges to provide welfare to millions of people who live in them. High rates of urbanization and limited administrative capacity of LA cities to plan and control urban growth have led to a critical deficit of urban green space, and therefore, to sub-optimal outcomes in terms of urban sustainability. This study seeks to assess the possibility of using real estate prices to provide an estimate of the monetary value of the ecosystem services provided by urban green space across five Latin American megacities: Bogota, Buenos Aires, Lima, Mexico City and Santiago de Chile. Using Google Earth images to quantify urban green space and multiple regression analysis, we evaluated the impact of urban green space, crime rates, business density and population density on real estate prices across the five mentioned megacities. In addition, for a subset of the data (Lima and Buenos Aires) we analyzed the effects of landscape ecology variables (green space patch size, connectivity, etc.) on real estate prices to provide a first insight into how the ecological attributes of urban green space can determine the level of ecosystem service provision in different urban contexts in Latin America. The results show a strong positive relationship between the presence of urban green space and real estate prices. Green space explains 52% of the variability in real estate prices across the five studied megacities. Population density, business density and crime had only minor impacts on real estate prices. Our analysis of the landscape ecology variables in Lima and Buenos Aires also show that the relationship between green space and price is context-specific, which indicates that further research is needed to better understand when and where ecological attributes of green space affect real estate prices so that managers of urban green space in LA cities can optimize ecological configuration to maximize ecosystem service provision from often limited green spaces. Keywords: Urban green space; real estate prices; Latin American megacities; Google Earth; hedonic price indices I. ARTICULO CIENTIFICO Article On the Use of Hedonic Price Indices to Understand Ecosystem Service Provision from Urban Green Space in Five Latin American Megacities Ursula Loret de Mola 1,*, Brenton Ladd 1,2, Sandra Duarte 3, Nils Borchard 4,5, Ruy Anaya La Rosa 6 and Brian Zutta 7 ID 1 Facultad de Ciencias Ambientales, Universidad Científica del Sur, Lima 15067, Perú; brenton.ladd@gmail.com 2 School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia 3 Departamento de Medio Ambiente y Sustentabilidad, Universidad Nacional de la Patagonia Austral, Rio Gallegos 9400, Santa Cruz, Argentina; s.duarteguardia@gmail.com 4 Institute of Geography, Soil Science/Soil Ecology, Ruhr-University Bochum, Universitätsstrasse 150, 44801 Bochum, Germany; nils.borchard@rub.de 5 Center for International Forestry Research, Jalan CIFOR, Situ Gede, Sindang Barang, Bogor 16115, Indonesia 6 Starfish Initiatives, Armidale, NSW 2350, Australia; ruy@biochar.international 7 Ministry of the Environment, National Forest Conservation Program for Climate Change Mitigation, Lima 15076, Perú; brian.zutta@gmail.com * Correspondence: eloretdemola.9@hotmail.com; Tel.: +51-998-019-990 Received: 7 October 2017; Accepted: 23 November 2017; Published: 5 December 2017 Abstract: Latin American (LA) megacities are facing enormous challenges to provide welfare to millions of people who live in them. High rates of urbanization and limited administrative capacity of LA cities to plan and control urban growth have led to a critical deficit of urban green space, and therefore, to sub-optimal outcomes in terms of urban sustainability. This study seeks to assess the possibility of using real estate prices to provide an estimate of the monetary value of the ecosystem services provided by urban green space across five Latin American megacities: Bogota, Buenos Aires, Lima, Mexico City and Santiago de Chile. Using Google Earth images to quantify urban green space and multiple regression analysis, we evaluated the impact of urban green space, crime rates, business density and population density on real estate prices across the five mentioned megacities. In addition, for a subset of the data (Lima and Buenos Aires) we analyzed the effects of landscape ecology variables (green space patch size, connectivity, etc.) on real estate prices to provide a first insight into how the ecological attributes of urban green space can determine the level of ecosystem service provision in different urban contexts in Latin America. The results show a strong positive relationship between the presence of urban green space and real estate prices. Green space explains 52% of the variability in real estate prices across the five studied megacities. Population density, business density and crime had only minor impacts on real estate prices. Our analysis of the landscape ecology variables in Lima and Buenos Aires also show that the relationship between green space and price is context- specific, which indicates that further research is needed to better understand when and where ecological attributes of green space affect real estate prices so that managers of urban green space in LA cities can optimize ecological configuration to maximize ecosystem service provision from often limited green spaces. Keywords: Urban green space; real estate prices; Latin American megacities; Google Earth; hedonic price indices Forests 2017, 8, 478; doi:10.3390/f8120478www.mdpi.com/journal/forests Forests 2017, 8, 478 3 of 15 1. Introduction Between 1700 and 2000, 55% of the Earth’s ice free land cover was transformed by human activities, leaving less than 45% of the terrestrial biosphere natural or semi natural [1]. Also, humans have changed the way they use the environment and their distribution within it. World population has gone from living mainly on semi natural lands in 1700 [2], to living mostly in dense settlements (cities) by 2016 [3]. Nowadays, cities are home for more than half of the world’s population [4], and cities are expanding on average at twice the rate of the human population [5–7]. In Latin America (LA), three-quarters of the population already lives in cities [8], making it one of the most urbanized regions in the world. Moreover, LA cities, and megacities increasingly play a key role in the economies of the region [9]. However, LA cities can also be characterized by deep social and spatial segregation, crime, income inequality, and poverty [10]. High population densities and the high concentration of human activity in LA megacities have led to a number of negative environmental impacts [11] and there are significant challenges in terms of meeting the demand for new physical infrastructure, which is often achieved at high social and/or-environmental costs [12]. Many city planners and policy makers consider urban green space and vacant lots as potential land to be converted to infrastructure [13,14] without taking into account the fact that cities depend on the ecosystem services that urban green space provides to sustain human well-being [15,16]. Urban green space, defined as vegetated natural and human-modified outdoor spaces [17] including parks and urban forests, greenways, trails, community gardens, street trees, cemeteries, and others [18], occurs as patches embedded in the urban matrix, where its connectivity and continuity is often endangered by other land allocation priorities [19]. Thus, cities in Latin America are characterized by a critical deficit of urban green space [8], which impairs human well-being [20]. Urban green spaces provide both environmental and social benefits, as they help to ameliorate several problems that occur in cities by supplying numerous ecosystem services. Ecosystem services are the direct and indirect contributions to human well-being, in this case, from urban ecosystems and their components [15] and their provision is related to an increased quality of life [21] and urban resilience [22]. Urban green spaces may improve air quality by filtration of pollutants, regulate water flux and urban temperature, reduce the heat island effect generated by concrete and combustion motors as well as reduce noise pollution [23,24]. Urban green space also improves the mental and physical health of citizens, and supports social interactions [18,25–27]. However, these benefits have a non-market price, so that they cannot be traded in an existing market [28], leading to insufficient consideration of green spaces in public urban-planning policies. One of the main challenges that LA urban planners face in order to achieve welfare for millions of urban residents in LA megacities is related to large, uncontrolled and informal urban development that places pressure on the provision of basic services, increases a city’s vulnerability and has a number of negative environmental consequences [29–31]. This informal growth also tends to occur independently and apart from formal urban expansion which leads to the consolidation of spatial segregation socioeconomically [32]. Green spaces are important in urban areas as they support ecological integrity of cities, provide ecosystem services, and improve the livability of cities [18,24,25], which clearly needs to be taken into account in urban planning activities [33,34]. If the economic value of urban green space could be demonstrated through a premium on real estate, the importance of the ecosystem services provided by urban green space would be reinforced in the political decision-making process [35]. This is important because financing public infrastructure and public services depends heavily on governmental institutional arrangements [36]. One approach to quantify the value of ecosystem services is through the use of hedonic price indices [37]. Hedonic price indices are based on correlations between prices in existing markets (i.e., the real estate market) and specific ecosystem services (i.e., air quality) or bundles of ecosystem services, as, for example, provided by urban green spaces [18,25]. An open question for the LA region is whether hedonic price indices can be used to estimate the value of ecosystem service provision from urban green space in its megacities. Another open question for LA cities is how the ecological attributes of Forests 2017, 8, 478 4 of 15 urban green space might impact the capacity of urban green spaces to provide ecosystem services. Here, we provide some insight into these two issues for LA cities. We specifically test the following hypotheses: (1) hedonic price indices can provide reliable estimates of the value of ecosystem service bundles generated by urban green space across Latin American megacities, and (2) the information that we can obtain from hedonic price indices may be context specific and vary across cities. 2. Methods 2.1. Characteristics of the Cities Used in the Comparison of Hedonic Price Indices Across Cities This study is focused on LA megacities (as opposed to cities more generally) due to the high concentration of the LA population in megacities compared to other regions of the world [38]. According to UN Habitat (2012), there are eight megacities in LA: Buenos Aires, Mexico City, Rio de Janeiro and São Paulo (with more than 10 million inhabitants) and Belo Horizonte, Bogota, Lima and Santiago (with populations approaching 10 million) [8]. Among these eight we were able to obtain real estate prices at district level for the following five megacities: Bogotá in Colombia, Buenos Aires in Argentina, Lima in Peru, Mexico City in Mexico and Santiago in Chile (AppendixA). Although these five megacities present common urban development challenges, the intensity of specific aspects related to sustainability and resilience changes from one city to another [10] (Table1). Bogotá is located in the center of Colombia, on the eastern flank of the Andes, at 2625 m above sea level [39]. The city has an extension of 1637 km2 [40] and is politically divided into 20 localidades [26]. Bogota has an annual population growth rate of 1.3%, with a mean population density of 4876 inhabitants/km2 [41]. It has a relative compact structure with high consolidation of population density, with an increased concentration of informal development in peripheral areas [42]. Growth of informal settlements and pronounced income inequality levels are the most important challenges in Bogota. The estimation of green space per inhabitant is 10 m2 [43]. Buenos Aires is located in the central-eastern region of the country, on the western shore of the La Plata river, on the Pampean plain, at 25 m above sea level [44]. The city’s extension is 204 km2 [45]. Buenos Aires is officially divided into 48 barrios; however, the political and administrative management of the city is distributed across fifteen comunas that, in most cases, cover more than one Buenos Aires barrio [44]. It has an annual rate of population increase of 1.5%, with a mean population density of 14,970 inhabitants/km2 [45]. Buenos Aires presents pronounced socio-spatial differentiation with a strong suburban and peri-urban growth [29]. Growth of informal and precarious housing, poverty, inequality, and crime are the most urgent challenges in Buenos Aires [10]. The estimation of green space per inhabitant is 6 m2 [44]. Lima is located on the coast and in the center of the country, on the shores of the Pacific Ocean and is limited by the coastal desert and formal construction of the city, which has occurred mostly on ex agricultural land on flood plains of three rivers: Chillón, Rímac and Lurín. Lima is 154 m above sea level [46]. The city has an extension of 2812 km2 and is divided into 43 distritos [47]. Lima’s annual population growth rate is 1.6% [48] and its population density is 3328 inhabitants/km2 [49]. Lima is a highly segregated metropolitan agglomeration with deep contrasts between high income and low- income sectors of the population [50]. In Lima there is a large amount of informal urban growth and an unsatisfied demand for basic services like drinking water, transport, and housing [10]. The city has 3 m2 of green space per inhabitant [51]. México City is located in the Mexican Valley at 2240 m above sea level [52]. The city has an extension of 1485 km2 and is divided into 16 delegaciones [53]. Mexico city’s annual population growth rate is 0.3% and its population density is 5967 inhabitants/km2 [54]. Growing insecurity, social-spatial fragmentation and precarious housing conditions are the most important social challenges in Mexico City [10]. Mexico City has 13 m2 of green space per inhabitant [55]. Santiago is located in the Santiago valley surrounded by the Andes, at 520 m above sea level [56]. The city has an extension of 640 km2 and is politically divided into 32 comunas [57]. It has an annual Forests 2017, 8, 478 5 of 15 population growth rate of 1.0% and a population density of 2304 inhabitants/km2 [58]. Santiago’s housing policies have reduced informal housing issues; nevertheless, these policies have led to profound social segregation [10,59]. Santiago has 4 m2 of green space per inhabitant. Table 1. Selected Latin American megacities and respective relevant information. Bogota Buenos Aires Lima Mexico City Santiago Country Colombia Argentina Peru Mexico Chile Mean altitude (AMSL) 2625 25 154 2240 520 Extension (km2) 1637 204 2812 1485 640 City administrative divisions 20 localidades 48 barrios 43 distritos 16 delegaciones 32 comunas Green space (m2/inhab) 10 6 3 13 4 Mean annual precipitation (mm) 818 1040 16 749 390 Mean annual temperature (◦C) 13.5 16.8 18.8 15.7 14.4 Population growth rate (%) 1.3 1.5 1.6 0.3 1.0 Informal settlement Lack of basic Spatial-Social Most Significant Challenge Insecurity growth segregation Insecurity services 2.2. Green Space Quantification across the 5 Megacities LA megacities occupy different amounts of physical space and have been built over different types of original land cover. Thus, three characteristics were considered to define the districts included in this study: (1) percentage of rural population; (2) location; and (3) size. Districts that had more than 50% rural population and/or were predominantly rural or adjacent to the sea shore were excluded. Within and between the five megacities, districts sizes are different. San Telmo in Buenos Aires is the smallest district with an area size of 1.2 km2; and Tlalpan in Mexico City is the largest district with 312 km2. Districts smaller than 3 km2 and/or larger than 100 km2 were excluded. To create a comparable sample, ten districts per city were randomly selected from the districts that were larger than 3 km2 and smaller than 100 km2. Google Earth Pro (version 7.3) was used to obtain land cover data with images from 2013. Total area sampled per district was 2.5 km2, divided into five randomly selected polygons of 0.5 km2, which accounted for the majority of the area in the majority of the districts sampled. In summary, this study considered five polygons per district equivalent to 50 polygons per city, with a total number of 250 polygons (Figure1). Green space measurements were established at an altitude of 500 m above ground level. Urban green spaces were defined as areas covered with any type of vegetation as described by Wolch et al. (2014) [18]. Thus, all types of vegetation cover were sampled inside all polygons, from single trees to urban forests to measure the total availability of green space including public and private spaces. Figure 1. Cont. Forests 2017, 8, 478 6 of 15 Figure 1. 250 sampled polygons distributed within five megacities. (a) Bogota; (b) Buenos Aires; (c) Lima; (d) México City; (e) Santiago. Image Landsat/Copernicus downloaded Google Earth. 2.3. Socio-Economic Variable Estimation across the Five Megcities Population Density: This variable was estimated as the number of inhabitants per km2 of each district area in 2013 (AppendixB). This variable was included in the analysis across cities because Garcia & Riera (2003) have shown that individuals are willing to pay in order to live in less densely populated suburbs [60]. Other studies also have shown that people prefer neighborhoods with low population density and low dwelling-unit density [61,62]. Business Density: This variable is the ratio resulting from dividing the total number of businesses by total district area in 2013 (AppendixB). This variable shows the probable relationship between a more economically active area and willingness to pay to live in said area. According to Des Rosiers et al. (2000) and Yu et al. (2012), there is a positive relationship between real estate prices and proximity to shopping centers, suggesting that the attractiveness of commercial facilities impacts on households’ decisions and translates into a higher demand, and therefore higher prices and rent [63,64]. Crime Density: This variable is calculated as the number of crimes recorded per km2 in each district in 2013 (AppendixB). Previous findings suggest that there is an important relationship between crime rates and property values. Ihlanfeldt & Mayock (2010) suggest that home buyers are willing to pay nontrivial premiums for housing located in neighborhoods with less aggravated assault, robbery and crime [65]. Indeed, crime rate reduction has an immediate benefit on real estate prices but also, benefits that are derived over a 4–6 year period [66]. 2.4. Case Study: The Effect of Landscape Ecology Variables on Hedonic Price Indices Buenos Aires and Lima were chosen as case studies for a more profound analysis in which we assessed how landscape ecology variables (mean patch size, patch connectivity, etc.) affect real estate prices in an effort to obtain a deeper understanding of how and when differences in the ecological attributes of green space can impact the potential for ecosystem service provision from urban green space. In this dataset, Buenos Aires and Lima represent the extremes of climatic conditions with the lowest mean annual precipitation in Lima (16 mm) and highest in Buenos Aires (1040 mm). Forests 2017, 8, 478 7 of 15 Satellite images of Buenos Aires and Lima from March 2013 were downloaded from Google Earth Pro (Version 7.3, Google Inc., Mountain View, CA, USA) and processed in ArcGIS 10.3 (ESRI, Redlands, CA, USA). Images were first georeferenced and projected in geographic form, and then areas of interest were extracted from the previous five randomly selected polygons of 0.5 km2 per district. We used the Segment Mean Shift from the Spatial Analyst toolbox to segment the images into objects of at least 10 pixels. We then manually selected objects that corresponded to our definition of urban green areas or “green patch” within the areas of interest. The area of each patch was calculated using the calculate geometry function in the attribute table of ArcGIS. Average Nearest Neighbor distance between green patches were calculated using Euclidean Distance in the Spatial Statistics toolbox. 2.5. Statistical Analysis To homogenize data and simplify interpretation for the analysis of hedonic price indices across cities, independent and dependent variables were normalized prior to statistical analysis with the following transformation: X ∗ i = Xi /Xmax where X ∗ i is each dependent variable after normalization. Xi is the variable prior to normalization and Xmax is the maximum value obtained for the variable in the respective city. For the case studies (Lima and Buenos Aires) no normalization of data was performed before analysis since differing scales are not an issue within single cities. Transformed and non-transformed data were analyzed using multiple linear regression using XLStat (Version 2014.5, Addin Soft, Paris, France). 3. Results 3.1. Comparison of Hedonic Price Indices across the Megacities From a multiple regression with four independent variables, we obtained a model with admissible accuracy (R2 = 0.62; Table2A; Figure2). The η2 values indicate that urban green space was the most important predictive variable explaining 52% of the variability in real estate prices across the megacities considered in this analysis. Business density also had a modest influence on real estate prices with a η2 value of 10, whereas crime rate and population density had non-relevant impacts (Table2). However, there was also considerable error with unexplained variability accounting for 39% of the sum of squares; therefore, a complete understanding of the factors that determine real estate pricing was not obtained from the analysis across cities. Table 2. (A) Summary statistics for the multiple regression models in which (A) real estate prices were correlated with four variables: green space, population density, business density and crime rate, (B) green space and green patch attributes vs. real estate prices in Lima, and (C) green space and green patch attributes vs. real estate prices in Buenos Aires. Variables: Av/at = green space/total area, TGS = Total green space, TNP = Total number of patches, MPS = Mean patch size, LPS = Largest patch size, ANN = average nearest neighbor, PopDen = Population Density, BusDen = Business Density, CrimDen = Crime Rate. Source df SS MS F Pr > F η2 (A) Analysis hedonic price indices across cities AV/AT * 1 1.14 1.14 60.16 <0.0001 51.5 Pop Den * 1 0.00 0.00 0.03 0.87 0.0 Bus Den * 1 0.22 0.22 11.5 0.002 9.8 Crim Den * 1 0.00 0.00 0.24 0.63 0.2 Error 45 0.85 38.5 Corrected Total 49 2.21 100 Forests 2017, 8, 478 8 of 15 Table 2. Cont. Source df SS MS F Pr > F η2 (B) Analysis of green space and patch attributes vs. real estate prices in Lima TGS 1 2,663,521.6 2,663,521.6 42.9 0.0 83.8 LPS 1 223,703.7 223,703.7 3.6 0.1 7.0 TNP 1 18,053.0 18,053.0 0.3 0.6 0.6 MPS 1 21,708.4 21,708.4 0.3 0.6 0.7 ANN 1 2851.4 2851.4 0.0 0.8 0.1 Error 4 248,443.6 62,110.9 7.8 Corrected Total 9 3,178,281.6 100.0 (C) Analysis of green space and patch attributes vs. real estate prices in Buenos Aires TGS 1 213,044.0 213,044.0 1.7 0.3 12.5 LPS 1 393,208.9 393,208.9 3.1 0.2 23.0 TNP 1 88,816.9 88,816.9 0.7 0.5 5.2 MPS 1 131,095.6 131,095.6 1.0 0.4 7.7 ANN 1 375,692.7 375,692.7 3.0 0.2 22.0 Error 4 508,002.4 127,000.6 29.7 Corrected Total 9 1,709,860.5 100 df, degrees of freedom; SS, sum of squares; MS, mean squares; F, F-test; Pr, probability, η2 percentage of variation of the R2 explained for each independent variable, η2 is obtained from the SS partial value between the SS total value per 100. * Variables after transformation (normalization) used for multiple regressions for first analysis. Normalization: X ∗ i = Xi /Xmax. Xi: variable prior to normalization. Xmax: maximum value obtained for the variable in the city in which the district occurs. 1.5 Lima Santiago Bogotá 1:1 México City Buenos Aires 1.0 0.5 0.0 0.0 0.5 1.0 1.5 Predicted Price Normalized Figure 2. Relationship between measured prices and predicted price at district level. Predicted values obtained via multiple regression: Predicted Price Normalized (NOR) = 0.35 + 0.71*av/at NOR − 0.053*Population Density NOR + 0.34*Business Density NOR − 0.061*Crime Density NOR. See Table2 for details. Diagonal line indicates the 1:1 relationship. The normalized data used in this analysis adhere closely to the 1:1 line (which indicates close correspondence between the measured and predicted values and shows the robustness of the multiple regression model. 3.2. Case Study: The Effect of Landscape Ecology Variables on Hedonic Price Indices Results from the case studies (Lima and Buenos Aires) provided a strong hint that the effectiveness of hedonic price indices for constraining the value of the ecosystem services provided by urban green space can be context specific. The Lima data showed a strong relationship between the amount of green space and real estate prices, in which the amount of green space explained 83.8% of the variability in real estate prices (Table2B). The positive and significant impact of total green space (TGS) further confirmed that the Measured Price Normalized Forests 2017, 8, 478 9 of 15 amount of urban green spaces is strongly valued by Lima’s residents. Largest patch size (LPS) of green space in Lima had a minor impact on real estate prices (η2 = 7), whereas mean patch size (MPS), total number of patches (TNP) and the average nearest neighbor (ANN) or connectivity between patches were irrelevant. In contrast, for Buenos Aires, the most important variables for explaining variation in real estate prices was largest patch size (LPS) with a η2 value of 23 (Table2C), followed by the average nearest neighbor (ANN) or connectivity with a η2 value of 22. Total green space (TGS) had a modest impact explaining 12.5% of variance in real estate prices. Mean patch size (MPS) of urban green space in Buenos Aires also had a small impact explaining 7.7% of the variance in real estate prices. In Buenos Aires, there was a considerable error with unexplained variability accounting for 29.7% of the variance. In contrast the error term for Lima was 7.8%. 4. Discussion and Conclusions 4.1. General Considerations The results of this study reinforce prior findings showing that urban green space is an important determinant of real estate prices [35,67–69], and that hedonic price indices can provide a robust estimate of the value of the ecosystem services provided by urban green spaces across LA cities. However, the unexplained variability in the multiple regression model for the analysis across cities (38.5%) was also significant and could be related to a variety of factors which were not taken into account in this study. Some social variables such as economic status perceptions [26], income dynamics [70], school quality [71] and/or cultural—spiritual values like sense of place, social cohesion [24,72] may provide further insights into the underlying factors that impact real estate prices. In addition, the use of landscape ecology variables in the case study of Lima and Buenos Aires demonstrates that despite regional similarities, the relationship between green space and real estate prices is context-specific, and that the ecological attributes of green space can impact strongly the potential for ecosystem service provision. These preferences could be related to several factors like city growth tendency, government policy or, as we suspect, the physical environment. It seems logical that amount of green space should be more highly prized in a hyper arid city like Lima than in a city like Buenos Aires where rainfall is sufficient to support growth of rainforest. This we believe can explain why we see a more nuanced valuation of green spaces in Buenos Aires focused on spatial patterns of the green space, specifically related to the size of green spaces and the connectivity. However differences in the quality of city planning may also be important in this specific comparison between Buenos Aires and Lima [50,73] and a larger analysis that includes data from more cities could provide further insight. Also, the relevance of ecosystem services differs according to the specific environmental and socio-economic characteristics of a city [24]. For example, urban green spaces can buffer extreme weather events like floods, which may be important for cities like Buenos Aires, but for cities like Lima with little to no rainfall, this ES is not important. Air quality regulation is critical for all megacities, but possibly most important in cities like Santiago and Mexico City where topography favors the concentration of aerosol contaminants. Urban forests stabilize slopes, preventing damage from natural hazards, which is crucial for Bogota, but not important for Buenos Aires, located on a coastal plain. A context-specific classification of ecosystem services in LA urban areas is needed to secure resilience-oriented planning [74] as it will be important to consider the role of culturally specific features of people-nature relationships for increasing the quality of life of LA city residents [26]. 4.2. Methodological Reflections Urban planners require up-to-date and accurate spatial data at a city scale to understand urban dynamics and processes [75]. In Latin America, robust data at city and local scale are difficult to obtain due to limited interest and investment in preliminary research and there is little effort to ensure open access of relevant data. Forests 2017, 8, 478 10 of 15 Our results demonstrate the important role that hedonic price indices can play in helping us to understand how green space generates ecosystem services in different situations (i.e., Buenos Aires vs. Lima). Further research is needed to better understand how when and why ecological attributes of green space impact real estate prices so we can optimize the way we invest in and manage green spaces in LA cities so that our investments in urban green space generate the maximum amount of ecosystem services possible. However, our ability to achieve this will depend heavily on data availability and a common frame for comparisons as we have attempted here across the LA megacities. Improving quality and quantity of data acquisition and availability could significantly improve urban planning processes for Latin American cities, which would be directly in line with the United Nations Sustainable Development Goals aiming to improve good health and well-being and supporting development of sustainable cities and communities (http://www.un.org/sustainabledevelopment/sustainable- development-goals/, accessed on 3 November 2017). 4.3. Relevance of the Results for Urban Planning Strategies High rates of urbanization in Latin America continue to pose direct threats to the preservation of urban green space [76] due to limited administrative capacity in LA cities to plan and control urban growth causing conversion of urban green space into houses and infrastructure. Further, creating new urban green space is becoming increasingly problematic due to a high levels of urban land consolidation [77] and overlapping demand of land for multiple purposes (e.g., urban green space vs. infrastructure) [78]. On the other hand, green space planning is not only guided by urban theories but also from the values people assign to green spaces [79]. Our results demonstrate that the people living in LA megacities value green space significantly, expressing their preferences through the real estate market. However as the case study clearly shows there is significant scope to conduct new research that will enable us to better understand how when and why the ecological attributes of urban green space can help to ensure, maintain or even enhance a range of different ecosystem services [80]. Being able to do more with less in terms of ecosystem service provision from urban green space is especially important for the urban poor in LA cities who generally live in the metropolitan peripheries with critical deficits of urban green space. Acknowledgments: We want to thank the Cámara Peruana de Construcción, Portal Inmobiliario, Cámara Colombiana de la Construcción, Metros Cúbicos, and Reporte Inmobiliario for their collaboration in providing essential data. We acknowledge support from the Publication Funds of Cientifica del Sur University. We also thank the editorial team and the peer reviewers who provided a wealth of constructive feedback that has significantly improved the quality of this paper. Author Contributions: B.L. conceived the research idea. U.L.d.M. obtained the real estate data and measured urban green space availability. B.L. and U.L.d.M. carried out the statistical analysis. B.L. supervised the whole research process and the findings of this work. S.D. processed climatic variables information and provided the maps. N.B. and R.A.L.R. provided critical feedback and helped shape the analysis and manuscript. B.R.Z. defined and performed the GIS analysis for case study section. U.L.d.M. wrote the paper with authors inputs. Conflicts of Interest: The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results. Forests 2017, 8, 478 11 of 15 Appendix A Table A1. Real Estate Data: US$/m2 at district level (2013). Country City Neighborhood Price (US$) Colombia Bogotá Antonio Nariño 870 Colombia Bogotá Barrios Unidos 1493 Colombia Bogotá Bosa 570 Colombia Bogotá Engativa 1228 Colombia Bogotá Fontibon 1263 Colombia Bogotá Kennedy 682 Colombia Bogotá Rafael Uribe Uribe 506 Colombia Bogotá San Cristobal 594 Colombia Bogotá Suba 1555 Colombia Bogotá Teusaquillo 1794 Argentina Buenos Aires Almagro 1830 Argentina Buenos Aires Caballito 1993 Argentina Buenos Aires Mataderos 1648 Argentina Buenos Aires Nueva Pompeya 1245 Argentina Buenos Aires Palermo 2523 Argentina Buenos Aires Parque Chacabuco 1749 Argentina Buenos Aires Parque Patricios 1603 Argentina Buenos Aires Recoleta 2730 Argentina Buenos Aires Saavedra 1995 Argentina Buenos Aires Villa Devoto 1809 Perú Lima Jesus Maria 1679 Perú Lima Los Olivos 1113 Perú Lima Miraflores 2293 Perú Lima Pueblo Libre 1345 Perú Lima San Borja 2148 Perú Lima San Isidro 2355 Perú Lima San Miguel 1504 Perú Lima San Juan Miraflores 914 Perú Lima Surquillo 1559 Perú Lima Villa María del Triunfo 568 México Mexico City Alvaro Obregon 1824 México Mexico City Miguel Hidalgo 2358 México Mexico City Azcapotzalco 1103 México Mexico City Benito Juarez 1712 México Mexico City Coyoacan 1502 México Mexico City Cuauhtemoc 1504 México Mexico City Gustavo A. Madero 998 México Mexico City Iztacalco 915 México Mexico City La Magdalena Contreras 1607 México Mexico City Venustiano Carranza 1081 Chile Santiago Vitacura 2795 Chile Santiago Santiago 1325 Chile Santiago Providencia 2333 Chile Santiago Peñalolen 1599 Chile Santiago Ñuñoa 2041 Chile Santiago Macul 1416 Chile Santiago Las Condes 2486 Chile Santiago La Reina 2134 Chile Santiago La Florida 1297 Chile Santiago Independencia 1089 Forests 2017, 8, 478 12 of 15 Appendix B Table A2. Independent social variables used in the study (2013). All data were divided between the area of each district. Population Business (Number of Security (Number of Country City Neighborhood (Inhabitants) * Businesses) ** Crimes Recorded) *** Colombia Bogotá Antonio Nariño 108,607 4660 1428 Colombia Bogotá Barrios Unidos 236,433 15,932 1502 Colombia Bogotá Bosa 612,754 11,803 1725 Colombia Bogotá Engativa 858,935 29,469 3327 Colombia Bogotá Fontibon 362,167 15,909 2153 Colombia Bogotá Kennedy 1,042,080 28,787 3889 Colombia Bogotá Rafael Uribe Uribe 376,767 8196 1520 Colombia Bogotá San Cristobal 408,477 5646 1285 Colombia Bogotá Suba 1,120,342 36,856 4669 Colombia Bogotá Teusaquillo 149,166 13,266 2316 Argentina Buenos Aires Almagro 93,571 2437 1062 Argentina Buenos Aires Caballito 183,662 5171 1837 Argentina Buenos Aires Mataderos 55,633 1176 797 Argentina Buenos Aires Nueva Pompeya 62,791 1142 1012 Argentina Buenos Aires Palermo 255,358 10,771 5751 Argentina Buenos Aires Parque Chacabuco 109,541 2423 1550 Argentina Buenos Aires Parque Patricios 62,791 1142 1012 Argentina Buenos Aires Recoleta 187,141 8987 2267 Argentina Buenos Aires Saavedra 49,910 1157 561 Argentina Buenos Aires Villa Devoto 49,443 1163 512 Perú Lima Jesus Maria 71,439 13,634 1021 Perú Lima Los Olivos 360,532 32,874 7929 Perú Lima Miraflores 83,649 27,303 2364 Perú Lima Pueblo Libre 76,743 9877 1114 Perú Lima San Borja 111,688 16,424 2634 Perú Lima San Isidro 55,792 19,445 1019 Perú Lima San Miguel 135,226 15,408 3565 Perú Lima San Juan Miraflores 397,113 26,725 4323 Perú Lima Surquillo 92,012 14,293 2318 Perú Lima Villa María del Triunfo 433,861 21,023 2542 México Mexico City Alvaro Obregon 734,290 20,170 10,902 México Mexico City Miguel Hidalgo 380,608 23,724 11,013 México Mexico City Azcapotzalco 410,475 16,928 8561 México Mexico City Benito Juarez 397,446 24,293 12,042 México Mexico City Coyoacan 618,265 22,142 11,826 México Mexico City Cuauhtemoc 536,086 66,587 26,542 México Mexico City Gustavo A. Madero 1,180,559 46,007 21,980 México Mexico City Iztacalco 380,259 16,955 7825 México Mexico City La Magdalena Contreras 242,355 6094 2385 México Mexico City Venustiano Carranza 424,962 30,763 10,337 Chile Santiago Vitacura 87,792 14,519 1382 Chile Santiago Santiago 331,325 56,651 16,459 Chile Santiago Providencia 144,169 44,699 4998 Chile Santiago Peñalolen 240,304 8690 4319 Chile Santiago Ñuñoa 212,163 15,328 4347 Chile Santiago Macul 122,966 5533 2389 Chile Santiago Las Condes 279,760 51,167 4893 Chile Santiago La Reina 101,358 6327 1633 Chile Santiago La Florida 387,352 15,696 8939 Chile Santiago Independencia 80,476 4868 1899 * Secretaría de Planeación—Alcaldía Mayor de Bogotá (Colombia), Dirección General de Estadística y Censos— Gobierno de la Ciudad Autónoma de Buenos Aires (Argentina), Dirección Técnica de Demografía e Indicadores Sociales del Instituto Nacional de Estadística e Informática—INEI (Peru), Consejo Nacional de Población—CONAPO (Mexico), Instituto Nacional de Estadísticas Chile—INE (Chile); ** Cámara de Comercio de Bogotá—CCB (Colombia), Subsecretaria de Trabajo, Industria y Comercio—Ministerio de Desarrollo Económico (Argentina), Instituto Nacional de Estadística e Informática—INEI (Peru), Instituto Nacional de Estadística y Geografía—INEGI (Mexico), Departamento de Estudios Económicos y Tributarios—Subdirección de Estudios del Servicio de Impuestos Internos (Chile); *** Observatorio de Seguridad en Bogota (Colombia), Instituto Superior de Seguridad Publica (Argentina), Observatorio Nacional de Seguridad Ciudadana—OBNASEC (Peru), Dirección General de Politica y Estadística Criminal—Procuraduría General de Justicia del Distrito Federal-PGJ DF (Mexico), Carabineros—Instituto Nacional de Estadísticas Chile—INE (Chile). References 1. Ellis, E.C.; Klein Goldewijk, K.; Siebert, S.; Lightman, D.; Ramankutty, N. Anthropogenic transformation of the biomes, 1700 to 2000. Glob. Ecol. Biogeogr. 2010, 19, 589–606. [CrossRef] Forests 2017, 8, 478 13 of 15 2. Martine, G. State of World Population 2007. Unleashing the Potential of Urban Growth; United Nations Population Fund: New York, NY, USA, 2007. 3. Department of Economic and Social Affairs—United Nations. The World’s Cities in 2016: Data Booklet; Department of Economic and Social Affairs—United Nations: New York, NY, USA, 2016. 4. UN-Habitat. Urbanization and Development: Emerging Futures; UN-Habitat: Nairobi, Kenya, 2016; p. 260. 5. Angel, S.; Parent, J.; Civco, D.L.; Blei, A.; Potere, D. The dimensions of global urban expansion: Estimates and projections for all countries, 2000–2050. Prog. Plan. 2011, 75, 53–107. [CrossRef] 6. Schneider, A.; Friedl, M.A.; Potere, D. Mapping global urban areas using modis 500-m data: New methods and datasets based on ‘urban ecoregions’. Remote Sens. Environ. 2010, 114, 1733–1746. [CrossRef] 7. Seto, K.C.; Fragkias, M.; Güneralp, B.; Reilly, M.K. A meta-analysis of global urban land expansion. PLoS ONE 2011, 6, e23777. [CrossRef] [PubMed] 8. UN-Habitat. The State of Latin American and Caribbean Cities 2012 towards a New Urban Transition; UN-Habitat: Nairobi, Kenya, 2012; p. 194. 9. Cadena, A.; Remes, J.; Manyika, J.; Dobbs, R.; Roxburgh, C.; Elstrodt, H.-P.; Chaia, A.; Restrepo, A. Building Globally Competitive Cities: The Key to Latin American Growth; McKinsey Global Institute: New York, NY, USA, 2011. 10. Rehner, J.; Samaniego, J.; Jordán Fuchs, R. Regional Panorama: Latin America. Megacities and Sustainability; ECLAC: Santiago, Chile, 2010; p. 203. 11. Krausmann, F.; Gingrich, S.; Eisenmenger, N.; Erb, K.-H.; Haberl, H.; Fischer-Kowalski, M. Growth in global materials use, gdp and population during the 20th century. Ecol. Econ. 2009, 68, 2696–2705. [CrossRef] 12. Cerón-Palma, I.; Sanyé-Mengual, E.; Oliver-Solà, J.; Montero, J.-I.; Ponce-Caballero, C.; Rieradevall, J. Towards a green sustainable strategy for social neighbourhoods in latin america: Case from social housing in Merida, Yucatan, Mexico. Habitat Int. 2013, 38, 47–56. [CrossRef] 13. Bomans, K.; Steenberghen, T.; Dewaelheyns, V.; Leinfelder, H.; Gulinck, H. Underrated transformations in the open space—The case of an urbanized and multifunctional area. Landsc. Urban Plan. 2010, 94, 196–205. [CrossRef] 14. Mell, I.C.; Henneberry, J.; Hehl-Lange, S.; Keskin, B. Promoting urban greening: Valuing the development of green infrastructure investments in the urban core of Manchester, UK. Urban For. Urban Green. 2013, 12, 296–306. [CrossRef] 15. The Economics of Ecosystems and Biodiversity (TEEB). Manual for Cities: Ecosystem Services in Urban Management. Available online:http://www.teebweb.org/wp-content/uploads/Study%20and%20Reports/ Additional%20Reports/Manual%20for%20Cities/TEEB%20Manual%20for%20Cities_English.pdf (accessed on 30 October 2017). 16. Chiesura, A. The role of urban parks for the sustainable city. Landsc. Urban Plan. 2004, 68, 129–138. [CrossRef] 17. Wendel, H.E.W.; Zarger, R.K.; Mihelcic, J.R. Accessibility and usability: Green space preferences, perceptions, and barriers in a rapidly urbanizing city in Latin America. Landsc. Urban Plan. 2012, 107, 272–282. [CrossRef] 18. Wolch, J.R.; Byrne, J.; Newell, J.P. Urban green space, public health, and environmental justice: The challenge of making cities ‘just green enough’. Landsc. Urban Plan. 2014, 125, 234–244. [CrossRef] 19. Department of Economic and Social Affairs—United Nations. World Population Prospects the 2012 Revision. Highlights and Advance Tables; Department of Economic and Social Affairs—United Nations: New York, NY, USA, 2013. 20. García-Ayllón, S. Rapid development as a factor of imbalance in urban growth of cities in Latin America: A perspective based on territorial indicators. Habitat Int. 2016, 58, 127–142. [CrossRef] 21. Kabisch, N.; Strohbach, M.; Haase, D.; Kronenberg, J. Urban green space availability in European cities. Ecol. Indic. 2016, 70, 586–596. [CrossRef] 22. McPhearson, T.; Andersson, E.; Elmqvist, T.; Frantzeskaki, N. Resilience of and through urban ecosystem services. Ecosyst. Serv. 2015, 12, 152–156. [CrossRef] 23. Bolund, P.; Hunhammar, S. Ecosystem services in urban areas. Ecol. Econ. 1999, 29, 293–301. [CrossRef] 24. Gómez-Baggethun, E.; Barton, D.N. Classifying and valuing ecosystem services for urban planning. Ecol. Econ. 2013, 86, 235–245. [CrossRef] 25. Sandifer, P.A.; Sutton-Grier, A.E.; Ward, B.P. Exploring connections among nature, biodiversity, ecosystem services, and human health and well-being: Opportunities to enhance health and biodiversity conservation. Ecosyst. Serv. 2015, 12, 1– 15. [CrossRef] 26. Scopelliti, M.; Carrus, G.; Adinolfi, C.; Suarez, G.; Colangelo, G.; Lafortezza, R.; Panno, A.; Sanesi, G. Staying in touch with nature and well-being in different income groups: The experience of urban parks in Bogotá. Landsc. Urban Plan. 2016, 148, 139–148. [CrossRef] 27. Krellenberg, K.; Welz, J.; Reyes-Päcke, S. Urban green areas and their potential for social interaction—A case study of a socio-economically mixed neighbourhood in Santiago de Chile. Habitat Int. 2014, 44, 11–21. [CrossRef] 28. Sengupta, S.; Osgood, D.E. The value of remoteness: A hedonic estimation of ranchette prices. Ecol. Econ. 2003, 44, 91–103. [CrossRef] 29. Inostroza, L.; Tábbita, J.H. Informal urban development in the greater Buenos Aires area: A quantitative-spatial assessment Forests 2017, 8, 478 14 of 15 based on households’ physical features using gis and principal component analysis. Procedia Eng. 2016, 161, 2138–2146. [CrossRef] 30. Irazábal, C. Revisiting urban planning in Latin America and the Caribbean. In Revisiting Urban Planning: Global Report on Human Settlements 2009; UN-Habitat: New York, NY, USA, 2009. 31. Inostroza, L. Informal urban development in Latin American urban peripheries. Spatial assessment in Bogotá, Lima and Santiago de Chile. Landsc. Urban Plan. 2017, 165, 267–279. [CrossRef] 32. Michelini, J.J.; Pintos, P. Metropolitan expansion and new socio-spatial segregation scenarios in contemporary Argentina. The case of Nordelta-Las Tunas (Buenos Aires). Habitat Int. 2016, 54, 40–49. [CrossRef] 33. Romero-Lankao, P.; Gnatz, D.M. Exploring urban transformations in Latin America. Curr. Opin. Environ. Sustain. 2013, 5, 358–367. [CrossRef] 34. Vásquez, A.; Devoto, C.; Giannotti, E.; Velásquez, P. Green infrastructure systems facing fragmented cities in Latin America—Case of Santiago, Chile. Procedia Eng. 2016, 161, 1410–1416. [CrossRef] 35. Nilsson, P. Natural amenities in urban space—A geographically weighted regression approach. Landsc. Urban Plan. 2014, 121, 45–54. [CrossRef] 36. MEA (Millennium Ecosystem Assessment). Ecosystems and Human Well-Being: Current State and Trends; Island Press: Washington, DC, USA, 2005; p. 47. 37. Heal, G. Valuing ecosystem services. Ecosystems 2000, 3, 24–30. [CrossRef] 38. Comision Económica para América Latina y el Caribe. América Latina y el Caribe: Desafíos, Dilemas y Compromisos de una Agenda Urbana Común; Comision Económica para América Latina y el Caribe: Santiago, Chile, 2016; p. 62. 39. Secretaría General Alcaldía Mayor de Bogotá. Available online:http://www.bogota.gov.co/ciudad/ ubicacion(accessed on 3 November 2017). 40. Alcaldía Mayor de Bogotá. Diagnostico de los Aspectos Fisicos, Demograficos y Socioeconomicos; Alcaldía Mayor de Bogotá: Bogotá, Colombia, 2009; p. 175. 41. Bogota Como Vamos. Informe de Calidad de vida de Bogota 2016; Bogota Como Vamos: Bogota, Colombia, 2017; p. 371. 42. Guzman, L.A.; Oviedo, D.; Bocarejo, J.P. City profile: The bogotá metropolitan area that never was. Cities 2017, 60, 202–215. [CrossRef] 43. Observatorio Ambiental de Bogota. Available online:http://oab2.ambientebogota.gov.co/es/indicadores? id=986&v=l(accessed on 3 November 2017). 44. Buenos Aires Ciudad. Available online:http://www.buenosaires.gob.ar/laciudad/ciudad(accessed on 3 November 2017). 45. Dirección General de Estadística y Censos. Anuario Estadístico Ciudad de Buenos Aires 2015; Dirección General de Estadísticay Censos: Buenos Aires, Argentina, 2016; p. 633. 46. Municipalidad de Lima. Available online:http://www.munlima.gob.pe/lima(accessed on 3 November 2017). 47. Instituto Nacional de Estadistica e Informatica. Lima Metropolitana Perfil Socio-Demográfico. Available online:http://proyectos.inei.gob.pe/web/biblioineipub/bancopub/Est/Lib0266/CAP01-04.htm (accessed on 3 November 2017). 48. Instituto Nacional de Estadística e Informática. Día Mundial de la Población—11 de Junio; Instituto Nacional de Estadística e Informática: Lima, Peru, 2015; p. 56. 49. Observatorio Ciudadano Lima Cómo Vamos. Evaluando la Gestión en Lima 2015; Observatorio Ciudadano Lima Cómo Vamos: Lima, Peru, 2016; p. 68. 50. Fernández de Córdova, G.; Fernández-Maldonado, A.M.; Del Pozo, J.M. Recent changes in the patterns of socio-spatial segregation in Metropolitan Lima. Habitat Int. 2016, 54, 28–39. [CrossRef] 51. Observatorio Ciudadano Lima Cómo Vamos. Cómo Vamos en Espacio Público; Observatorio Ciudadano Lima Cómo Vamos: Lima, Peru, 2016; p. 7. 52. Instituto Nacional de Estadística y Geografía (INEGI). Anuario Estadístico y Geográfico de los Estados Unidos Mexicanos 2016; Instituto Nacional de Estadística y Geografía: Aguascalientes, Mexico, 2016; p. 965. 53. Gobierno de Ciudad de México. Available online:http://www.cdmx.gob.mx/cdmx/sobre-nuestra-ciudad (accessed on 3 November 2017). 54. Instituto Nacional de Estadística y Geografía (INEGI). Mujeres y Hombres en México 2016; Instituto Nacional de Estadística y Geografía: Aguascalientes, Mexico, 2016; p. 260. 55. Checa-Artasu, M.M. Las áreas verdes en la ciudad de méxico. Las diversas escalas de una geografía urbana. Rev. Bibliogr. Geogr. Cienc. Soc. 2016, 21, 1–22. 56. Biblioteca del Congres o Nacional de Chile (BCN). Available online:http://www.bcn.cl/siit/nuestropais/ region13/relieve.htm(accessed on 3 November 2017). 57. Instituto Nacional de Estadísticas. División Político—Administrativa y Censal 2007; Instituto Nacional de Estadísticas: Santiago, Chile, 2008; p. 358. 58. Intendencia de la Región Metropolitana. Información Geográfica. Available online:http://www. intendenciametropolitana.gov.cl/informacion_geografica_2.html(accessed on 3 November 2017). 59. Reyes Päcke, S.; Figueroa Aldunce, I.M. Distribución, superficie y accesibilidad de las áreas verdes en santiago de chile. EURE (Santiago) 2010, 36, 89–110. [CrossRef] 60. Garcia, D.; Riera, P. Expansion versus density in barcelona: A valuation exercise. Urban Stud. 2003, 40, 1925–1936. Forests 2017, 8, 478 15 of 15 [CrossRef] 61. Song, Y.; Knaap, G.-J. New urbanism and housing values: A disaggregate assessment. J. Urban Econ. 2003, 54, 218–238. [CrossRef] 62. Dunse, N.; Thanos, S.; Bramley, G. Planning policy, housing density and consumer preferences. J. Prop. Res. 2013, 30, 221–238. [CrossRef] 63. Des Rosiers, F.; Thériault, M.; Villeneuve, P.-Y. Sorting out access and neighbourhood factors in hedonic price modelling. J. Prop. Investig. Financ. 2000, 18, 291–315. [CrossRef] 64. Yu, T.-H.; Cho, S.-H.; Kim, S.G. Assessing the residential property tax revenue impact of a shopping center. J. Real Estate Financ. Econ. 2012, 45, 604–621. [CrossRef] 65. Ihlanfeldt, K.; Mayock, T. Panel data estimates of the effects of different types of crime on housing prices. Reg. Sci. Urban Econ. 2010, 40, 161–172. [CrossRef] 66. Pope, D.G.; Pope, J.C. Crime and property values: Evidence from the 1990s crime drop. Reg. Sci. Urban Econ. 2012, 42, 177–188. [CrossRef] 67. Kong, F.; Yin, H.; Nakagoshi, N. Using gis and landscape metrics in the hedonic price modeling of the amenity value of urban green space: A case study in Jinan city, China. Landsc. Urban Plan. 2007, 79, 240–252. [CrossRef] 68. Morancho, A.B. A hedonic valuation of urban green areas. Landsc. Urban Plan. 2003, 66, 35–41. [CrossRef] 69. Panduro, T.E.; Veie, K.L. Classification and valuation of urban green spaces—A hedonic house price valuation. Landsc. Urban Plan. 2013, 120, 119–128. [CrossRef] 70. Yao, L.; Liu, J.; Wang, R.; Yin, K.; Han, B. Effective green equivalent—A measure of public green spaces for cities. Ecol. Indic. 2014, 47, 123–127. [CrossRef] 71. Nguyen-Hoang, P.; Yinger, J. The capitalization of school quality into house values: A review. J. Hous. Econ. 2011, 20, 30–48. [CrossRef] 72. Pastur, G.M.; Peri, P.L.; Lencinas, M.V.; García-Llorente, M.; Martín-López, B. Spatial patterns of cultural ecosystem services provision in southern Patagonia. Landsc. Ecol. 2016, 31, 383–399. [CrossRef] 73. Himschoot, P.; Areco, M.M. Cambio Climático en Buenos Aires, Riesgo de Desastre y Pobreza Urbana; Agencia de Protección Ambiental del Gobierno de la Ciudad de Buenos Aires: Buenos Aires, Argentina, 2015; p. 49. 74. Frantzeskaki, N.; Kabisch, N.; McPhearson, T. Advancing urban environmental governance: Understanding theories, practices and processes shaping urban sustainability and resilience. Environ. Sci. Policy 2016, 62, 1–6. [CrossRef] 75. Parés-Ramos, I.; Álvarez-Berríos, N.; Aide, T. Mapping urbanization dynamics in major cities of Colombia, Ecuador, Perú, and Bolivia using night-time satellite imagery. Land 2013, 2, 37–59. [CrossRef] 76. Inostroza, L.; Baur, R.; Csaplovics, E. Urban sprawl and fragmentation in latin america: A dynamic quantification and characterization of spatial patterns. J. Environ. Manag. 2013, 115, 87–97. [CrossRef] [PubMed] 77. Haaland, C.; van den Bosch, C.K. Challenges and strategies for urban green-space planning in cities undergoing densification: A review. Urban For. Urban Green. 2015, 14, 760–771. [CrossRef] 78. Perrotti, D.E.; Sánchez, R. La Brecha de Infraestructura en América Latina y el Caribe; NU. CEPAL. División de Recursos Naturales e Infraestructura: Santiago, Chile, 2011. 79. Ives, C.D.; Oke, C.; Hehir, A.; Gordon, A.; Wang, Y.; Bekessy, S.A. Capturing residents’ values for urban green space: Mapping, analysis and guidance for practice. Landsc. Urban Plan. 2017, 161, 32–43. [CrossRef] 80. Swanwick, C. Society’s attitudes to and preferences for land and landscape. Land Use Policy 2009, 26, S62–S75. [CrossRef] © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). II. VERSION INICIAL Communication Quantifying the impact of urban green space on real estate prices in five mega cities of Latin America Ursula Loret de Mola 1,*, Brenton Ladd 1,2, Sandra Duarte 3, Nils Borchard 4 and Ruy Anaya La Rosa5 1 Facultad de Ciencias Ambientales, Universidad Científica del Sur. Lima 33, Perú 2 School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia. 3 Universidad Nacional de la Patagonia Austral (9400) Rio Gallegos, Santa Cruz, Argentina 4 Ruhr-University Bochum, Institute of Geography, Soil Science/Soil Ecology, Universitätsstrasse, 150, 44801 Bochum, Germany 5 Starfish Initiatives, Armidale, NSW 2350, Australia * Correspondence: eloretdemola.9@hotmail.com; Tel.: +51-998-019-990 Academic Editor: name Received: date; Accepted: date; Published: date Abstract: This study seeks assesses the possibility of using real estate prices to provide an estimate of the monetary value of ecosystem services provided by urban green space across five Latin American mega cities: Bogota, Buenos Aires, Lima, Mexico City and Santiago de Chile. Using Google Earth to map urban green space with images from 2013 and through analysis with a multiple regression model, we evaluated the influence of urban green space, crime rates, business density and population density on real estate prices. The results show a strong positive relationship between the presence of urban green space and real estate prices. Green space explains 52% of the variability in real estate prices across the five studied megacities. Population density, business density and crime had only minor impacts on real estate prices. These results reinforce the idea that urban green spaces are important providers of ecosystem services and that the values of these ecosystem services can be estimated using willingness to pay for real estate. The results further also the need for additional study of ecosystem service provision by urban green space in LAC cities which should include study of how ecological and social factors interact to determine the level of ecosystem service provision from urban green spaces. Keywords: Urban green space; real estate prices; Latin America mega cities; Google Earth 1. Introduction Urbanization is a global phenomenon that is happening rapidly: cities are expanding on average at twice the rate that population is growing [1-3]. In Latin America and the Caribbean, three-quarters of the population already live in cities [4], and cities play a key role in the economies of the region [5]. High population densities and the high concentration of human activity in LAC cities generate negative environmental impacts [6] and cause significant challenges in terms of meeting the demand for new infrastructure, which is often achieved at high social and/or-environmental cost [7]. Many cities consider urban green space and undeveloped areas as potential land to be converted to infrastructure [8,9] without taking into account that the fact that cities depend on the ecosystem services that urban green space provide to sustain human well-being [10,11]. Ecosystem services provided by urban green space are numerous. For example, urban green space may improve air quality, regulate water flow and urban temperature, reduce noise pollution etc. [12-14]. Hence, if the economic value of urban green space could be demonstrated through a premium on real-estate, the importance of the ecosystem services provided by urban green spaces would be reinforced in the political decision-making process [15]. This is important because financing public infrastructure and public services depends heavily on governmental institutional arrangements [16]. One approach to quantify the value of ecosystem services is through the use of hedonic price indices [17]. Hedonic price indices are based on correlations between prices in existing markets (i.e. the real estate market) and specific ecosystem services (i.e. air quality) or bundles of ecosystem services, as for example may be provided by urban green space [13,14]. An open question for Latin America and the Caribbean is whether hedonic price indices can be used to estimate the value of ecosystem service provision from urban green space in LAC cities. In addition to services provided by urban green space, other factors might influence real estate prices, such as opportunities for shopping (or business density), population density, and/or crime frequency. Hence, this study seeks to estimate the influence of urban green space on real estate prices (US$ per m2) at district level in five Latin American mega cities: Bogota, Buenos Aires, Lima, Mexico City and Santiago de Chile. The influence of urban green space on real estate prices is also compared with the influence of crime rates, and business and population density. Thus, this study provides information to urban planners, practitioners and stakeholders about the willingness of communities to invest in preserving urban green space in order to provide a first measure of willingness to pay for urban ecosystem services in LAC cities and compare this to willingness to pay for ease of access to commerce, low crime rates and population density 2. Materials and Methods 2.1. Study area Based on national data compiled by Economic Commission for Latin America and the Caribbean - ECLAC (2013) [18] and real estate information available for each city, five mega cities were selected: Bogotá in Colombia, Buenos Aires in Argentina, Lima in Peru, Mexico City in Mexico and Santiago in Chile. Each city is politically divided into districts or administrative units with 20 districts in Bogota, 48 districts in Buenos Aires, 43 districts in Lima, 16 districts in Mexico City, and 32 districts in Santiago. For this study, districts and administrative units that were predominantly rural or close to water bodies, and/or smaller than 3 km2 and/or larger than 100 km2 were excluded. Size and location of urban green space in remaining districts or administrative units, were collected from 10 randomly selected districts or administrative units. 2.2. Data There is no free access to real estate information in Latin America. Real estate companies and government institutions do not publish their databases. Not all real estate companies generate data for all districts within their respective cities, and there is only a very limited set of institutions that own this information on real estate prices. It therefore required significant research efforts to determine who owned information on real estate prices in the LAC mega cities. In addition, these institutions are generally unwilling to provide free access to their databases since they generate income from them. Hence, convincing these organizations to collaborate with us on this study involved much effort and constant follow-up. The price per square meter of real estate in 2013 at district level (US$ per m2) was used as the dependent variable and provided by the Cámara Peruana de Construcción (CAPECO) for Lima, Portal Inmbolibiario for Santiago de Chile, Cámara Colombiana de la Construcción (CAMACOL) for Bogota, Metros Cúbicos for Mexico City, and Reporte Inmobiliario for Buenos Aires (Appendix A). Independent variables used in this study were urban population density in 2013 (inhabitants per km2), business density in 2013 (number of businesses per km2; any type of business) and security data (number of crimes recorded 2013 per km2; any type of crime) for each district or administrative unit (i.e. comunas in Buenos Aires). These data were provided by the following institutions: Secretaría de Planeación - Alcaldía Mayor de Bogotá (Colombia), Cámara de Comercio de Bogotá – CCB (Colombia), Observatorio de Seguridad en Bogota (Colombia), Gobierno de la Ciudad Autónoma de Buenos Aires (Argentina), Ministerio de Desarrollo Económico (Argentina), Instituto Superior de Seguridad Publica (Argentina), Instituto Nacional de Estadística e Informática – INEI (Peru), Observatorio Nacional de Seguridad Ciudadana – OBNASEC (Peru), Consejo Nacional de Población - CONAPO (Mexico), Instituto Nacional de Estadística y Geografía – INEGI (Mexico), Procuraduría General de Justicia del Distrito Federal - PGJ DF (Mexico), Instituto Nacional de Estadísticas Chile – INE (Chile), Departamento de Estudios Económicos y Tributarios (Chile), and Instituto Nacional de Estadísticas Chile – INE (Chile) (Appendix B). 2.3. Urban green space Google Earth was used to obtain land cover data with images from 2013. Total area sampled per district was 2.5 km2, divided into five randomly selected polygons of 0.5 km2, which accounted for the majority of the area in the districts and administrative units sampled. In summary, this study considered five polygons per district equivalent to 50 polygons per city, with a total number of 250 polygons (Figure 1.). Altitude differs between and within cities, and so to control for this we established as a measuring point at an altitude 500 meters above ground level. In this study, urban green areas are areas covered with any type of vegetation as described by Wolch et al (2014) [14]. Thus, all types of vegetation cover were sampled inside all polygons, from single trees to urban forests using FreeMapTools.com to assess their size in m2. Figure 1. Sampled polygons. From top to bottom: Bogotá, Buenos Aires, Lima, Mexico City, and Santiago de Chile 2.4. Statistical analysis To homogenize data and simplify interpretation, we normalized both the independent (Table 1.) and dependent (price of real estate in US$ per m2) variables prior analysis with the following transformation: 𝑿 𝑵𝑶𝑹 𝒊 = 𝑿𝒊/𝑿𝒎𝒂𝒙 (1) Where Xi is the variable prior to normalization and Xmax is the maximum value obtained for the variable in the respective city. Transformed data were analyzed using multiple linear regression using XLStat (Addin Soft, Paris). Table 1. Variables after transformation (normalization) used for multiple regressions. Av/at = Area Verde / Area total muestreada (green space / total area), Pop.Den = Densidad Poblacional (Population Density), Bus.Den = Densidad Empresarial (Business Density), Crim.Den = Tasa de Delincuencia (Crime Rate) Variable Name Definition av/at NOR Green space per total area Pop.Den NOR Inhabitants per district area Bus.Den NOR Business per district area Crim.Den NOR Crime per district area Normalization: 𝑋 𝑁𝑂𝑅 𝑖 = 𝑋𝑖/𝑋𝑚𝑎𝑥 . Xi: variable prior to normalization. Xmax: maximum value obtained for the variable in the city in which the district occurs. 3. Results From a multiple regression with four independent variables, we obtained a model with admissible accuracy (R2=0.62; Table 2.). The ƞ² values indicate that urban green space was the most important predictive variable explaining 52% of the variability in real estate prices. Business density also had a modest influence on real estate prices with a ƞ² value of 10, whereas crime rate and population density had little impact (Table 2.). However, there was also considerable error with unexplained variability accounting for 39% of the sum of squares, therefore a complete understanding of the factors that determine real estate pricing remains unknown (ƞ²=38.5). Table 1. Summary statistics for the multiple regression model which correlated real estate prices with four variables in review: Green space, Population density, Business density and Crime rate Source dƒ SS MS F Pr>F ƞ² av/at NOR 1 1.14 1.14 60.16 < 0.0001 51.5 Pob.Den NOR 1 0.00 0.00 0.03 0.8690 0.0 Bus.Den NOR 1 0.22 0.22 11.50 0.0015 9.8 Crim.Den NOR 1 0.00 0.00 0.24 0.6283 0.2 Error 45 0.85 38.5 Corrected Total 49 2.21 100.0 SS sum of squares, df degrees of freedom, Pr probability, ƞ2 percentage of variation of the R2 explained for each independent variable. The normalized data used in this study adheres closely to the 1:1 line (Figure 2.) which indicates close correspondence between the measured and predicted price values, and shows the robustness of the multiple regression model. 1.5 Lima Santiago Bogota 1:1 Mexico City Buenos Aires 1.0 0.5 0.0 0.0 0.5 1.0 1.5 Predicted Price Normalized Figure 2. Relationship between measured vs. predicted price values of m2 at district level. Predicted values using multiple regression: Predicted Price NOR = 0.35 + 0.71*av/at NOR - 0.053*Pob.Den NOR + 0.34*Buss.Den NOR - 0.061*Crim.Den NOR. See Table 2 for details. Diagonal line indicates 1:1 relationship. 4. Discussion This study presents a multiple regression model derived from 50 neighborhoods located in five Latin American mega cities. The model showed a strong positive relationship between presence of urban green space and real estate prices, and revealed minor impacts of business density, population density and crime rates on real estate prices. Thus, our results reinforce prior findings showing that urban green space is an important determinant of real estate prices which can provide a robust estimate of the value of the ecosystem services provided by urban green spaces [15]. The unexplained variability in the multiple regression model (38.5%) was also significant and could be related to a variety of factors which were not taken into account in this study. A more sophisticated classification of urban green space, for example, differentiation between different categories of green space could lead to a more accurate model [19]. Candidates for inclusion in a more sophisticated classification of urban green space include “ecological factors” such as “stem diameter structure, leaf area index, basal area, soil quality” and the relative representation of different plant functional traits. Social factors such as changing perceptions that occur with economic status [20] may likewise provide further insights into how green spaces generate ecosystem services in LAC cities. This in turn would provide natural resource managers in LAC cities with the information needed to optimize the configuration urban green space to maximize ecosystem service provision from the limited green areas LAC cities contain. High rates of urbanization in Latin America and the Caribbean continue to pose direct threats to the preservation of urban green space [21] due to limited administrative capacity of LAC cities to plan and control urban growth causing conversion of urban green space into houses and infrastructure. Further, the complexity of creating new urban green space has been increasingly problematic due to a high level of urban land consolidation [22] and overlapping demand of land for multiple purposes (e.g. urban green space vs. infrastructure) [23]. Thus, cities Measured Price Normalized in Latin America and the Caribbean are characterized by a critical deficit of urban space [4], which impairs human well-being [24]. Green spaces are important in urban areas as they support ecological integrity of cities, provide ecosystem services, and improve the livability of cities [13,14,25], which clearly needs to be taken into account in urban planning activities [26,27]. Urban planners require up-to-date accurate spatial data at city scale [28] to understand urban dynamics and processes [29]. In Latin America, robust city data is difficult to obtain due to limited interest in investing in basic research and there is little effort to ensure open access of relevant data. Improving quality and quantity of data acquisition and availability could significantly improve urban planning processes for Latin American and Caribbean cities, which would be directly in line with the United Nations Sustainable Development Goals http://www.un.org/sustainabledevelopment/ sustainable-development-goals/, accessed on 28th April 2017) aiming to improve good health and well-being and supporting development of sustainable cities and communities. Acknowledgments: We want to thank the Cámara Peruana de Construcción, Portal Inmbolibiario, Cámara Colombiana de la Construcción, Metros Cúbicos, and Reporte Inmobiliario for their collaboration by providing essential data that made this study possible. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. References 1. Angel, S.; Parent, J.; Civco, D.L.; Blei, A.; Potere, D. The dimensions of global urban expansion: Estimates and projections for all countries, 2000–2050. Progress in Planning 2011, 75, 53-107. 2. Schneider, A.; Friedl, M.A.; Potere, D. Mapping global urban areas using modis 500-m data: New methods and datasets based on ‘urban ecoregions’. Remote Sensing of Environment 2010, 114, 1733-1746. 3. Seto, K.C.; Fragkias, M.; Güneralp, B.; Reilly, M.K. A meta-analysis of global urban land expansion. PloS one 2011, 6, e23777. 4. Habitat, U. The state of latin american and caribbean cities 2012 towards a new urban transition. Nairobi, Kenia 2012. 5. Cadena, A.; Remes, J.; Manyika, J.; Dobbs, R.; Roxburgh, C.; Elstrodt, H.-P.; Chaia, A.; Restrepo, A. Building globally competitive cities: The key to latin american growth. McKinsey Global Institute 2011. 6. Krausmann, F.; Gingrich, S.; Eisenmenger, N.; Erb, K.-H.; Haberl, H.; Fischer-Kowalski, M. Growth in global materials use, gdp and population during the 20th century. Ecological Economics 2009, 68, 2696- 2705. 7. Cerón-Palma, I.; Sanyé-Mengual, E.; Oliver-Solà, J.; Montero, J.-I.; Ponce-Caballero, C.; Rieradevall, J. Towards a green sustainable strategy for social neighbourhoods in latin america: Case from social housing in merida, yucatan, mexico. Habitat International 2013, 38, 47-56. 8. Bomans, K.; Steenberghen, T.; Dewaelheyns, V.; Leinfelder, H.; Gulinck, H. Underrated transformations in the open space—the case of an urbanized and multifunctional area. Landscape and Urban Planning 2010, 94, 196-205. 9. Mell, I.C.; Henneberry, J.; Hehl-Lange, S.; Keskin, B. Promoting urban greening: Valuing the development of green infrastructure investments in the urban core of manchester, uk. Urban forestry & urban greening 2013, 12, 296-306. 10. Berghöfer, A.; Mader, A.; Patrickson, S.; Calcaterra, E.; Smit, J.; Blignaut, J.; de Wit, M.; van Zyl, H. Teeb manual for cities: Ecosystem services in urban management. TEEB the economics of ecosystems and biodiversity 2011. 11. Chiesura, A. The role of urban parks for the sustainable city. Landscape and urban planning 2004, 68, 129- 138. 12. Bolund, P.; Hunhammar, S. Ecosystem services in urban areas. Ecological economics 1999, 29, 293-301. 13. Sandifer, P.A.; Sutton-Grier, A.E.; Ward, B.P. Exploring connections among nature, biodiversity, ecosystem services, and human health and well-being: Opportunities to enhance health and biodiversity conservation. Ecosystem Services 2015, 12, 1-15. 14. Wolch, J.R.; Byrne, J.; Newell, J.P. Urban green space, public health, and environmental justice: The challenge of making cities ‘just green enough’. Landscape and Urban Planning 2014, 125, 234-244. 15. Nilsson, P. Natural amenities in urban space–a geographically weighted regression approach. Landscape and Urban Planning 2014, 121, 45-54. 16. Assessment, M.E. Current state and trends. Washington, DC 2005. 17. Heal, G. Valuing ecosystem services. Ecosystems 2000, 3, 24-30. 18. LATINA, C.E.P.A.A.; CARIBE, E. Statistical yearbook for latin america and the caribbean 2011. Santiago do Chile: CEPAL 2012. 19. Derkzen, M.L.; Teeffelen, A.J.; Verburg, P.H. Quantifying urban ecosystem services based on high‐ resolution data of urban green space: An assessment for rotterdam, the netherlands. Journal of Applied Ecology 2015, 52, 1020-1032. 20. Scopelliti, M.; Carrus, G.; Adinolfi, C.; Suarez, G.; Colangelo, G.; Lafortezza, R.; Panno, A.; Sanesi, G. Staying in touch with nature and well-being in different income groups: The experience of urban parks in bogotá. Landscape and Urban Planning 2016, 148, 139-148. 21. Inostroza, L.; Baur, R.; Csaplovics, E. Urban sprawl and fragmentation in latin america: A dynamic quantification and characterization of spatial patterns. Journal of environmental management 2013, 115, 87- 97. 22. Haaland, C.; van den Bosch, C.K. Challenges and strategies for urban green-space planning in cities undergoing densification: A review. Urban forestry & urban greening 2015, 14, 760-771. 23. Perrotti, D.E.; Sánchez, R. La brecha de infraestructura en américa latina y el caribe. 2011. 24. García-Ayllón, S. Rapid development as a factor of imbalance in urban growth of cities in latin america: A perspective based on territorial indicators. Habitat International 2016, 58, 127-142. 25. Gómez-Baggethun, E.; Barton, D.N. Classifying and valuing ecosystem services for urban planning. Ecological Economics 2013, 86, 235-245. 26. Romero-Lankao, P.; Gnatz, D.M. Exploring urban transformations in latin america. Current Opinion in Environmental Sustainability 2013, 5, 358-367. 27. Vásquez, A.; Devoto, C.; Giannotti, E.; Velásquez, P. Green infrastructure systems facing fragmented cities in latin america-case of santiago, chile. Procedia Engineering 2016, 161, 1410-1416. 28. Parés-Ramos, I.; Álvarez-Berríos, N.; Aide, T. Mapping urbanization dynamics in major cities of colombia, ecuador, perú, and bolivia using night-time satellite imagery. Land 2013, 2, 37. 29. McPhearson, T.; Pickett, S.T.; Grimm, N.B.; Niemelä, J.; Alberti, M.; Elmqvist, T.; Weber, C.; Haase, D.; Breuste, J.; Qureshi, S. Advancing urban ecology toward a science of cities. BioScience 2016, 66, 198-212. © 2017 by the authors. Submitted for possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/). III. CARTAS DE PRESENTACIÓN Y RESPUESTAS A LOS REVISORES Ursula Loret de Mola Universidad Científica del Sur Km 19, Villa, Panamericana Sur, Lima 33, Perú 511 610 6400 eloretdemola.9@hotmail.com Giovanni Sanesi Guest Editor Special Issue - Forests October, 7th, 2017 Dear Giovanni Sanesi: We are very grateful for the opportunity to resubmit an improved version of our manuscript identified with ID forests-222942 and to have the manuscript considered for the Special Issue "What Role Can Ecological Factors Play in Supporting Ecosystem Services in Urban and Peri-Urban Forests?". We have followed reviewer one’s very helpful recommendation and performed an additional more detailed analysis/ case study for two of the five selected cities: Lima and Buenos Aires; including the following landscape ecology variables: largest Patch size (m2), total number of patches of green space, mean patch size of green spaces(m2), and Average Nearest Neighbor between patches (m). The results from the new case studies of revealed that the relationship between green space and real estate prices is context-dependent. The most important variable in Lima is green space / total space, explaining 83.8% of real estate price variability; in Buenos Aires, the most relevant variables are Total number patches (23%) and Average nearest neighbor, or connectivity between patches of green space (22%). After modifying the manuscript in line with the helpful comments from the reviewers, especially the recommendation to expand the analysis with a more detailed case study we believe the new set of results better align with the central theme of the special issue showing context dependence of ecosystem service provision from urban green space and a potentially productive strategy for future research to better understand how and when spatial patterns of urban green spaces matter for ecosystem service provision Below are point by point explanations of how we have modified the manuscript in response to the reviewer comments. Thank you for your consideration Sincerely, Ursula Undergraduate Thesis Candidate Universidad Científica del Sur, Lima, Perú Comments for the Author: Reviewer #1: My major concern on the paper is related to its focus……In case of a “Communication” – as it was presented here – the reader would expect no “extensive” data analysis but rather a critical literature review (…). Another possibility – that in my point of view would be very valuable – is to extend this manuscript in terms of content and data analysis and to resubmit it as “Research Article”. ANSWER: We thank the reviewer for helpful observation. We followed this advice and performed a more detailed analysis / case study for two of the five selected cities: Lima and Buenos Aires. In the manuscript we are thus extending both analysis and text so that the manuscript now has the form of a regular research article in forests (as opposed to a short communication). To achieve this, we performed a more in-depth GIS analysis of a subset of our data which not only included Green area (m2) divided by total area, but also the following landscape ecology variables: largest Patch size (m2), total number of patches of green space, mean patch size of green spaces (m2), and Average Nearest Neighbor between patches (m). I would strongly recommend to extend the introduction section ANSWER: This has been done in the revised manuscript (line. 46-92) Furthermore, I would like to see a “Case Study Area” section with profound information on the 5 LAC and describing the specific challenges in these 5 LAC cities. Why it is worth to compare them? In terms of results the reader would expect more detailed data analysis, e.g. in terms of the quality of green spaces (parks vs. green lanes), size of green areas (smaller vs. large areas), and differences between LAC in general as well as similar district sizes – all in all analyzing the whole facets of the interrelation between urban green spaces and real estate values in these 5 LAC cities. ANSWER: As detailed in our response to point 1 of reviewer 1 we have now augmented the data analysis and manuscript with the suggested case study. Which as the reviewer suggested it might, provides insight into where and when the amount of green space is important vs. when the ecological attributes of green space is important. We again thank the reviewer for this helpful and insightful recommendation. I’ve some doubts with regard to the other proposed variables (businesses and crime) as these are poor explained and do not contribute to a broader discussion. ANSWER: Crime rate, business and population densities were included in the cross-city analysis because the evaluation committee for my thesis, which is the basis of the work presented here, suggested/required it, arguing that real estate prices could be affected by these variables also. If the thesis reviewers raise this question we think it likely that other readers may do the same. So, we think it is important to leave these variables in the analysis across cities. We do however agree fully with the reviewer that the analysis across cities shows that these other variables are almost irrelevant and that real estate prices can be used as hedonic price indices to quantify the value of urban green spaces and the ecosystem services they provide. And so, following the reviewer’s advice we have dropped crime rates and business and population density in the case study analysis and instead use relative green area and landscape ecology variables (see table 2). Again, as the reviewer suggests this provides deep insight into where and when the amount of green space is important vs. when the ecological attributes of green space is important. Again, we are extremely thankful to the reviewer for their insight and willingness to share it. Against this background and independently of the authors’ decision, the manuscript would require a profound rework, considering and integrating my concerns/comments. As the topic is highly relevant, I would like to encourage the authors to take the time needed for restructuring the manuscript and developing a clear focus as well as data analysis in case of a research article. ANSWER: We addressed second option of reviewer’s request by: 1) extending the introduction including ecosystem services and urban green spaces definitions, their relevance in a global and regional context. Explaining the LAC challenges related to green spaces in sustainability criteria and the importance of having a regional trend for policy formulation and comparison (line. 46-92). 2) Developing two study cases: Lima and Buenos Aires. These cities represent the extremes of climatic characteristics, related to precipitation and temperature, for the selected cities. We measured variables: Total Green area (m2), Largest Patch (m2), Total number of patches, Mean Patch Size (m2), Average Nearest Neighbor (m) (table 2) 3) Describing every variable selected and its relevance (line. 191-208) 4) Including an extension of discussion and conclusions based on this new analysis (line. 268-307). Reviewer #2: This a fascinating short communication that attempts to scribe a value to ecosystem services to real estate prices in five Latin American mega-cities. I have taken the liberty to provide some critical comments for the authors to consider that I hope will improve the readability and impact of this research for the journal Forests. These are: p.1, line 16: ‘this study seeks to assess’ (insert word) ANSWER: Corrected p.1, line 21: past tense applies? ANSWER: Corrected p.1 line 26: ‘The results further also the need for the additional….’ (insert word) ANSWER: Corrected p.1 line 27: LAC acronym not explained ANSWER: Corrected p.1 line 33: urbanization is not exactly a recent phenomenon! What defines a ‘mega-city’? Many urban areas are now conceptualized as internally heterogeneous, functionally differentiated, and highly interconnected ecosystem mosaics. ANSWER: We agree that urban expansion is not exactly a recent phenomenon and that cities are highly heterogeneous, discussion of this is beyond the scope of what we aim for or could achieve in this study and so we use the (simple) UN habitat definition, which we now cite in the case study in a modified way to incorporate the reviewer’s correct observation (line. 96-99) p. 1 line 39: Is it the cities themselves that consider, or rather those who implement governance such as city planners? Here you can expand on wider issues to do with social and environmental interfaces – for example, in the UK we have a concept called ‘natural capital’ which tries to link green infrastructure (GI) and access to greenspace with mental and physical well-being of citizens. The Total Economic Value (TEV) conceptual framework in Europe gives the National Ecosystem Assessment a monetary value. For GI to work it needs to be inter- connected. ANSWER: We corrected the expression for city planners. p. 1 line 43: other factors? e.g.: urban heat island affect, public health benefits (as explained above). ANSWER: We mentioned more factors (line. 69-78). p. 2 line 53: LAC acronym not explained. Are all factors that influence real estate prices considered adequately? Hedonic pricing is often linked to quality of life indicators such as in Europe proximity to suitable schools, local employment opportunities etc. In fact there are a whole bunch of multi-linked influencing social, economic, physical/environmental and Governmental drivers at work. The regression model you propose needs a more implicit justification here so that it is clear that the causal factors you have selected are the most important factors for consideration in the urban areas in question. ANSWER: The acronym is now explained. With respect to the factors that influence real estate prices we have now provided more bibliographic justification in the manuscript. The selection of the variables of population density and crime rates was based on the results from other studies where these variables are considered as determinants of real estate prices (Dunse, Thanos, & Bramley, 2013; Garcia & Riera, 2003; Ihlanfeldt & Mayock, 2010; Kryvobokov, 2007; Pope & Pope, 2012; Wassmer & Baass, 2006). The selection of business density was based on Yu, Cho, & Kim’s (2012) results, describing a positive impact of shopping centers on real estate prices and above all, a considerable annual property tax revenue. Second, as an indicator of commercial dynamism that can justify employment opportunities, and implicit income index. Also, it is worth to mention, that schools of high quality in LAC megacities are generally private and compiling data on their spatial location would require significant work. We agree however that it would have been an excellent additional variable for the analysis across cities (line. 191-208) p. 2 line 57: why select these five? There is an assumption here that they are comparable, but they all exist in different countries and therefore are influenced by very different economic, political and fiscal measures. ANSWER: The selection of these five megacities was based on availability of real estate data and social-demographic and economic data in 2013. Also as the majority of the research team is from Lima, a city built in a hyper arid location, we were interested to know how the relation between urban green space and willingness to pay for it changes across megacities where rainfall varies. With the new case study added to the manuscript following the recommendation of reviewer 1 the rationale behind city selection we think is now clearer (line. 100- 103). p.2 line 61: does your paper provide a ‘first measure of willingness to pay’ as you suggest? There is a growing amount of literature available. Is this also what you will really show? Residents may be blissfully unaware of this in certain contexts. ANSWER: agreed the possibility for spurious correlations exists however the consistency of the pattern across megacities does provide support for the conclusion that the correlations presented are not spurious. To address the reviewer’s correct observation, we have modified text to acknowledge the possibility of spurious correlation (line 275-289). p.2 line 74: streamlining of data mining needs justification. Will this pick up heterogeneous districts – examination of Fig 1 seems to indicate these samples are not random. ANSWER: After the first filter (rural and close to water districts), the remaining districts per city were written in a different sheet of an Excel file. The number of ten (10) districts was determined in order to be able to establish comparisons because of the differences in the numbers of districts per city. We opted for the random function in Excel, which gives you numbers with long decimals between 0 and 1 and typing it in the next column. When selecting both column and ordering from minor to major, we selected the first ten (10) districts per city and with them perform the measurements and analysis required. p.2 line 76: label for section 2.2 too brief! What kind of data?! ANSWER: Corrected p.2 lines 77-84: Do you need this section? Elsewhere in the world this information is freely available. ANSWER: Probably this type of information is freely available in most of the countries around the world. However, the reality in Latin America is quite different. Peru does not publish house price statistics, and generally its statistics are weak. The Ministerio de Economía y Finanzas has basic statistics on production, prices, interest rates, international reserves and the balance of payments. There are financial statistics at the Banco Central de Reserva del Perú but just for some districts. The Instituto Nacional de Estadística e Informática produces data, but not on-line time-series. The Peruvian Chamber of Construction (CAPECO) produces data, but it is not publicly or freely available. No housing price statistics are published in Argentina; Reporte Inmobiliario, a private firm, produces house price statistics but they are not publicly available. General statistics come from the Banco Central de la República Argentina, the Secretary of Economic Policy, the Instituto Nacional de Estadística y Censos and the Dirección Nacional de Cuentas Internacionales. No housing price or rent statistics are published in Mexico. The Instituto Nacional de Estadística y Geografía is weak in this aspect, and appears to concentrate on census data. Most Colombian house price statistics are from Colombia's central bank, Banco de la Republica Colombia. The time-series are quarterly, from 1988 and there are no rentals statistics. Chile publishes house price statistics through the Chilean Chamber of Construction (CChC) but the information is for period and not all is publicly available. Reporte Inmobiliario also publishes reports on Chilean property market from time to time. So, private firms have a niche market that they can exploit for their own business and for the public, in that sense, information access is restricted. However, we will follow your suggestion and we will place this paragraph in the acknowledgements section p. 2 lines 92-101: this is better placed in the acknowledgements section. ANSWER: Corrected. p.3 line 103: citation for Google Earth should include year of issue. Citing images for use in publications is worth checking – a convention might be software and version, date of original imagery, location name, geographic coordinates, altitude, data set (if applicable), image source(s) and copyright(s), website URL and date accessed – check this. Interestingly some land use data available on GIS for featured countries from: https://freegisdata.rtwilson.com/. ANSWER: Thanks for the GIS information, we appreciate it. About Google Earth images, we made the details (scale, date) more visible and wrote a more specific description of the figure. p. 3 line 104: why 2.5 km2? Can this be justified with reference to the work of others who used a similar approach? Does the 250 polygons mean that data is upscaled from 5 or did you measure data within all of these? ANSWER: Within the five megacities, Buenos Aires has the smallest district size. After random selection, the smallest district size was 3.5 Km2, because of that we decided to establish 2.5 km2 as total measured area. The measure of green space was done within all 250 polygons. p. 3 line 110: how was vegetation cover quantified? Link to urban forests here. ANSWER: From Google Earth images, all vegetation from single trees to urban forests were mapped by using the tool polygon in Google Earth program. p. 4 Fig 1: clustering of points seems non-random. Are these scaled (to size) polygons? Please note previous comments above. No scale bar on the maps – label each image (a), (b), (c) etc and link to the relevant city. There will be spatial heterogeneity across all the cities (e.g.: some areas would have higher crime rates than others), so the relative contribution of these factors will change. ANSWER: The Figure 1 shows the central point of each sampled polygons in order to visualize the distribution of all sampled polygons within the selected districts of each city. It could be perceived as non-random distribution; however, there must be considered the two filters used (ruralness and districts located close to beaches; and districts smaller than 3 km2 and bigger than 100 km2) (Figure. 2). p.4 line 116: OK, interesting. Data transformation of the independent variable (X) will not affect the distribution of Y, so can be made with impunity providing the transformation has the desired affect (ie: normalises data to meet assumptions for homoscedasticity). Transforming the value of Y (in your case, price of real estate) will affect least squares considerations. Log Y data transformations are commonplace. Can you clarify that the transformation method you ascribe to the Y variable is valid? Here again I would ask you to justify why these four independent variables were selected when others could have been used? A model is only as useful as what is fed into it. Would a multiple stepwise regression not work? ANSWER: The normalization was made for all the variables (Y and X), due to the variability of price ranges from one city to another. Data relativization ensures that this source of variability is taken into account outside the statistical analysis, ensuring that our ability to evaluate the variable of interest across cities is maximized. The practice of normalization (making data relative) is applied across a wide range of disciplines including studies focused on urban issues, e.g. Bettencourt, L., & West, G. (2010). A unified theory of urban living. Nature, 467(7318), 912-913. p.5 line 129: we? Personal term; it is probably my lack of familiarity with XLStat, but I am unfamiliar with the symbol denoted as a column header in the final column of Table 1. What is this? (also mentioned on line 132). ANSWER: The ƞ² is the percentage of variation of the R2 explained for each independent variable. ƞ² is obtained from the SS partial value between the SS total value per 100. The table legend has been expanded to include this information (table 2) p.5 line 135: comments here might be linked to data transformation approach of Y variable, the way data was subdivided and sampled or that fact that not enough causal factors were included in the model. Remember – here I am playing devil’s advocate. There seems to be an insistence that higher prices are linked to more greenspace, but this is kind of predictable. Prices will always generally be higher in residential areas that are less developed and more affluent. Value of Table 1 could be questioned. ANSWER: Agreed. The new analysis/case study we did follow the suggestion of reviewer 1 which we believe increases the value of the analysis we have performed. We also would like to reiterate the fact data availability is extremely limited and complicated in our region and the compilation of the data we provide in the appendix we believe is valuable. p.5 line 141 and Figure 2 p. 5: could relationships between measured and predicted price values be quantified e.g.: spearman rank? Could the authors explain the validity of plotting actual values versus predicted values, when R2 value is already calculated? Is there something about undertaking some intra-city comparisons that would explain why there is (perhaps) regional variation here? I know the 10 were randomly selected but this might make your results go further. Could you not also explore trends within individual capital cities, as this might show the spatial heterogeneity you might find in any cityscape. ANSWER: We agree. The reviewers comment aligns closely with comment 1 of reviewer 1. We now do intra-city comparisons (Lima and Buenos Aires) and as the reviewer suggests it adds significantly to the manuscript. Thank you for the helpful suggestion and see our response to comment 1 of reviewer 1 p.6 line 144: use of vs.? ANSWER: Corrected p.6 line 149: first use of the word neighborhood? According to your regression model inhabitants and crime had no effect on prices contradicting your statement on lines 151-152. ANSWER: Corrected p.6 line 152: you say findings but feature only one reference. Have you really determined the value of ecosystem services? What you seem to have done is tried to explain real estate prices using an (incomplete) multiple regression model. ANSWER: We included more references. We assumed that where people prefer to live is reflected in the monetary and non-monetary value that they attribute to their district. Focusing on the value of green space, we can assume that urban green spaces provide bundles of Ecosystem Services (line. 269-280). p.6 line 157: why can’t you do this? If you do, this article becomes far more publishable. The network via connecting nodes is critical for GI – this is worth adding. ANSWER: We can and have done this as per this recommendation and the first suggestion of reviewer 1. Thanks for the hint. p.6 line 161: surely public health agendas are worth a mention? In the UK it is proven that having access to greenspace can reduce the bill for the National Health Service. Doctors are even prescribing green prescriptions for combating mental health problems. ANSWER: We included health aspect in discussion (line. 298-299). p.6 line 162: by assuming homogeneity between and within cities, you could have provided an answer for this. Most of what features in the next paragraph is largely conjecture and not borne from the work you have undertaken. ANSWER: We are not assuming homogeneity between cities, what we assumed based on the document published by ECLAC named “REGIONAL PANORAMA Latin America Megacities and Sustainability” is that LAC megacities present common problems and challenges related to sustainability, even when the intensity of specific aspects changes from one city to another (line. 302-303) One would assume that to be published in a journal called Forests, there would need to be more mention of urban forestry and trees – there is little mention of this in the actual paper. ANSWER: We extended and focused more on this topic in the case studies of Lima and Buenos Aires. Reviewer #3: I feel that the context of the study with existing theoretical, practical and philosophical work in the area of ecosystem services valuation is a little obscure and could benefit from more explicit description. Why is it a useful addition and what are the limitations of the approach? ANSWER: We agreed. We mention in case study results, that it might be useful to consider more research to better understand when and where ecological attributes of green space affect prices so we can confidently use hedonic price indices for holistic valuations of green space (line. 300-307) It would be useful to have more of a thread throughout the manuscript re: the importance of urban green space / global context, and the kinds of decision- makers that it / this work would be most relevant to. This is covered in the discussion but isn’t clearly set out in the introduction. ANSWER: Thank you. We extended the introduction section in order to add more useful information (line. 46-92). It might be good to break the discussion down more clearly in to (for example) technical considerations, considerations for different decision-makers, broader relevance of the work and future research. At present each statement seems rather isolated and lines 181-184 in particular are important but comes across a little tokenistic when included only as a final thought. ANSWER: Thank you. We followed your recommendation to improve the flow and readability of the discussion (line. 269-307). 18-19: mapping vs / plus quantification or qualification? ANSWER: Corrected. 20-21: why was this particular set of other variables chosen? This isn’t really made completely clear in the body of the manuscript later either ANSWER: This comment aligns closely with comment p2-line 53 of Reviewer 2, please see our response to comment p2-line 53 of reviewer 2. 22: which value are you showing here, with what accuracy etc.? ANSWER: The value was explained and the accuracy was added. 26: Yes - it is one element in the full picture of the way in which environment and society are enriched by urban green spaces. It would be worth briefly setting out the philosophical context of this approach - not here, but early in the manuscript ANSWER: Thank you, we agreed. Now, we have presented a broader picture of the benefits that society receives with the incorporation of urban green spaces in the introduction and mentioned this in the discussion (line. 69-78) 27: ‘LAC cities’ - define this acronym at least once in the text and try to be consistent throughout with use of the full text / acronym ANSWER: The acronym is now explained. Keywords: it would be useful to include some keywords that refer to the method and approach used, e.g. hedonic pricing ANSWER: Hedonic price indices was added as a keyword. 34: ‘Latin America and the Caribbean (LAC)’ ANSWER: Corrected 36: As above ANSWER: Corrected 38: What kinds of infrastructure are you referring to here? It is a pretty broad term referring to a range of physical and non-physical elements. I presume you mean things like civil engineering interventions: water, energy and transport infrastructure and housing? ANSWER: We referred to physical civil infrastructure, because Latin American and the Caribbean has a considerable deficit of infrastructure. Closing the region's infrastructure deficit requires US $ 250 billion annually over the next five years, equivalent to 5% of GDP(BID,2016). This has now been defined clearly in the manuscript (line 108-154). 43-46: Yes… but how and why? While it is beyond the scope of the manuscript, it might be useful to consider (or hint at the consideration of) the implications and limitations that hedonic pricing methods have for indicating a holistic value of green space. What kind of political decisions do you have in mind here? Local planning decisions or more general, high level decisions around environmental policies? ANSWER: We extended and expanded this point in the discussion (line. 300- 307) 47: What about private infrastructure developments? How significant are these in comparison to the public sector in these spaces? ANSWER: Private infrastructure developments related to urban green space are scarce, and they usually take place in private-public alliances. The best examples of this are in Belo Horizonte, Brasil and more often in Chile. However, they are not widespread in all the region because of several factors (politics, governance, capacity, profitability, etc) 51-53: I am sure that this is true, but in which ways are you considering this an open question: Because it hasn't been done before in LAC? Or at all? Because other studies in other cities are non-transferrable / there are particular differences that you expect to be the case in LAC? ANSWER: We considered it as an open question related to a regional context, because policy design for Latin American megacities requires an identification of common urban trends in the region. This we have emphasized now in the revised manuscript (line 108-154) 54-55: Why were these particular variables chosen? It would be good to briefly outline why. ANSWER: We have now provided more bibliographic justification in the manuscript as we mentioned above (see response to comment p2 line 53 of reviewer 2). 57: Formatting: m2 ANSWER: Corrected 59: Variable selection - as above ANSWER: Explained (line. 191-208) 61: Invest in having green space… does the word ‘preserving’ give too much future weighting to the method? ANSWER: Agreed, we corrected. 63: Variable selection - as above ANSWER: Explained 67: In what way was the real estate data used to select the cities? This is unclear. Is this just a question of data availability, or was there a considered ‘selection process’? ANSWER: The reviewers comment aligns closely with comment p2: lines 77-84 of reviewer 2. As we explained, real estate data in Latin American region is very difficult to obtain. From the eight megacities, we were able to obtain information for only five. We consider that one of the greatest achievements of the research presented here is to make this information available. 72: How did you judge level of ‘rural’-ness, or proximity to water bodies? Distance, scales etc. ANSWER: The city administration defines rural zones within the city according to data obtained in census. However, due to pragmatic limitations, we considered “rural and close to sea shore districts” districts in which present 50% of the district was rural or adjacent to a beach. 74-75: How was the random selection undertaken? ANSWER: After the first filter (rural and close to water districts), the remaining districts per city were written in a different sheet of an Excel file. The number of ten (10) districts was determined in order to be able to establish comparisons because of the differences in the numbers of districts per city. We opted for the random function in Excel, which gives you numbers with long decimals between 0 and 1 and typing it in the next column. When selecting both column and ordering from minor to major, we selected the first ten (10) districts per city and with them perform the analysis. 78: Presumably because they only have jurisdiction over a certain number of areas is which they are marketing property? I do not know this system well, and presume that other readers may benefit from a brief explanation of how the real estate system operates in these cities ANSWER: We provided a brief explanation in acknowledgment section. 83-84: Actually, really useful to know! ANSWER: Thank you. We are sure this will help others to identify partnerships. 104: Why were these quantities chosen? ANSWER: Within the five megacities, Buenos Aires has the smallest district size. After random selection, the smallest district size was 3.5 Km2, because of that we decided to establish 2.5 km2 as total measured area because it fit within this area. 108: 500m ‘above ground level’? Unclear on the meaning of this reference point - this is referring to where the remote images were taken? So all of your randomly selected points were also constrained by a requirement to be (at?) 500m within your methodology? ANSWER: The 500 m above ground level (m.a.g.l.) is the altitude where the measure took place, due to the differences between cities’ altitudes. For example, Lima’s average altitude is 154 m.a.s.l. and Mexico city’s altitude is 2.250 m.a.s.l. In that sense, the urban green space measure was done at 654 m for Lima and 2750 m for Mexico City to ensure standardization of measurments. 111: Is this a validated methodology? Has it been used in other notable studies? ANSWER: The use of Google Earth is valid (Mering, Baro, & Upegui, 2010; Clark & Aide, 201; Dorais & Cardille, 2011; Duhl, Guenther, & Helmig, 2012; Wardaya & Ridha, 2014). 130: Refers to 'Table 2'? ANSWER: Corrected 133-135: So this just refers to all of the possible factors that were not fed in to the model? - including, presumably, a lot of factors that are irrespective of the environmental context ANSWER: Yes, in the case studies now included deeper insights are provided. 152-154: Further references would be useful here. As noted previously: " it might be useful to consider (or hint at the consideration of) the implications and limitations that hedonic pricing methods have for indicating a holistic value of green space.” It gives valuable insight through a particular lens: the perspective of standard, contemporary economic approaches for valuing commodities ANSWER: Agreed. We mentioned in discussion section in order to point out the limitations of the methods (line. 261-267) 156: As comment for 133-135 ANSWER: We extended the analysis in order to provide better insights. 158-160: This is a useful list, but seems rather specific when some general aspects such as ‘biodiversity’ on the environmental side, and ‘amenity value, or specifics like value for recreation' are not included ANSWER: Agreed. We mentioned some additional factors like biodiversity and culture/recreational values. 162-165: A valid point. Are these the kind of decision-makers that you have in mind for this work in general? Just to be clear - is this a statement about natural resource managers having a better understanding of the most important factors (and potentially ‘tradeoffs’) between different kinds of ecosystem services depending on the way in which they are valued? ANSWER: Yes. We measured urban green spaces as a bundle of ecosystem services from which society benefits, although the relevance of an ecosystem services in a given city depends on the environmental and socio-economic characteristics of each city. 170: WRT ‘urban land consolidation’ - presumably this is the case in existing urban ‘boundaries’, but not necessarily in the peripheral areas? ANSWER: we agree the imbalances and asymmetries of development in the Latin American and the Caribbean region are expressed territorially in the human settlements that exist in it, where the increase and concentration of wealth in contrast to the increase and diffusion of poverty are the main examples that illustrate the urban land consolidation phenomenon in the region, this complex issue is at least now mentioned in the revised manuscript (line 108-154) 174-176: Another group of decision-makers. This statement re: the importance of green spaces feels like it could have been made earlier in the introduction for greater impact ANSWER: Thank you. We follow your recommendation and extended the idea in the introduction (line. 4-108) 177-178: Your message is a bit mixed here, between urban green space priorities and calls for better data and open access. (How, where would this data be generated and stored? Any examples?) ANSWER: Agreed. We divided the two ideas into separate parts and explained in a more detailed way (line. 269-307) 190: Formatting: m2 ANSWER: Corrected Universidad Científica del Sur Km 19, Villa, Panamericana Sur, Lima 33, Perú 511 610 6400 eloretdemola.9@hotmail.com Giovanni Sanesi Guest Editor Special Issue - Forests November, 9th, 2017 Dear Giovanni Sanesi: We are very grateful for the opportunity to resubmit a revised and improved version of our manuscript identified with ID forests-235954 and to have the manuscript considered for the Special Issue "What Role Can Ecological Factors Play in Supporting Ecosystem Services in Urban and Peri-Urban Forests?". Also, we would like to express our thanks to the reviewers for the positive feedback and helpful comments for improving the manuscript. We have modified the manuscript in line with the helpful comments from the reviewers and addressed main concerns about manuscript structure to guide the readers to better understand the context dependence of ecosystem service provision from urban green space and a potentially productive strategy for future research to better comprehend how and when spatial patterns of LAC urban green spaces matter for ecosystem service provision. We believe the manuscript have resulted in an improved revised study benefited from these insightful revision suggestions. Below are point by point explanations of how we have modified the manuscript in response to the reviewer comments. Thank you for your consideration Sincerely, Ursula Undergraduate Thesis Candidate Universidad Científica del Sur, Lima, Perú Comments for the Author: Reviewer #1: Most elements have been addressed from the original critical review, and the revisions have improved the academic value of this paper. ANSWER: We are very grateful with the reviewer’s comment. Reviewer #2: Thank you very much for the improved manuscript and I’m happy to read an extended manuscript. Most of the reviewer’s comments were integrated into the resubmitted version that lead to a better understanding of the topic. However, I have some concerns /comments on the paper that hopefully will be useful to the author/s in improving this manuscript: Introduction section: A general overview of the topic is presented; a definition of urban green space is given and the research topic and challenges are generally addressed. But what the reader is missing is the central research question / hypothesis of the paper. The last paragraph provides some insides. However, no research question is provided that guides through the manuscript. ANSWER: We thank the reviewer for helpful recommendations in order to improve the academic quality of our manuscript. Our hypothesis referred to where people prefer to live is reflected in the monetary and non-monetary value that they attribute to their district. On this basis, if there is a relationship between green space and real estate prices, it will lead to ensure the provision of ecosystem services through specific economic tools. Focusing on the value of green space, we can assume that urban green spaces provide bundles of ecosystem services. Our both research questions described in the last paragraph of the introduction section pretend to guide the readers to the first step of a permanent urban research agenda. Additionally, some minor remarks on this section: Line 46-49: Those sentences refer to Latin America or in general to the world? This should be clarified. ANSWER: We clarified the information by specifying the word “world” where needs to be. Line 49: You refer to “dense settlements (cities) by 2000.” This source is quite old considering that urbanization is a very fast process. It would be great if a more current source would be cited. ANSWER: It was corrected, and an updated reference was cited. Line 60: What do you mean with “undeveloped areas”? Please provide a more precise definition or wording. ANSWER: We follow reviewer recommendation and reword to Vacant Lots to be more precise Line 76: I would kindly recommend to consider the following publication that refers especially to the potential of urban green spaces and social interaction in Santiago de Chile: Krellenberg, K., Welz, J., Reyes-Päcke, S. 2014: Urban green areas and their potential for social interaction – A case study of a socio- economically mixed neighbourhood in Santiago de Chile. Habitat International, Volume 44, Pages 11-21, ISSN 0197-3975, https://doi.org/10.1016/j.habitatint.2014.04.004. ANSWER: We thank the recommendation and cite the study Krellenberg, et al (2014). Line 79-81: These two sentences do not contribute to the overall discussion on challenges for urban planners in LAC cities as these are to general/obvious. What is the real challenge behind? The reader would expect a more profound discussion on the challenges that LAC cities faces today. I would kindly recommend being more precise and try to formulate interesting hypothesis. ANSWER: We follow reviewer’s recommendation and precise the urban challenges in LAC context, as follow: “One of the main challenges that LAC urban planners face in order to achieve welfare for millions of urban residents in LAC megacities is related to large, /uncontrolled and informal urban development that places pressure on the provision of basic services, increases a city’s vulnerability and has a number of negative environmental consequences. This informal growth also tends to occur independently and apart from formal urban expansion which leads to the consolidation of spatial segregation socioeconomically” Line 88-92: Are these your research question? What is the main aim of the paper? How is it structured … this would guide the reader through the manuscript. ANSWER: The aim of the study is to provide a first insight into the estimation of the value of ecosystem service provision from urban green space across LAC megacities using hedonic price indices; and the identification of how the ecological attributes of urban green space might impact the capacity of urban green spaces to provide ecosystem services. As well, we mention that it might be useful to consider more research to better understand when and where ecological attributes of green space affect prices so we can confidently use hedonic price indices for holistic valuations of green space because it is beyond our scope. We agreed about manuscript structure and make subdivision through all document to provide a better understanding. Methods: Although the authors indicated that a specific focus on Lima and Buenos Aires is given the manuscript still focuses on a comparative analysis. Therefore, it is absolutely necessary to present a profound description why these specific 5 LAC cities were selected. Why 5 and not 7? Which kind of criteria sustains your selection? Thus, what are the central aspects/topics of comparison in order to highlight similarities and differences of the research studies? Line 108 and further on some broad information is given on the 5 LAC cities, however it is still not clear why these 5 cities were selected for a statistical analysis. Are these cities very similar to each other or are they very different? Why is it worth to compare them? ANSWER: The main reason about the selection of five cities instead of eight was the possibility to obtain real estate prices at district level. The three megacities excluded are located in Brazil, and the access to real estate prices there is more complex. Additionally, I would recommend removing Figure 1 as this does not provide any important information. Instead of Figure 1, a table should be presented that gives the most important information on the 5 cities and that better explains the criteria for comparison. Furthermore, be careful in presenting the 5 cities. At least a consistent structure should be pursued. For example, annual population growth is only presented for Bogotá and Buenos Aires; and only in Bogotá a percentage of housing deficit is presented. Thus, be precise in presenting the cities, use similar criteria (maybe those of section 2.3?) and those that are very important for your analysis and research question. Furthermore, please extend your bibliographic references for this section. Many of your statements are not supported by literature. ANSWER: We addressed reviewer’s recommendation by removing Figure 1 and presenting a Table 1 with the five cities information in the same structure and same indicators for better comparison. Also, we extended the references for this section. (lines. 111-147). In section 2.2 (Land Cover Data) you refer only to Lima and Buenos Aires. Why Lima and Buenos Aires were selected? For the reader it is very important to know why and it is quite important to explain your decision. Why do you need the other 3 cities? ANSWER: We agree that the justification is not exactly precise. The reason of city selection is now explained (lines.192-195), and literally says: “Buenos Aires and Lima where chosen for the case study because within the group of five cities considered in this analysis these cities represent the extremes of climatic conditions captured in the analysis. Mean Annual Precipitation (MAP) for example in Lima is 16.4mm, whereas in Buenos Aires the figure is 1040mm” Discussion and conclusion: This section aims to respond the research question, thus it should stretch the bow to the introduction. Unfortunately this is not explicitly done here. The discussion is very simple, e.g. line 282 states that “classifications of urban green spaces can lead to a deeper understanding” --- What do you mean with deeper understanding? How far your analysis can contribute to this kind of deeper understanding? Furthermore, in line 284 and further on you give an explanation for the different results in Lima and Buenos Aires and highlight the climatic conditions. This kind of specific criteria was not discussed in the case study area section and this seems very surprising here. ANSWER: We improved the discussion section by providing further information of Buenos Aires and Lima context to better understand the impact of green spaces attributes and the value people’s preference about these spaces on real estate prices. In addition, we introduced climatic conditions and differences in the specification of case study section. Line 296-298 is a kind of definition that could be moved to the introduction section. It should not be part of the discussion section, unless it is to be questioned here. ANSWER: The sentences were moved to the introduction section. (lines. 84-86) Line 303 seems very obvious – “relationship between green spaces and real estate prices is context-specific” – It would be interesting to know what kind of context make the difference? ANSWER: We have now augmented the information of Buenos Aires and Lima, in order to provide a better contextualization of the relationship between green space provision/attributes and real estate prices. (lines. 279-287) Finally, within the discussion I would acclaim a differentiation (e.g. in terms of sub-sections) of a theoretically part (what does the results mean for theoretical framework), methodological reflection (e.g. data availability, further statistical analysis) and relevance of the results for further urban planning strategies. ANSWER: This has been done in the revised manuscript in all section to keep a coherent way for better understanding. We again thank the reviewer for this helpful recommendation. Ursula Loret de Mola Universidad Científica del Sur Km 19, Villa, Panamericana Sur, Lima 33, Perú 511 610 6400 eloretdemola.9@hotmail.com Giovanni Sanesi Guest Editor Special Issue - Forests November, 18th, 2017 Dear Giovanni Sanesi: We are extremely grateful for your consideration accepting our manuscript identified with ID forests-235954 for the Special Issue "What Role Can Ecological Factors Play in Supporting Ecosystem Services in Urban and Peri-Urban Forests?". We have addressed all your comments and constructive feedbacks. Thank you for your consideration Sincerely, Ursula Undergraduate Thesis Candidate Universidad Científica del Sur, Lima, Perú Comments for the Author: I ask the authors to select the right term between 'Latin American and Caribbean megacities' and 'Latin American megacities' ANSWER: We selected Latin American to stay coherent in terms along the study. I also ask whether the term 'green spaces' refers only to public spaces or the total availability of green spaces. In my opinion this aspect is unclear and the authors have to explain how they were able to select only public spaces, if any. ANSWER: Thank you for the hint. We have added an explicit statement at the end of section 2.2 Green space quantification across the 5 megacities precising that we referred to the availability of green spaces including private and public green spaces. (Line 168) I would like to add further elements to the debate on ecosystem services that green areas can provide to citizens in this geographic context. ANSWER: We have added a paragraph in discussion section pointing out the relevance of ecosystem services depending to the specific characteristics of a city. (Line 290) The bibliography also contains several typing errors. ANSWER: We corrected typing errors in bibliography following Forests reference format and style. IV. VERSION EN ESPAÑOL Artículo Sobre el Uso de Índices de Precios Hedónicos para Comprender la Provisión de Servicios Ecosistémicos de las Áreas Verdes Urbanas en Cinco Megaciudades de América Latina 1. Introducción Entre los años 1700 y 2000, el 55% de la cubierta terrestre libre de hielo fue transformada por las actividades humanas dejando menos del 45% de la biosfera terrestre, natural o seminatural [1]. Asimismo, los seres humanos hemos cambiado no solo la forma de utilizar nuestro entorno sino también nuestra distribución dentro de él. La población mundial ha pasado de vivir principalmente en tierras semi naturales en 1700 [2], a vivir substancialmente en densos asentamientos, o ciudades, en el 2016 [3]. En la actualidad, las ciudades albergan más de la mitad de la población mundial [4], y se expanden en promedio el doble de la tasa de crecimiento poblacional [5–7]. En América Latina (AL), tres cuartas partes de la población ya vive en ciudades [8], convirtiéndola en una de las regiones más urbanizadas en el mundo. Mas aun, las ciudades y megaciudades de AL desempeñan un papel muy importante en la economía de la región [9]. No obstante, las ciudades de AL también también se caracterizan por una profunda segregación socio-espacial, altos niveles de delincuencia, desigualdad de ingresos y pobreza [10]. La alta densidad poblacional y la alta concentración de actividades humanas en las megaciudades de AL han conllevado un número de impactos ambientales negativos [11], así como varios retos en términos de satisfacer la demanda por nueva infraestructura, la cual se logra generalmente a un alto costo socio-ambiental [12]. Muchos planificadores urbanos y políticos consideran las áreas verdes urbanas y los lotes disponibles como zonas potenciales para ser convertidas en infraestructura [13,14], sin tomar en cuenta el hecho que las ciudades dependen de los servicios ecosistémicos que las áreas verdes urbanas proveen para sostener el bienestar humano [15,16]. Las áreas verdes urbanas, definidas como espacios exteriores con vegetación natural o espacios humanamente modificados [17] incluyen parques y bosques urbanos, vías verdes, senderos, jardines comunitarios, árboles en la calle, cementerios y otros [18], aparecen como parches inmersos en la matriz urbana, donde su conectividad y continuidad esta muchas veces en peligro por otras prioridades de adjudicación de tierras [19]. Por estas razones, las ciudades de América Latina se caracterizan por un crítico déficit de áreas verdes urbanas [8], las cuales perjudican la calidad de vida de sus habitantes [20]. Las áreas verdes urbanas proveen beneficios ambientales y sociales; al mismo tiempo, ayudan a aminorar muchos problemas urbanos al proveer numerosos servicios ecosistémicos. Los servicios ecosistémicos son las contribuciones directas e indirectas al bienestar humano, en este caso, de ecosistemas urbanos y sus componentes [15], y su provisión está relacionada con un incremento en la calidad de vida [21] y en la resiliencia urbana [22]. Las áreas verdes urbanas pueden mejorar la calidad del aire a través de la filtración de contaminantes, regular el flujo de agua y la temperatura urbana, reducir el efecto de isla de calor generado por el concreto y la combustión de motores como también reducir la contaminación acústica [23,24]. Las áreas verdes urbanas también mejoran la salud mental y física de los ciudadanos, y facilitan la interacción social de los residentes urbanos [18,25–27]. Sin embargo, estos servicios ecosistémicos no poseen un precio de mercado; por lo tanto, no pueden ser comercializados en un mercado existente [28], provocando una inadecuada consideración de las áreas verdes en las políticas públicas de planificación urbana. Uno de los principales desafíos que los planificadores urbanos enfrentan para alcanzar el bienestar de millones de residentes en las megaciudades de AL, está relacionado con el masivo crecimiento urbano, descontrolado e informal; el cual ejerce presión en la provisión de servicios básicos, aumentando la vulnerabilidad de las ciudades y otras consecuencias ambientales negativas [29–31]. Este crecimiento informal tiende a ocurrir independientemente y por separado de la expansión urbana formal, conduciendo a la consolidación de la segregación espacial socioeconómica [32]. Las áreas verdes son importantes en las zonas urbanas debido a que contribuyen a su integridad ecológica, proveen servicios ecosistémicos, y mejoran la habitabilidad en las ciudades [18,24,25]; lo cual requiere ser considerado en las actividades de planificación urbana [33,34]. Si el valor económico de las áreas verdes pudiese demostrarse a través de una prima en los precios inmobiliarios, la importancia de los servicios ecosistémicos provistos por las áreas verdes se vería reforzada en el proceso de la toma de decisiones políticas [35]. Ello es realmente importante porque la financiación de la infraestructura y los servicios públicos depende en gran medida de los arreglos institucionales gubernamentales [36]. Un enfoque para cuantificar el valor de los servicios ecosistémicos es mediante el uso de índices de precios hedónicos [37]. Los índices de precios hedónicos se basan en correlaciones entre los precios de mercados existentes (ejem., el mercado inmobiliario) y servicios ecosistémicos específicos (ejem., la calidad del aire) o paquetes de servicios ecosistémicos, como por ejemplo las áreas verdes urbanas [18,25]. Una pregunta abierta para la región de AL es si los índices de precios hedónicos pueden ser utilizados para estimar el valor de la provisión de servicios ecosistémicos a parir de las áreas verdes en estas megaciudades. Otra pregunta abierta para las ciudades de AL sería cómo los atributos ecológicos de las áreas verde urbanas pueden impactar la capacidad de estas áreas en la provisión de servicios ecosistémicos. En esta investigación, se proporciona algunos alcances en los dos temas mencionados para las ciudades de AL. Específicamente se prueban las siguientes hipótesis: (1) los índices de precios hedónicos pueden proveer estimaciones confiables del valor de los paquetes de servicios ecosistémicos generados por las áreas verdes urbanas a través de las megaciudades de AL, y (2) la información obtenida de los índices de precios hedónicos puede ser específica al contexto y variar a través de las ciudades. 2. Métodos 2.1. Características de las ciudades empleadas en la comparación de índices de precios hedónicos a través de las ciudades Esta investigación está enfocada en las megaciudades de AL (opuesto a las ciudades, utilizadas generalmente) debido a la alta concentración de población latinoamericana en megaciudades en comparación con otras regiones del mundo [38]. De acuerdo con UN Habitat (2012), existen ocho megaciudades en AL: Buenos Aires, Ciudad de México, Rio de Janeiro and São Paulo (con más de 10 millones de habitantes) y Belo Horizonte, Bogotá, Lima y Santiago (con poblaciones aproximadas a los 10 millones) [8]. De estas ocho ciudades, fue posible obtener los precios inmobiliarios a nivel de distritos para las siguientes cinco megaciudades: Bogotá en Colombia, Buenos Aires en Argentina, Lima en Perú, Ciudad de México en México y Santiago en Chile (Anexo A). A pesar que estas cinco megaciudades presentan desafíos comunes en el desarrollo urbano, la intensidad de los aspectos específicos relacionados a la sostenibilidad y la resiliencia varían de una ciudad a otra [10] (Tabla 1). Bogotá está ubicada en el centro de Colombia, en el flanco este de los andes, a 2625 msnm [39]. La extensión de la ciudad es de 1637 km2 [40] y está dividida políticamente en 20 localidades [26]. Bogotá tiene una tasa de crecimiento poblacional anual de 1.3%, con un promedio de densidad poblacional de 4,876 habitantes/km2 [41]. Posee una estructura relativamente compacta con alta consolidación de densidad poblacional, y con una gran concentración de desarrollo informal en áreas periféricas [42]. El crecimiento de los asentamientos informales y los pronunciados niveles de desigualdad de ingresos son los desafíos más importantes en Bogotá. La estimación de áreas verdes por habitante es de 10 m2 [43]. Buenos Aires se encuentra ubicada en la región centro-oriental del país, en la orilla occidental del río de la Plata, en la llanura pampeana, a 25 m sobre el nivel del mar [44]. La extensión de la ciudad es 204 km2 [45]. Buenos Aires está oficialmente dividido en 48 barrios; sin embargo, la administración política y administrativa de la ciudad se distribuye a través de quince comunas que, en la mayoría de los casos, cubren más de un barrio de Buenos Aires [44]. Tiene una tasa de crecimiento poblacional anual de 1,5%, con una densidad poblacional media de 14,970 habitantes/km2 [45]. Buenos Aires presenta una pronunciada diferenciación socio-espacial con un fuerte crecimiento suburbano y periurbano [29]. El crecimiento urbano informal y precario, la pobreza, la desigualdad y el crimen son los desafíos más urgentes en Buenos Aires [10]. La estimación de las áreas verdes por habitante es de 6 m2 [44]. Lima se encuentra en la costa central del país, a orillas del Océano Pacífico, limitada por el desierto costero y la construcción formal de la ciudad, la cual se ha desarrollado mayormente en tierras agrícolas de las llanuras inundables de tres ríos: Chillón, Rímac y Lurín. Lima está a 154 m sobre el nivel del mar [46]. La ciudad tiene una extensión de 2812 km2 y está dividida en 43 distritos [47]. La tasa de crecimiento poblacional anual es de 1,6% [48] y su densidad poblacional media es de 3,328 habitantes/km2 [49]. Lima es una aglomeración metropolitana altamente segregada con profundos contrastes entre los sectores de altos ingresos y de bajos ingresos [50]. En Lima hay una gran cantidad de crecimiento urbano informal y una demanda insatisfecha de servicios básicos como agua potable, transporte y vivienda [10]. La ciudad cuenta con 3 m2 de áreas verdes por habitante [51]. Ciudad de México se encuentra en el Valle de México a 2240 m sobre el nivel del mar [52]. La ciudad posee una extensión de 1485 km2 y está dividida en 16 delegaciones [53]. La tasa de crecimiento poblacional anual de la Ciudad de México es de 0.3%, con una densidad poblacional de 5,967 habitantes/km2 [54]. La creciente inseguridad, la fragmentación socio-espacial y las condiciones de vivienda precarias son los desafíos sociales más importantes en la Ciudad de México [10]. La Ciudad de México tiene 13 m2 de áreas verdes por habitante [55]. Santiago se encuentra en el valle de Santiago, rodeado por los Andes, a 520 m sobre el nivel del mar [56]. La ciudad tiene una extensión de 640 km2 y está políticamente dividida en 32 comunas [57]. Tiene una tasa de crecimiento poblacional anual del 1.0% y una densidad poblacional media de 2,304 habitantes/km2 [58]. Las políticas de vivienda de Santiago han reducido los problemas de vivienda informal; sin embargo, estas políticas han llevado a una profunda segregación social [10,59]. Santiago posee 4 m2 de áreas verdes por habitante. Tabla 1. Megaciudades de América Latina seleccionadas y su respectiva información relevante Ciudad de Bogotá Buenos Aires Lima Santiago México País Colombia Argentina Perú México Chile Altitud media (msnm) 2625 25 154 2240 520 Extensión (km2) 1637 204 2812 1485 640 16 División administrativa 20 localidades 48 barrios 43 distritos 32 comunas delegaciones Área verde (m2/hab) 10 6 3 13 4 Precipitación media anual 818 1040 16 749 390 (mm) Temperatura media anual 13.5 16.8 18.8 15.7 14.4 (°C) Tasa de crecimiento 1.3 1.5 1.6 0.3 1.0 poblacional (%) Crecimiento Falta de Segregación de Desafíos mahs relevantes Inseguridad servicios Inseguridad socio- asentamientos básicos espacial informales 2.2. Cuantificación de las áreas verdes a través de las 5 megaciudades Las megaciudades de AL ocupan diferentes cantidades de espacio físico, así como su construcción se ha dado sobre diferentes tipos de cubiertas terrestres. Por lo tanto, se consideraron tres características para definir los distritos incluidos en este estudio: (1) porcentaje de la población rural; (2) ubicación; y (3) tamaño. Se excluyeron los distritos que tenían más del 50% de población rural y/o que eran predominantemente rurales o adyacentes a la orilla del mar. Dentro y entre las cinco megaciudades, los tamaños de los distritos son diferentes. San Telmo en Buenos Aires es el distrito más pequeño con un área de 1.2 km2; y Tlalpan en la Ciudad de México es el distrito más grande con 312 km2. Se excluyeron los distritos menores a 3 km2 y/o mayores a 100 km2. Para crear una muestra comparativa, diez distritos por ciudad fueron seleccionados al azar de los distritos que eran más grandes que 3 km2 y más pequeños que 100 km2. Google Earth Pro (versión 7.3) se utilizó para obtener los datos de cobertura terrestre con imágenes del 2013. El área total muestreada por distrito fue de 2.5 km2, dividida en cinco polígonos seleccionados al azar de 0.5 km2, los cuales representaron la mayor parte del área en la mayoría de los distritos muestreados. En resumen, este estudio consideró cinco polígonos por distrito equivalente a 50 polígonos por ciudad, con un número total de 250 polígonos (Figura 1). Las mediciones del área verde se establecieron a una altitud de 500 m sobre el nivel del suelo. Las áreas verdes urbanas se definieron como áreas cubiertas con cualquier tipo de vegetación según lo descrito por Wolch et al. (2014) [18]. Por lo tanto, todos los tipos de cubierta vegetal se muestrearon dentro de todos los polígonos, desde árboles individuales hasta bosques urbanos para medir la disponibilidad total de las áreas verdes, incluyendo los espacios públicos y privados. Figura 1. 250 polígonos muestreados distribuidos a través de las cinco megaciudades (a) Bogotá; (b) Buenos Aires; (c) Lima; (d) Ciudad de México, (e) Santiago. Imagen Landsat/Copernicus descargada de Google Earth. 2.3. Estimación de variables socio-económicas a través de las cinco megaciudades Densidad Poblacional: Esta variable fue estimada como el número de habitantes por km2 de cada distrito en año 2013 (Anexo B). Esta variable fue incluida en el análisis entre ciudades debido a que García & Riera (2003) han demostrado que las personas están dispuestas a pagar para vivir en barrios menos densamente poblados [60]. Otros estudios también han expresado que las personas prefieren barrios con baja densidad poblacional y baja densidad de unidades de vivienda [61,62]. Densidad empresarial: Esta variable es el resultado de dividir el número total de negocios por el área total del distrito en el 2013 (Anexo B). Esta variable muestra la relación probable entre un área más económicamente activa y la disposición a pagar para vivir en dicha área. De acuerdo con Des Rosiers et al. (2000) y Yu et al. (2012), existe una relación positiva entre los precios inmobiliarios y la proximidad a los centros comerciales, lo que sugiere que el atractivo de las instalaciones comerciales repercute en las decisiones de los hogares, lo que se traduce en una mayor demanda y, por lo tanto, precios y rentas más altas [63,64]. Tasa de criminalidad: Esta variable se calcula como el número de delitos registrados por km2 en cada distrito en el año 2013 (Anexo B). Los hallazgos previos sugieren que existe una importante relación entre las tasas de criminalidad y los valores de las propiedades. Ihlanfeldt y Mayock (2010) sugieren que los compradores de vivienda están dispuestos a pagar primas no triviales por viviendas ubicadas en barrios con menores incidencias de asaltos agravados, robos y delitos [65]. De hecho, la reducción de la tasa de criminalidad tiene un beneficio inmediato en los precios de los bienes raíces, pero también, los beneficios pueden derivarse por períodos de 4 a 6 años [66]. 2.4. Caso de estudio: Efecto de las variables de ecología del paisaje en los índices de precios hedónicos Buenos Aires y Lima fueron elegidos como casos de estudio para un análisis más profundo en el que evaluamos cómo las variables de ecología del paisaje (tamaño medio de parche, conectividad de parches, etc.) afectan los precios inmobiliarios, en un esfuerzo por obtener una comprensión más profunda de cómo y cuándo los atributos ecológicos de las áreas verdes pueden afectar el potencial de la provisión de servicios ecosistémicos de las áreas verdes urbanas. En este conjunto de datos, Buenos Aires y Lima representan los extremos de las condiciones climáticas, con una precipitación anual media más baja en Lima (16 mm) y la más alta en Buenos Aires (1040 mm). Las imágenes de satélite de Buenos Aires y Lima de marzo del 2013 se descargaron de Google Earth Pro (Versión 7.3, Google Inc., Mountain View, CA, EE. UU.) y se procesaron en ArcGIS 10.3 (ESRI, Redlands, CA, EE. UU.). Primero se georreferenciaron las imágenes y se proyectaron en forma geográfica. Luego se extrajeron las áreas de interés de los cinco polígonos seleccionados al azar de 0.5 km2 por distrito. Utilizamos Segment Mean Shift de la caja de herramientas de Spatial Analyst para segmentar las imágenes en objetos de al menos 10 píxeles. A continuación, seleccionamos manualmente los objetos que correspondían a nuestra definición de áreas verdes urbanas o "parche verde" dentro de las áreas de interés. El área de cada parche se calculó utilizando la función de cálculo geométrico en la tabla de atributos de ArcGIS. La distancia promedio del vecino más cercano entre los parches verdes se calculó utilizando la Distancia euclidiana en la caja de herramientas Estadísticas espaciales. 2.5. Análisis estadístico Para homogenizar los datos y simplificar la interpretación del análisis de índices de precios hedónicos a través de las ciudades, las variables independientes y dependientes fueron normalizadas previo al análisis estadístico tomando la siguiente fórmula: X*i = Xi/Xmax donde X*i es cada variable dependiente después de la normalización. Xi es la variable previa a la normalización y Xmax es el máximo valor obtenido para cada variable en su respectiva ciudad. Para el estudio de caso (Lima and Buenos Aires) no se realizó una normalización de los datos antes del análisis ya que las diferentes escalas no son un problema dentro de las ciudades individuales. Los datos transformados y no transformados se analizaron mediante regresión lineal múltiple utilizando XLStat (Versión 2014.5, Addin Soft, París, Francia). 3. Resultados 3.1. Comparación de los índices de precios hedónicos a través de las megaciudades A partir de una regresión múltiple con cuatro variables independientes, obtuvimos un modelo con precisión admisible (R2=0,62, Tabla 2A, Figura 2). Los valores ŋ2 indican que el área verde urbana fue la variable predictiva más importante explicando el 52% de la variabilidad de los precios inmobiliarios en las megaciudades consideradas en este análisis. La densidad de negocios también tuvo una modesta influencia en los precios inmobiliarios con un valor de ŋ2 de 10, mientras que la tasa de criminalidad y la densidad poblacional tuvieron impactos no relevantes (Tabla 2). Sin embargo, también hubo un error considerable con una variabilidad inexplicada que representa el 39% de la suma de cuadrados; por lo tanto, no se obtuvo una comprensión completa de los factores que determinan los precios inmobiliarios en el análisis a través de las ciudades. Tabla 2. (A) Estadísticas resumidas para los modelos de regresión múltiple en los que (A) los precios inmobiliarios se correlacionaron con cuatro variables: área verde, densidad poblacional, densidad empresarial y tasa de criminalidad, (B) área verde y atributos de parche verde vs. precios inmobiliarios en Lima, y (C) área verde y atributos de parche verde vs. precios inmobiliarios en Buenos Aires. Variables: Av/at = área verde/área total, TGS = área verde total, TNP = Número total de parches, MPS = Tamaño medio del parche, LPS = Parche más grande, ANN = distancia al vecino más cercano, PopDen = Densidad poblacional, BusDen = Dendisad empresarial, CrimDen = Tasa de criminalidad. df, grados de libertad; SS, suma de cuadrados; MS, cuadrados medios; F, F-test; Pr, probabilidad, ŋ2 porcentaje de variación del R2 explicado por cada variable independiente, ŋ2 es obtenido de la suma de cuadrados parcial entre la suma de cuadrados total por 100. * Variables después de transformación (normalización) utilizadas en la regresión múltiple para el primer análisis. Normalización: X*i = Xi/Xmax. Xi: variable antes de la normalización Xmax: máximo valor obtenido para la variable para cada ciudad. Figura 2. Relación entre los precios medidos y los precios estimados a nivel de distrito. Valores estimados obtenidos por regresión múltiple: Precio Estimado Normalizado (NOR) = 0.35 + 0.71 * av / at NOR - 0.053 * Densidad de Población NOR + 0.34 * Densidad de Negocios NOR - 0.061 * Tasa criminalidad NOR. Ver la Tabla 2 para más detalles. La línea diagonal indica la relación 1:1. Los datos normalizados utilizados en este análisis se adhieren estrechamente a la línea 1: 1 (lo que indica una correspondencia estrecha entre los valores medidos y estimados, demostrando la solidez del modelo de regresión múltiple). 3.2. Caso de Estudio: Efecto de las variables de ecología del paisaje en los índices de precios hedónicos Los resultados de los estudios de caso (Lima y Buenos Aires) proporcionaron un fuerte indicio de que la efectividad de los índices de precios hedónicos para restringir el valor de los servicios ecosistémicos proporcionados por el espacio verde urbano puede ser específico al contexto. Los datos de Lima mostraron una fuerte relación entre el área verde total y los precios inmobiliarios, explicando el 83.8% de la variabilidad de los precios inmobiliarios (Tabla 2B). El impacto positivo y significativo del área verde total (TGS) confirmó además que la cantidad de área verde urbana es muy valorada por los residentes de Lima. El tamaño de parche más grande (LPS) del área verde en Lima tuvo un impacto menor en los precios inmobiliarios (ŋ2 = 7), mientras que el tamaño medio de parche (MPS), el número total de parches (TNP) y el vecino promedio más cercano (ANN) o conectividad entre parches fueron irrelevantes. En contraste, para Buenos Aires, las variables más relevantes respecto a la variación en los precios inmobiliarios fueron el tamaño de parche más grande (LPS) con un valor ŋ2 de 23 (Tabla 2C), seguido por distancia al vecino más cercano (ANN) o conectividad con un ŋ2 valor de 22. El área verde total (TGS) tuvo un impacto modesto que explica el 12.5% de la variabilidad en los precios inmobiliarios. El tamaño promedio del parche (MPS) del espacio verde urbano en Buenos Aires también tuvo un pequeño impacto, explicando el 7.7% de la variabilidad en los precios inmobiliarios. En Buenos Aires, hubo un error considerable con una variabilidad inexplicada que representa el 29.7%. En cambio, el término de error para Lima fue 7.8%. 4. Discusión y Conclusiones 4.1. Consideraciones generales Los resultados de este estudio refuerzan los hallazgos previos que muestran que el área verde urbana es un determinante importante de los precios inmobiliarios [35,67-69], y que los índices de precios hedónicos pueden proporcionar una estimación robusta del valor de los servicios ecosistémicos proporcionados por las áreas verdes urbanas a través de las ciudades de AL. Sin embargo, la variabilidad inexplicada en el modelo de regresión múltiple para el análisis a través de las ciudades (38.5%) también fue significativa y podría estar relacionada con una variedad de factores que no fueron tomados en cuenta en este estudio. Algunas variables sociales, como las percepciones del estado económico [26], la dinámica de los ingresos [70], la calidad escolar [71] y / o los valores culturales y espirituales como el sentido del lugar y la cohesión social [24,72] pueden proporcionar información adicional sobre los factores subyacentes que impactan los precios inmobiliarios. Además, el uso de variables de ecología del paisaje en el estudio de caso de Lima y Buenos Aires demuestra que, a pesar de las similitudes regionales, la relación entre el área verde y los precios inmobiliarios depende del contexto, y que los atributos ecológicos del área verde pueden afectar el potencial para la provisión de servicios ecosistémicos. Estas preferencias podrían estar relacionadas con varios factores, como la tendencia de crecimiento de la ciudad, la política del gobierno o, como sospechamos, el entorno físico. Parece lógico que la cantidad de área verde sea más apreciada en una ciudad hiper árida como Lima que en una ciudad como Buenos Aires, donde las precipitaciones son suficientes para mantener el crecimiento del bosque. Esto creemos que puede explicar por qué vemos una valoración más matizada de los espacios verdes en Buenos Aires centrada en los patrones espaciales del área verde, específicamente relacionados con el tamaño de las áreas verdes y la conectividad. Sin embargo, las diferencias en la calidad de la planificación de la ciudad también pueden ser importantes en esta comparación específica entre Buenos Aires y Lima [50,73] y un análisis más amplio que incluya datos de más ciudades podría proporcionar una mayor comprensión. Además, la relevancia de los servicios ecosistémicos difiere según las características ambientales y socioeconómicas específicas de una ciudad [24]. Por ejemplo, las áreas verdes urbanas pueden amortiguar eventos climáticos extremos como inundaciones, que pueden ser importantes para ciudades como Buenos Aires, pero para ciudades como Lima con poca o ninguna lluvia, este servicio ecosistémico no es relevante. La regulación de la calidad del aire es fundamental para todas las megaciudades, pero posiblemente sea esencial en ciudades como Santiago y la Ciudad de México, donde la topografía favorece la concentración de contaminantes en aerosol. Los bosques urbanos estabilizan las laderas, previniendo el daño de los peligros naturales, lo cual es crucial para Bogotá, pero no relevante para Buenos Aires, que se encuentra en una llanura costera. Para asegurar la planificación orientada a la resiliencia, se necesita una clasificación específica al contexto de los servicios ecosistémicos en las áreas urbanas de AL [74]; ya que será importante considerar el papel de las características culturales específicas de las relaciones entre personas-naturaleza para aumentar la calidad de vida de residente de AL [26]. 4.2. Reflexiones metodológicas Los planificadores urbanos requieren datos espaciales actualizados y precisos a escala de ciudad para comprender la dinámica y los procesos urbanos [75]. En América Latina, los datos sólidos a escala local y de ciudad son difíciles de obtener debido al limitado interés e inversión en la investigación preliminar; así como el poco esfuerzo para garantizar el acceso abierto de los datos relevantes. Nuestros resultados demuestran el importante papel que los índices de precios hedónicos pueden desempeñar para ayudarnos a comprender cómo el área verde genera servicios ecosistémicos en diferentes situaciones (por ejem., Buenos Aires vs. Lima). Se requiere investigación adicional para comprender mejor cuándo y por qué los atributos ecológicos de las áreas verdes afectan los precios inmobiliarios y así optimizar la forma en que invertimos y administramos las áreas verdes en las ciudades de América Latina, garantizando que las áreas verdes urbanas generen la máxima cantidad de servicios ecosistémicos posibles. Sin embargo, nuestra capacidad para lograr esto dependerá en gran medida de la disponibilidad de los datos y un marco común para las comparaciones, como lo hemos intentado aquí en las megaciudades de América Latina. Mejorar la calidad y cantidad de la adquisición y disponibilidad de datos podría mejorar significativamente los procesos de planificación urbana para ciudades latinoamericanas, lo que estaría directamente en línea con los Objetivos de Desarrollo Sostenible de las Naciones Unidas que apuntan a mejorar la buena salud y el bienestar y apoyar el desarrollo de ciudades y comunidades sostenibles (http://www.un.org/sustainabledevelopment/sustainabledevelopment-goals/, consultado el 3 de noviembre de 2017). 4.3. Relevancia de los resultados para las estrategias de planificación urbana Las altas tasas de urbanización en América Latina continúan presentando amenazas directas a la preservación de las áreas verdes urbanas [76]. La limitada capacidad administrativa en las ciudades de AL para planificar y controlar el crecimiento urbano viene causando la conversión de las áreas verdes urbanas en viviendas e infraestructura. Además, la creación de nuevos espacios verdes urbanos se está volviendo cada vez más problemática debido a los altos niveles de consolidación de tierras urbanas [77] y la superposición de la demanda de tierras para fines múltiples (por ejem., espacio verde urbano vs. infraestructura) [78]. Por otro lado, la planificación del espacio verde no solo está guiada por las teorías urbanas, sino también por los valores que las personas asignan a los espacios verdes [79]. Nuestros resultados demuestran que las personas que viven en las megaciudades de AL valoran significativamente el área verde, expresando sus preferencias a través del mercado inmobiliario. Sin embargo, como el estudio de caso muestra, hay un amplio margen para realizar nuevas investigaciones que nos permitan comprender mejor cómo y por qué los atributos ecológicos de los espacios verdes urbanos pueden ayudar a asegurar, mantener o incluso mejorar una gama de diferentes servicios ecosistémicos [80]. Ser capaz de hacer más con menos, en términos de la provisión de servicios ecosistémicos por el área verde urbana, es especialmente importante para los residentes urbanos más pobres en las ciudades de AL, los cuales generalmente viven en las periferias metropolitanas con déficits críticos del espacio verde urbano. VI. ABREVIATURAS % percentage °C centigrade scale / Celsius scale AMSL above mean sea level ANN average nearest neighbor AV/AT green space/total area BUSDEN Business Density CRIMDEN Crime Rate DF degrees of freedom F, F-test INHABITANTS/KM2 – inhabitants per square kilometers KM2 square kilometers LA Latin America LPS Largest patch size M2/INHABIT - square meters per inhabitant MM millimeter MPS Mean patch size MS mean squares POPDEN Population Density PR probability R2 Coefficient of determination SS sum of squares TGS Total green space TNP Total number of patches VII. ANEXOS ANEXO 1: CERTIFICADO DE PUBLICACION POR REVISTA FORESTS ANEXO 2: CORREO ELECTRONICO – NOTIFICACION DE ACEPTACION DE PUBLICACION ANEXO 3: CONSTANCIA DE PAGO DE PUBLICACION