1. Introduction
The world’s urban population is expected to increase from 55% in 2019 [
1] to 68% by 2050 [
2], thus growth of demand for ecosystem services in cities would ensue. The relation between demand and supply of ecosystem services varies with scale. Locally generated ecosystem services are more closely related to the living quality of the resident, and some of them are irreplaceable by other distant sources of ecosystem services (for example, mitigation of heat island effect) [
3]. Considering the numerous population size in cities, the social and economic value of ecosystem services within cities can be surprisingly high [
4]. Besides, a global assessment highlighted how massive urbanization is impacting biodiversity and ecosystems around the world negatively [
5]. Therefore, an improvement of urban ecosystem services could potentially benefit city residents and mitigate the loss of ecosystem services globally.
Yet despite the importance of urban ecosystem services evaluation [
3], most of the studies and the implementation of the research findings into land use policy, are in North America, Europe, and China [
6,
7] (case studies see New York City [
8] and Berlin [
9]). Besides, even though related evaluation tools like i-Tree have been widely applied in many cities around the world, urban ecosystem services research in Japan has been less addressed. A pilot study evaluated the ecosystem services of street trees in Kawasaki City in Japan using i-Tree [
10]. Other than that, only some case studies using a similar approach were found [
11,
12,
13].
Ecosystem services are estimated with a variety of methods, including indicators and valuation. Indicators are used to quantify the state and change of the objects of interest. Some of the commonly used indicators are crop yield for food production, carbon storage and carbon sequestration for climate change mitigation, and runoff reduction for hydrological regulation. Regarding the valuation, two methods are applied to estimate ecosystem services’ monetary value. One is the traditional economic method using firsthand data, including the stated preference method and revealed preference method. Though the empirical, field-based method can provide more accurate results [
14], it is time-consuming and limited on the scale. Therefore, the other method, value transfer (or ‘benefit transfer’) is widely used in ecosystem evaluation [
15], for which the monetary value estimation of one location (the ‘reference ecosystem’) is transferred to another (the ‘target location’) [
16]. The value transfer method is frequently applied in regional services estimation based on the area of land use/land cover types and per unit area ecosystem service value of each type. In these studies, cities are categorized as ‘urban area’ or ‘built-up area,’ and the ecosystem service of the category is estimated with a constant per unit area ecosystem service value. Particularly, the per unit area ecosystem service value for urban ecosystem from Costanza et al. [
17] has been widely applied (e.g. see [
18,
19]). Some other research modified the per unit area ecosystem service value based on the local context like scarcity value effect [
20]. However, the land use/land cover-based value transfer method could cause uncertainty in urban ecosystem service estimation since it ignores the high heterogeneity in cities and rapid change of land use/land cover [
6,
15]. To get a more specific per unit area ecosystem service value for urban ecosystems, within-city research and inter-city comparison research is needed.
Among the service providers in urban ecosystems (e.g., forest patches, waterways and lakes, parks, brownfields, urban agriculture [
6,
21,
22]), the urban forest is one of the foremost. As a crucial local ecosystem services provider in cities, urban forest functions in many services like carbon storage and sequestration, noise reduction, air quality improvement, energy conservation, and recreation [
3,
23]. While the ecosystem services of urban forests might have been underestimated since many previous studies focused on remnant forests or street trees (e.g., [
24,
25]), partially due to data availability. However, the dispersed green spaces such as private gardens have been less studied, despite the fact that their importance to urban ecosystem services has been proved [
21,
22,
26].
To estimate the ecosystem services of urban forests more precisely, i-Tree Eco has been applied worldwide in more than one hundred countries. Developed by the United States Department of Agriculture, i-Tree Eco allows users to calculate several ecosystem services (carbon storage and sequestration, pollutants removal, runoff reduction, etc.) of each tree with field investigation data of tree species, size, and condition. Though i-Tree Eco enhances users to manage urban forest more accurately, even at a single-tree level, most research only presented the results of inferred total ecosystem services of the whole research area (e.g., see [
27]) or results by species [
28,
29]. One possible reason is being guided by the automatically generated report of the tool. These results, however, provide little information on the link between within-city heterogeneity and urban ecosystem services. Only a few research reported ecosystem services across land use/land cover within cities [
21,
30].
To address the gaps mentioned above, we conducted an urban ecosystem services evaluation at a Japanese city, Kyoto. The study is partially aimed at enriching the database of urban ecosystem services with detailed ground-based investigation data and the i-Tree Eco tool. Another main objective of this article is to link urban heterogeneity and urban ecosystem services by comparing ecosystem services across land use. We expected that ecosystem services would differ across land use types.
In this study, a pre-stratified sampling method based on the area of land use classes was applied for field data collection, then the i-Tree Eco tool was used to calculate the urban forest structure, tree compensatory value, and ecosystem services. The ecosystem services, including carbon storage and sequestration, air pollutants removal, and runoff reduction, were estimated for the entire study area and allocated to each tree, then further grouped by quadrat. We compared ecosystem services at both quadrat level and single-tree level across land use classes. For a better understanding of the link between heterogeneity and ecosystem services, we also compared the results of Kyoto City with the studies of other cities.
2. Materials and Methods
2.1. Study Area
Kyoto City (35°19′16″ N–34°52′30″ N, 135°33′33″ E–135°52′43″ E), the capital of Kyoto Prefecture, is located in Kyoto Basin of Kansai region, Honshu Island, Japan, with an area of 828 square kilometers. The city is dominated by a humid subtropical climate with hot, humid summers, and cold, dry winters. It is one of the ‘Cities designated by government ordinance of Japan’ with a population of 1.47 million (0.73 million households) in 2019. The area of the built-up area of the city is 144 square kilometers.
As a planned capital, Kyoto city was founded when Emperor Kammu relocated the capital in 794. The Japanese borrowed the basic city layout from Chang’an, China in the Tang-dynasty, part of which is a grid spatial system dividing the city into blocks. The land use pattern of the city, however, mainly formed in modern times. During the infrastructure promotion at the end of the Meiji era (1868–1912), the city center was constructed based on the traditional commercial area. The administration boundary expanded significantly in the Taisho era (1912–1926), and an expansion and construction of the industrial area and residential area was achieved based on the urban planning laws [
31]. Kyoto city is now, overall, a mono-centric city. The city center is mainly used as a commercial area. The industrial area is mostly located in the west and south of the city. The residential area is in the surrounding area (
Figure 1).
2.2. Tree Data Collection
According to the urban planning system and City Planning Law of Japan, urban land use is categorized into 12 classes, with a regulation on the architectural form and use of the buildings constructed [
32]. We aggregated them into 6 classes from city fringe to city center (
Table 1): ResLow (Low-rise residential zone), ResHigh (mid/high-rise residential zone), ResOther (Other residential zone), Ind (Industrial zone), ComNbr (Neighborhood commercial zone), Com (Commercial zone). For field investigation, 200 quadrats (20 m × 20 m) were established, including the alternative ones, by stratified sampling method based on the area of the land use classes [
33]. The field investigation was conducted between May and August in 2019. The number of the quadrats accessed and investigated (n = 175 [
34]) for each land use class is shown in
Figure 1.
Following the i-Tree Eco workbook [
33], the information of the quadrats and trees was collected. For each quadrat, we took photos of the surrounding environment and the vegetation. Information of each tree (woody plants higher than 2 meters) in a quadrat, including species, height, diameter at breast height (DBH), canopy missing percentage, crown size, crown health condition, and crown light exposure, were collected. A total of 1240 trees (of 118 species) was recorded in 151 out of the 175 quadrants (
Table 1).
2.3. Evaluation of Ecosystem Services and Monetary Value
i-Tree model has been widely used to help managers and researchers to quantify urban forest structure, ecosystem services, and tree monetary value. We calculated three values of each tree: compensatory value, representing compensation for the loss of a tree [
35,
36]; monetary value of carbon storage, representing the cumulative result of net carbon sequestration for years; annual ecosystem services, including carbon sequestration, air pollutants removal, and runoff reduction. Though cultural service is one of the most critical components of ecosystem services in cities, i-Tree is not capable of calculating it for now. We will briefly introduce the method for structure and ecosystem services evaluation, and valuation of tree monetary value in the following sections; for more details refer to i-Tree method documentation [
37]. To improve the accuracy of results, a modified i-Tree model with local parameters of Kyoto City was applied (see
Table S1 for model details and parameters list).
2.3.1. Structure
Leaf area is estimated based on species, total height, crown base height, crown width, and percent crown missing. The method is a species-specific regression equation with a shading coefficient (percent light intensity intercepted by foliated tree crowns) for deciduous urban species, while a shading coefficient of 0.91 is applied for conifer trees [
38]. Leaf area index (LAI) is calculated with leaf area and adjusted with the overlap of tree crowns or light exposure. Leaf biomass is calculated based on leaf area with species-specific convert factor. Total biomass for each tree is calculated using species-specific allometric equations from the literature with DBH and total height (see [
39,
40] for the attributes and references of the equations).
2.3.2. Carbon Storage and Carbon Sequestration
Carbon storage is estimated based on biomass and carbon content. For evergreen and palm species, leaf biomass is added. Carbon sequestration is estimated based on the growth rate. The growth rates are estimated with the measurement of radial growth increments [
40], length of the growing season, and the growth adjustment factor of crown health and crown light exposure [
38]. For valuation of the ecosystem services, the social cost for carbon in Japan (10,600 Yen, which is about 96 US dollars per ton carbon) from the Japanese government document [
41] was applied.
2.3.3. Air pollutants Removal and Health Benefits
Air pollution removal is estimated using the percent tree cover and leaf area index. The pollutants estimated include nitrogen dioxide (NO
2), ozone (O
3), particulate matter less than 2.5 µm (PM
2.5), and sulfur dioxide (SO
2). In the locations supported more sufficiently in i-Tree Eco (e.g., cities in the US and Canada), the tree data is merged with local pre-processed weather and air pollution concentration data for the evaluation of pollutants removal. However, in this case, since Kyoto City is not officially supported by default, we input the local weather data from local monitor stations manually. The value of air pollutant removal is assessed by the BenMAP method [
42] that estimates avoided costs for adverse health incidences based on the air quality improvement and medical records across the US.
2.3.4. Runoff Reduction
Runoff reduction in i-Tree Eco is estimated based on the difference between the runoff with current tree cover and that without trees [
40]. In the simulation, rainfall interception of trees and runoff are calculated mainly by precipitation, leaf area index, and infiltration with a time step of an hour [
43]. One limitation of the model is that the water reaching the pervious surface is assumed to be absorbed by the soil, while the water reaching the impervious surface is assumed to become urban surface runoff. Besides, though the impervious cover rate is estimated by JAXA satellite imagery, the number is assumed to be constant across the research area. To reflect the local economic benefit of the ecosystem service, we used the stormwater control facilities cost estimation of Suita City of Japan (719 yen per m
3, which is about 7 US dollars per m
3) [
13] for the valuation.
2.3.5. Compensatory Value
The compensatory value of trees is estimated using the guideline of the Council of Tree and Landscape Appraisers [
36] in i-Tree Eco [
37]. The compensatory value of a tree is determined by replacement cost, DBH, and a location-specific per unit trunk area cost. For palm trees, the cost to clear trunk is also considered. The values of these parameters have been compiled for numerous states in the US; while for other countries, an average value of replacement cost and per unit trunk area cost is applied.
2.4. Data Analysis
The collected and calculated data for every single tree was then added to get a quadrat dataset, including average DBH, average LAI, the total number of trees, and the total ecosystem services of each quadrat. Since the assumption of normality for the metrics is violated in this case, the non-parametric statistic method, Kruskal–Wallis rank-sum test, was used to analyze the difference of DBH, LAI, the number of trees, and each ecosystem service among land use classes. For the statistical group comparison where a significant difference was detected, Dunn’s test was then applied for a post hoc pairwise comparison. Similarly, at the single-tree level, Kruskal–Wallis rank sum test and Dunn’s test were used to test the differences of DBH, LAI, and each ecosystem service across land use classes respectively.
Furthermore, a species-specific analysis was used to compare the single-tree ecosystem services across land use classes by species. To achieve a robust result, only widespread species presenting across a sequence of land use classes with at least 3 individuals for each land use class were analyzed. The target species include Acer palmatum Thunb., Ginkgo biloba L., 1771, Ligustrum lucidum Ait., Nandina domestica Thunb., Osmanthus fragrans Lour., Podocarpus macrophyllus (Thunb.) Sweet, 1818, Prunus x blireana, Quercus x alvordiana, and Zelkova serrata (Thunb.) Makino.
All the analysis was conducted in R (version 4.0.3), and the difference was considered significant at p < 0.05. The function was applied for Kruskal–Wallis rank sum test and dumn.test function from dumn.test package was used for the post hoc comparison.
5. Conclusions
The main purpose of this study is to demonstrate the link between heterogeneity of a city and urban ecosystem services, and the potential contribution of dispersed green to urban ecosystem services. This study captured the structure, ecosystem services, and monetary values of the urban forest in Kyoto City. For urban forest structure analysis, the Ind zone has more mature trees with higher DBH than the other land use; residential zones have a higher proportion of trees with larger LAI. The comparison across land use classes shows that ecosystem services are different across land use at the single-tree level, though no significant difference was detected at the quadrat level. The results indicate that the comparison varies with scale. Though less addressed in previous research and not statistically significant, the residential zones have higher average and median ecosystem services values than the other land use. The result suggests a potentially important contribution of dispersed green space (like private yards) to urban ecosystem services. For a more comprehensive and precise evaluation of ecosystem services of urban forests, further research considering heterogeneity, scale effect, and varieties of green space type are needed.
The results also provided insight for practice. We identified a mismatch of air pollutants removal and emission across land use types that the air cleaning services of urban forests in commercial areas should be improved. Furthermore, a species-specific method can help in making urban planning aimed at increasing ecosystem services.