Provision of Allotment Gardens and Its Influencing Factors: A Case Study of Tokyo, Japan

Abstract

Allotment gardens (AGs) are widely used in metropolitan areas around the world to offer agricultural opportunities to urban residents. However, there are not enough individual plots for residents to rent for urban gardening, and research on AGs from a city-wide perspective is ongoing. In addition, AGs have a long history in Tokyo, yet few international studies on the current situation of AGs have addressed Asian cities. Thus, this study intends to analyze the provision of AGs and its influencing factors in Tokyo. Using ArcGIS combined the 472-points dataset created by geo-coordinate mapping with urban GIS data to reveal spatial characteristics in four dimensions. Results demonstrate that most AGs are in the urbanization promotion area; most municipalities have AGs; AGs are concentrated within 20 to 30 km from the center of Tokyo; the AGs’ clusters are located at the municipal boundaries. We conducted multiple regressions to determine the influencing factors at the municipal level, with the provision that AGs are related to population density, land price, and the ratio of productive green space. The policy implication of this study is that policymakers need to consider the siting strategy of AGs based on spatial characteristics of AGs.

1. Introduction

Allotment gardens (AGs) are land-based gardening practices in urban areas [1,2,3,4,5,6]. AGs belong to urban agriculture, offered to households in small plots mainly for recreational gardening, especially for growing vegetables and fruits [7,8]. The family usually pays rent for the plot and involves responsibility and commitment to the individual plot. Land use planning of AGs is encoded in the planning system [9,10]. Typically, plots are sited collectively and administered by the municipality, city, or organization, although there is also private provision [11,12]. AGs’ spaces are semi-public, the paths through the site and other facilities are public, and the single plots are private [9]. In the early history of urban civilization, the courtyard spaces of private houses usually had vegetable gardens. In the context of rapid urbanization, it is difficult for urban dwellers to have their vegetable gardens; therefore, single private plots of allotment gardens are receiving more and more attention [13].

AGs offer a variety of benefits and have been the subject of numerous scientific studies. During the Industrial Revolution, AGs responded to the unhealthy living conditions and poverty in the workers’ settlements [14]. After the Second World War, the benefits of gardens were primarily to provide food [15,16,17]. Historically, allotment gardens emerged as welfare for low-income families based on the idea of self-sufficiency. Over time, allotment gardens achieved multifunctional urban land use [18]. They began transforming from social welfare to leisure activities, becoming a place between agricultural production and passive recreation [19]. Many indirect benefits came to the fore and continue to exist, such as improving the quality of life and urban environment. Previous studies have shown that AGs provide a stress-relieving sanctuary [20] and a valuable connection to nature [21]. They also contribute to a healthier lifestyle [22] and social opportunities [23]. Allotment gardening is implemented as leisure [24] and recommended as a health-promoting activity for the general population [8]. Therefore, scholars regard the allotment garden as a part of urban green infrastructure [25,26,27,28,29].

Unfortunately, urban green infrastructure is rarely evenly provided. The inequality of green space in South Africa has existed since apartheid [30], the provision of urban green space is unfair among different social groups in Berlin [31], family gardens in Portland, Oregon, USA gather in the high-grade area in the urban core [32], and there are socioeconomic differences in the accessibility of healthy food [33]. Furthermore, urban authorities give prominence to nonagricultural uses on urban land allocation preference [34]. The location of AGs in the urban environment has been a problem since the beginning, resulting from the conflict with urban development needs. Access to AGs spaces is a challenge [35] and a rare opportunity for urban residents to participate in gardening activities [20], but this further generates multiple undesirable outcomes among residents, such as spatial dislocation, gentrification, and social isolation [36,37,38,39,40,41].

However, the rational spatial planning of AGs can control the disorderly urban sprawl and improve land-use efficiency [42]. The creation of AGs in peri-urban areas of cities shows the positive valuation [43]. Multiple ecosystem services and resident satisfaction from AGs can also increase land values and rents in the urban environment [44]. Recently, in countries with economic crises, AGs have become a new planning strategy embedded in spatial development visions to increase the investment attractiveness of cities [45,46,47]. Therefore, urban planners and city management need to examine the spatial planning of AGs more broadly.

Although urban AGs support sustainable planning strategies to revitalize urban space [45], and people being increasingly interested in urban agriculture and urban food [48,49], knowledge on the provision of urban AGs remains very limited. In a study conducted in 2014, the geographical characteristics of AGs in three European cities (Ljubljana, Milan, and London) were identified by analyzing aerial images [50]. In Germany, the spatial characteristics of 276 AGs in Leipzig were analyzed to assess their ecosystem services [11]. In Poland, the significance of the 86 AGs in Poznań in the urban green space system’s spatial development was examined and classified with a method of valuation that could be used in urban planning [51]. In the UK, a study has shown that the provision of AGs in London does not meet demand and that the 682 AGs are not evenly distributed across London [52]. Overall, the above studies on the provision of AGs are fragmented and do not identify the geographical distribution characteristics of urban allotment sites, concentrating in European cities. It is necessary to study the provisioning status of AGs in Asian cities, Tokyo, which was selected.

To cover broader publications, we surveyed the Japanese language literature. Research on AGs in Tokyo has focused on three main themes, examining financial support policies for AGs from the perspective of landowners [53]; examining organic waste utilization systems from the perspective of resident participation [54]; and developing programs to support AGs from the perspective of analyzing user motivation [55]. Research related to the spatial provision of AGs in Japan mainly focuses on four aspects, how landowners’ attitudes affect the location of AGs [56]; the relationship between geographical location and demand for AGs [57]; the relationship between vacancy rates of AGs and geographical location [58]; and the possibility of AGs using park space [59]. In general, the analysis of the spatial provision of AGs in Tokyo offered by previous studies is insufficient and requires the accumulation of basic knowledge.

The present study attempts to fill the aforementioned knowledge gaps in AGs research, selects Tokyo as the study area, and the overall objectives are: (1) to identify the characteristics of the provision of AGs based on geospatial distribution status; (2) to explore the factors influencing the provision of AGs at the municipal level. The remainder of this paper is structured as follows. Section 2 contains a brief review of the development of AGs in Japan and describes the study area and data. Section 3 mainly explains the methodology, including GIS-based methods and statistical analyses strategies. Section 4 analyzes the characteristics of the spatial provision of AGs in four dimensions and identifies the factors that influence supply in terms of the municipal dimension. The results are discussed in Section 5. Finally, Section 6 summarizes the conclusion and highlights the implications of this study.

2. Study Area and Materials

Different countries have different measures to support allotment gardening, with the UK allotment incorporated into the Local Agenda 21 [60] and Germany developing planning document for AGs [10]. This section first clarified the development of AGs in Japan and presented an overview of AGs in Tokyo. We then explain the data sources in detail, which are available for consultation for other cities in Japan.

2.1. Study Area

2.1.1. Development of AGs in Japan

We have chosen the case of Japan, and a discussion of the historical background and relevant legislation is inevitable to capture the development of AGs. On this basis, we will further summarize the current characteristics of Japanese allotment sites.

The first attempts at AGs in Japan date back to the 1920s, when Japanese scholars, influenced by Howard’s Garden City [61] and European AGs [62], proposed the “Kaen Toshi” (Garden City) and built house with AG in Osaka, Japan, to improve the poor residential environment of the time [63]. From the 1930s to the 1940s, Japanese planners made repeated attempts to develop garden plots. There are two fundamental attempts: one is to set up allotment plots in the urban central park, and the other is to set up AGs in the peri-urban [64]. Subsequently, The Japanese Agricultural Land Law 1952 restricted the sale and lease of agricultural land, which meant that AGs became impossible at this time [65]. In the 1960s, however, there were instances of farmers privately leasing farmland to urban dwellers in the peri-urban [66], and since 1975, driven by this bottom-up activity, the government took steps to allow the development of AGs, and legislations were successively enacted (see

Table 1. Relevant legal basis for the provision of AGs in Japan.

Year	Relevant Legislation	Content
1975	Recreational Agricultural Land Notice	▪ Urban residents can enter the agricultural land to experience gardening by signing a contract.
1989	Agricultural Land Rental Law	▪ Make Land Available, municipal institutions, and Japan Agricultural Co-operatives (JA) are allowed to open AGs.
1990	Allotments Promotion Law	▪ Enhance Facilities, such as toilets or rest chalets.
2005	Amendment to Agricultural Land Rental Law	▪ It became possible for landowners to rent out land for use as AGs.
	Amendment to Allotments Promotion Law	▪ Enterprises and Non-profit organizations (NPOs) are allowed to open AGs.
2018	Facilitation of loan for Agricultural Land	▪ The application processes of opening AGs in productive green spaces are simplified.

Source: Author’s elaboration based on [62,63,64,65,66,67].

).

In the past few decades since the relevant legislations of AGs came into force, AGs in Japan have developed steadily and formed their characteristics (see Figure 1). They are characterized as follows: (a) small plots located on public or private land in urban areas; (b) set up by municipal institutions, JA, landowners, enterprises, or NPOs; (c) semi-public (usually with a simple fence and individual plot for tenant use only); (d) multifunctional land use (cultivate vegetables and flowers integrating other uses of space such as recreational activities, the living worth of the elderly, educational or social practices for the students and children); (e) only a symbolic payment; (f) limited the duration of the lease (ranging from 1 to 5 years); (g) used by the surrounding neighborhood; (h) with certain facilities (toilet, rest hut, water supply, tools storage); (i) distribute cultivation manual (guide cultivate and waste disposal, limit the use of pesticides); (j) the harvest belongs to the tenant (it can be sold not for profit); (k) no permission for overnight stay. The rationale behind AGs in Japan is to provide a place for people to relax and sustainably use and protect the city’s farmland [68].

2.1.2. Overview of AGs in Tokyo, Japan

Tokyo is located in Japan’s Kanto region and the only administrative district in Japan named after the “Metropolitan”. Since 1943, Tokyo Metropolitan Government has administered the prefecture’s 23 special wards, the Tama area, and two outlying island chains. In addition, Tokyo is an essential world-class city in Asia and is traditionally one of the world’s top four world-class cities. According to the TOKYO STATISTICAL YEARBOOK, Tokyo’s park per capita was approximately 5.71 m² in 2018, while the park per capita in the 23 wards was only 4.32 m². It would appear that the per capita urban green space in Tokyo is far lower than a minimum target of 9 m² and an ideal value of 50 m² set by the World Health Organisation [69,70] and, statistically, around 80% of residents in Tokyo live in apartments without private gardens [71]. In this context, as the spatial dimension of functional urban green space [72], the research on its current configuration is essential for future planning and management.

In highly urbanized areas, such as the Tokyo Metropolitan Area (TMA), people strive for meaningful and regular nature encounters, recognizing and enjoying the benefits of nature experiences [73]. A survey conducted by the Tokyo Metropolitan Government indicated that Tokyo residents have high expectations for urban gardening [74]. However, Tokyo’s AGs are only available to a limited percentage of households and can be viewed as a limited family space. Those wishing to rent a private plot for gardening activities must apply to the municipal authorities. The municipality receives applications from residents during a centralized period, and if the number of households applying is greater than the number of single plots provided, a lottery is enforced to decide on the allotment gardeners. According to survey results from the Bureau of Industrial and Labor Affairs of Tokyo Metropolitan Government in 2017, the average competition ratio for AGs accepting applications in that year was 1.6 times, while the area with the highest competition ratio in Tokyo was the Ward Area, with an average of 1.8 times. The competition ratio of an AG in 2019, up to 3.4 times, was observed in Suginami city. It is evident that people are becoming increasingly interested in “farming” and that AGs have become widely popular and not easily accessible [75]. At the same time, this also means that the provision of AGs in Tokyo is insufficient.

In Tokyo, citizens usually participate in two types of AGs, “shimin noen” and “taiken noen”. Based on the development and functions of the gardens, Terada (2017) identified them as conventional allotment gardens (CAGs) and guided allotment gardens (GAGs), respectively [76]. There are some differences between CAGs and GAGs. The CAG is a garden where municipalities or agricultural cooperatives rent agricultural land and divide it into small plots of approximately 10–15 m² for gardening and leisure activities by urban residents, with the tenant free to grow vegetables or flowers or fruit, usually for a period of 1 to 3 years. The GAG is a garden set up by the landowners themselves and used by urban residents, where tenants experience gardening on a 30 m² plot under the careful guidance of farmers, with no strict duration restrictions, and can cultivate for several years. In this study, CAGs and GAGs were analyzed indiscriminately, as both AGs provide urban dwellers with plots to experience gardening.

Figure 2 shows the extent of the study area (total 51 municipalities: 23 special wards, 26 cities, three towns, and one village), which excludes the island section of TMA because of its geographic structure characterized by islands and its geospatial non-contiguity.

2.2. Data Source

In this study, we obtained AGs data and municipal data. AGs data are a dataset manually sorted based on open-source government data. Municipal data included urban geographic information system data and socioeconomic data, and the municipalities’ measurement is currently the smallest measurement unit of socioeconomic statistics in Tokyo.

2.2.1. AGs Data

An overview of the AGs data sources is shown in

Table 2. Data sources.

Sources	Items	Number of AGs	Periods
Homepage of Ministry of Agriculture, Forestry and Fisheries (MAFF) 1	▪ Name: List of Allotment Gardens. (31 March 2019 updated)	AGs: 320	April 2020~December 2020
	▪ Division: Urban Rural Communication Division of Rural Policy Planning Department of Rural Development Bureau.
Homepage of Tokyo Metropolitan Government 2	▪ Number of CAGs, Number of plots and total area of municipalities that have CAGs in Tokyo.	CAGs: 434	September 2020~December 2020
	▪ Number of GAGs, Number of plots and total area of municipalities that have GAGs in Tokyo.	GAGs: 111
Homepage of the 53 municipalities in study area	▪ Search for “shimin noen”.	CAGs: 286	April 2020~December 2020
	▪ Search for “taiken noen”.	GAGs: 51
	▪ Record the detailed address of AGs.
	▪ Record information on the number of zones, area and competition ratio for each AG.
	▪ Record relevant policies.
	▪ Record agricultural activities.
Homepage of National Association of Farming Experience Farms	▪ Record the detailed address of GAGs.	GAGs: 87	September 2020~December 2020
	▪ Record information on the number of zones and area for each GAG.

¹ Data are recorded with detailed information about each AG. ² Only statistics within each municipality are available here, not for each AG.

. The following data were collected for individual AGs: detailed address information of AGs, number of plots, area, area of the most plots, the usage fee for most plots, competition ratio for AG, and whether residents outside the municipality can use the service. We organized AG data obtained from four sources, deleted duplicate AGs, and obtained as much information as possible on AGs. Finally, we identified 472 AGs in the study area, which were the targets of this study (see Figure 2).

Table 3. Descriptive statistics for attributes of AGs (n = 472) in Tokyo.

Attribute	Mean	Median	Std.	Min.	Max.
Number of Plots	59.97	50	42.37	4	400
Area (m2)	1513.46	1094	1413.73	48	15,000

presents descriptive statistics for two critical attributes of the AGs. As one plot in the AGs is allocated for a household, the number of plots represents households with access to AGs. The area is the most fundamental attribute of environmental space, explaining the scale of the place. Therefore, we consider these two attributes to be critical for studying spatial distribution. For the area, it is essential to note that the Ministry of Agriculture, Forestry and Fisheries (MAFF) list has two indicators: the whole area and the farm area. In the current study, the area was represented by the farm area based on the AG’s farming service function. The significant difference between the maximum, minimum, and median values indicates AGs with a significant difference in scale. The value of the standard deviation reflects the large fluctuations between these samples.

2.2.2. Municipal Data

Multiple sources of geographic information and land-use data were used. First, we used numerical maps (basic national land information) published by the Geospatial Information Authority of Japan (GSI) and nationwide city, town, and village boundary data from ESRI Japan. Second, we downloaded the national land survey data from the Ministry of Land, Infrastructure, Transport and Tourism (MLIT) for urban and agricultural areas. Finally, in the regression analysis at the municipal level, we obtained the data of socioeconomic variables from the TOKYO STATISTICAL YEARBOOK(2019) [77]. The specific socioeconomic variables are described in the following section.

3. Methodology

In a nutshell, our workflow can be summarized as follows: (1) dataset processing; (2) spatial analysis; (3) exploratory regression. The first and second stages used GIS-based methods, as shown in Figure 3. The third stage was designed to explore the factors influencing AGs’ provision at the municipal level, and statistical analysis was carried out with SPSS software (IBM SPSS Statistics 26).

3.1. Method for Dataset Processing

As preparation for the analysis, we inputted the exact addresses of the 472 AGs into Google Maps to determine the geographical coordinates (WGS-84) of each AG and create a text dataset of points. We then used ArcGIS 10.7 software for geo-coordinate mapping and embedded information on each AG’s number of plots and area to create a point vector database to study the spatial distribution.

3.2. GIS-Based Spatial Analysis Methods

We used ArcGIS 10.7 for the analysis of the spatial distribution characteristics of AGs’ provision in Tokyo, and all operations were conducted in the projection coordinate system (JGD_2000_Japan_Zone_9), using the following four fundamental measures of spatial statistics.

3.2.1. Overlay Analysis

The overlay of points and polygons was primarily used to perform the spatial relationship judgment of whether the AG is in the specified parcel. After the overlay analysis and attribute information processing, it is possible to query which parcel the specified AG is distributed in by the attribute. We can also calculate how many AGs are in each specified parcel and count the number of plots and the total area within the parcel.

3.2.2. Mean Center

The mean center identifies the geographical center for the point dataset of AGs and measures the central tendency of point features. In this study, it represents the spatial distribution center of the 472 AGs.

3.2.3. Standard Deviational Ellipse (SDE)

The standard deviational ellipse captures the directional bias of AG distribution, describing three parameters: deviation along the longer axis (Major axis), deviation along the shorter axis (Minor axis), and rotation angle [78]. The longer axis reflects the direction of the maximum expansion of AGs, the shorter axis reflects the direction of the minimum expansion of AGs, and the rotation angle corresponds to the geographic direction of AG distribution. Ayhan et al. (2010) used SDE to study the spatial distribution of mosques in historical cities in Turkey [79]. Satoh et al. (2011) used SDE to compare the distribution of former and current residential relocation of residents who purchased their own homes in Tokyo [80]. In ArcGIS, the parameter “Ellipse Size” has three standard deviation levels. This study used 1_STANDARD_DEVIATION, indicating that the generated ellipse polygon encompasses approximately 68% of all input feature centroids of AGs.

3.2.4. Kernel Density Estimation (KDE)

KDE is commonly used to reflect the relative concentration of the spatial distribution of point elements. Sengoku et al. (2011) used KDE to determine the spatial extent of shopping areas [81]; Hong et al. (2018) used KDE to determine the density and layout of urban-type living houses around a university campus [82]. In this study, KDE was used to check the clusters of AGs. The calculation formula of kernel density is as follows [83]:

$$f\left(x_i,y_i\right)=\frac{1}{nh}\underset{i=1}{{\displaystyle \sum }^n} k\left(\frac{d}{h}\right)$$

(1)

where $f\left(x_i,y_i\right)$ stands for the kernel density value of AG at spatial location $\left(x_i,y_i\right)$; $h$ is the search radius; $n$ is the number of AGs whose distance from location $\left(x_i,y_i\right)$ is less than or equal to $h$; and the $k$ function represents the spatial weight function; and $d$ is the point distance between the current AG and $\left(x_i,y_i\right)$. Numerous studies have shown that the choice of the distance decay threshold needs attention [84], i.e., the setting of the parameter “Search radius” in ArcGIS, which is mainly related to the characteristics of the geographical phenomenon and its scale of analysis. For the parameter “output cell size”, the smaller the value, the smoother the generated density raster. According to AG distribution, these need to be tested several times to arrive at the desired value and effect. In our study, based on the density of raster location points and the research refinement requirements, we repeatedly tested and changed the default search radius to 3000 m and the output cell size of the raster image to 10 m to obtain more detailed results.

3.3. Regression Analysis

First of all, we need to emphasize that the AGs in Tokyo are highly territorial. Residents can only rent AGs provided by their municipality. Hence, exploring the factors influencing the provision of AGs is a study conducted at the municipal level.

This study used ordinary least squares (OLS) regression model to determine which independent variables explained variation in the provision of AGs at the municipal level. Given that AGs are rented on a household basis, we used the number of plots per 1000 households and the area of AGs per 1000 households as dependent variables. By referring to the existing literature, our original models included a range of socioeconomic variables related to income, population, housing, land, and urban agriculture for a total of eight potential independent variables. Tax levied per household was considered a potential independent variable because low-income people meet more of their needs for products from their gardens [32], and tax is a proxy for household income. We have included the population density variable because the demand for multifunctional green space is exceptionally high in areas with high population density and limited space [85]. Since previous publications suggest that the allotment gardeners have little access to gardens at home and more often live in flats or rented accommodation [86], two potential independent variables were included, the percentage of households living in rented houses and the percentage of families living in apartments. Many farmers in Tokyo rely on running flats and car parks for a large proportion of their income [87], using private agricultural land for real estate, which leaves less and less land available for AGs, so the average posted residential land prices should be included. The three potential independent variables related to urban agriculture are the population ratio of farm households, the area ratio of agricultural land, and the area ratio of productive green space. In Tokyo, the farmland provided for agricultural use within the city planning area is identified as the productive green space, an area in the City Planning Act that seeks to preserve planned urban farmland and is decided by the municipality for planning purposes. As the application processes of opening AGs in productive green spaces are simplified as described in Table 1 above, we include the land area ratio of productive green space.

We compared variance inflation factors (VIF) to test for multicollinearity and removed variables with VIF > 2 and then used stepwise multiple linear regression to eliminate the remaining non-significant (p > 0.10) variables. The applied stepwise regression was designed to leave a minimum set of independent variables in the regression model while maximizing the adjusted determination coefficient. Two final models of the different dependent variables included three identical independent variables. The expression of the final model is:

$$A{G}_i={\mathsf{\beta }}_0+{\mathsf{\beta }}_{1}POP{D}_i+{\mathsf{\beta }}_{2}RL{P}_i+{\mathsf{\beta }}_{3}PGL{R}_i+{\mathsf{\epsilon }}_i$$

(2)

where $A{G}_i$ represents the provision of AGs in Tokyo (either the number of plots per 1000 households or area per 1000 households), $\mathsf{\beta }$₀ is the intercept, $POP{D}_i$ is the population density, $RL{P}_i$ is the average price posted for residential land, $PGL{R}_i$ is the productive green land area ratio, and ε is the error term; $i$ = 1,…, n and n is the number of observations. Due to incomplete and unpredictable socioeconomic data for Tokyo towns and villages, we removed four observations with missing values, so the number of observations for this experiment is 49 (23 special wards and 26 cities, See Figure 2).

4. Results

The proposed approach was conducted as described in the previous sections. This section presents the empirical analysis of Tokyo, one of the most urbanized and densely populated spots on earth [88]. In addition, to obtain correct results and reduce the role of the random factor, the analysis has covered the most extensive number of existing AGs from official data; the 472 mappable AGs are used as the research object for this study. We focus on the spatial distribution characteristics of AGs’ provision under four dimensions: in the context of land use, at the municipal level, in the multi-distance zones from the center of Tokyo, and spatial clustering. Finally, stepwise multiple linear regression was used to explore the significant drivers for provision in the municipal dimension.

4.1. In the Context of Land Use

Figure 4 shows the city planning land use and AG locations in Tokyo. In 1968, the City Planning Act of Japan was enacted, which divided the city planning area (CPA) into urbanization promotion area (UPA) and urbanization control area (UCA). Agricultural promotion area (APA), aimed at preserving and cultivating agricultural land in the surrounding areas of urban areas, was established in UCA in 1970. In APA, agriculture-based maintenance is conducted, and urbanization is controlled. In addition, agricultural land area (ALA) has been established in APA to control urbanization more strongly. Using the overlay analysis in ArcGIS, we determined that 462 AGs were located in UPA, accounting for 97.88%. Nine AGs are located in the APA, with three AGs in ALA and six AGs outside ALA. In addition, one AG is situated outside the city planning area.

4.2. At the Municipal Level

Figure 5 shows the quantitative distribution characteristics in different municipalities. Results show 12 special wards without AGs. Hinohara-mura does not have an AG. An analysis of the dataset revealed that 75.47% of municipalities within the study area had opened AGs, with the most significant number in Nerima ward (46 AGs), followed by Itabashi ward (34 AGs), Edogawa ward (33 AGs), and Setagaya ward (26 AGs). We observed that the geographical locations of the municipalities with a higher number of AGs (the darker colored areas in Figure 5) were at the edge of the 23 wards of central Tokyo.

Following the study of the number of AGs opened in municipalities, we analyze two fundamental indicators of the attributes of AGs: plots and area. We have drawn the corresponding box-and-whisker plot. The horizontal coordinates of the box-and-whisker plot represent municipalities. The vertical coordinates represent the number of plots and area on the two graphs, respectively. Each point represents an AG, and the number of points also reflects the number of AGs in each municipality.

Figure 6 shows the difference between the number of AGs plots. The number of plots varies widely. AG with the lowest number of plots is in Kunitachi-shi; AG with the most significant number of plots in Hachioji-Shi is an abnormally high value, reaching 400. The vast majority of AGs have plots below 100, but there are 16 municipalities with abnormally high values. The municipality with the highest fluctuation is Kodaira-shi, followed by Fussa-shi, and Koto-Ku has the highest median value. The situation of the AGs area is shown in Figure 7. Most of the median values are below 4000 m². The AG in Hachioji City, which has the highest number of plots, also has the maximum area: 15,000 m². The most considerable fluctuation in the area is in Nerima-ku, followed by Suginami-ku, Tachikawa-shi, and Akiruno-shi. The fluctuation in the number of plots and area is, on the one hand, since the AGs are developed and built on existing farmland and, on the other hand, the lack of rational planning.

Most of the municipalities have opened AGs. There was a need to perform an overall analysis of the 472 AGs. Therefore, using ArcGIS, we determined the mean center of the vector point dataset, as shown in Figure 8. The mean center of AGs is located just west of the 23 wards of central Tokyo, which can be interpreted in terms of the extent of the overall study area. The mean center is also tilted towards the 23 wards. This tilt is, as noted in the previous section, more conducive to providing services to residents. It is also important to note that the mean center reflects the point group’s central geographic location, which is a reflection of the group effect and is interpreted in a global context.

The standard deviational ellipse in Figure 9 explains the spatial orientation distribution of the 472 AGs. The ellipse encompasses 68% of AGs, with a rotation angle of 86.759°, close to the east-west distribution. From the length of the major and minor axes compared with the extent of the study area, AGs cover a wide geographical area, and as previously stated, most of the municipalities have opened AGs.

4.3. In the Multi-Distance Zones from the Center of Tokyo

Our municipality-based analysis found that 12 wards were without AGs, and these 12 wards were mainly concentrated in the central location of the 23 wards. So next, we explored the geographical relationship between the spatial distribution of AGs and the urban center.

Sakai et al. (2015) [89] referred to the location in front of the Tokyo railway station as the center of the Tokyo Metropolitan Area and pointed out that this location is the most expensive land in Japan and the central point of the ring and radial road networks. The center of Tokyo was also used in our study.

Viewing the center of Tokyo as the center of a circle, multiple concentric circles were drawn to form multiple distance bands of a 5-km interval, as shown in Figure 10. There were no AGs within 5 km from the center of Tokyo. At a distance of 5–10 km, AGs began to appear. From 10 km, the number of AGs increased gradually, and when the distance exceeded 30 km, the AG number decreased gradually.

We counted the total number of AGs, total number of plots, and total area of each distance zone to analyze this change, as shown in Figure 11. The number of AGs in the 10–20 km distance zone was significantly higher for all three indicators. There was a slight drop in the number of AGs in the 20–25 km distance zone; however, this is related to this area’s small size. After a slight increase in values at 30–35 km, the values gradually decreased, with only two AGs in the area greater than 50 km from the center of Tokyo.

4.4. Spatial Clustering

We used kernel density estimation to detect AG clusters, and the results are shown in Figure 12. The natural breaks (Jenks) method is used to classify the densities into nine grades, with each class given a different color. “K1” to “K8” represent AG clusters.

“K1” is the highest-density cluster, located at the northwest corner of Itabashi-ku. Almost all the AGs in Itabashi-ku are concentrated in this cluster at the municipality’s edge.The density of “K2” is second only to “K1”. “K2” is located at the junction of Kunitachi-shi, Hino-shi, and Fuchu-shi.

“K3”–“K8” are the third level of AG clusters. “K3” is located in Nerima-ku, which is closely associated with urban agriculture in Tokyo. The number of AGs in Nerima-ku is more than that of Itbashi-ku, but the density is less than that of Itbashi-ku, mainly because the spatial distribution of AGs in Nerima-ku is more dispersed. “K3” is located on the west side of Nerima-ku. “K4” and “K5” are located in Katsushika-ku and Edogawa-ku, and both are at the eastern boundary of the municipality. “K6” and “K7” are located on the south side of Tokyo. “K6” is located at the junction of Setagaya-ku, Chofu-shi, and Komae-shi. “K7” is on the west side of “K6”, also located at the intersection of three municipalities, Chofu-shi, Fuchu-shi, and Inagi-shi. “K8” is on the northwest side of Tokyo, at the junction of Ome-shi and Hamura-shi.

From the Tokyo-wide view, the geographical locations of the clusters are mainly concentrated in the geographic midsection of Tokyo. After analyzing each clustering point individually, we found that they all appear at the edges of municipalities. This implies that the municipalities where clusters appear tend to open their AGs close to the land boundary. Some clusters are at the boundary of one municipality. Others fall at the junction of several municipalities, i.e., they are situated at the boundary of several municipalities.

4.5. Regression Analysis

At the municipal level, either the number of plots per 1000 households or area per 1000 households, the stepwise multiple linear regression analysis showed three identified influencing factors: (1) population density; (2) the average posted residential land prices; (3) the land area ratio of productive green space. Descriptive statistics are shown in

Table 4. Variables and statistical description.

	Variables	Definition	Mean	Median	Std.	Min.	Max.
Dependent Variables	PPH	Number of plots per 1000 households	5.8094	4.7900	5.41795	0	20.7
	APH	Area per 1000 households	73.9933	61.3600	65.22197	0	241.42
Independent Variables	POPD	Population density (1000 persons per km2)	11.7986	11.9900	5.57808	1.1	22.25
	RLP	Average posted land prices (residential land, JPY 10,000)	48.412	33.100	47. 3553	9.5	268.9
	PGLR	The ratio of the area of productive green land to the total area in the municipality (%)	3.0898	1.2900	3.81194	0	17.03

.

The regression analysis (see model_1 in

Table 5. Results of OLS regression.

	Model_1	Model_2
Variables	PPH	APH
POPD	−0.426 ** (−0.439)	−4.888 ** (−0.418)
RLP	−0.034 * (−0.298)	−0.423 ** (−0.307)
PGLR	0.331 * (0.233)	5.125 ** (0.300)
Constant	11.465 **	136.326 **
Observations	49	49
R-squared	0.555	0.614

POPD\RLP\PGLR marked with unstandardized Beta (standardized Beta). *: p < 0.05 **: p < 0.01.

) indicates that significant negative relationships between the number of plots per 1000 households and population density (p < 0.01) and average posted land prices (p < 0.05). However, there was a significant positive relationship between plots per 1000 households and the area ratio of productive green land (p < 0.05). Similarly (see model_2 in Table 5), there was a significant negative relationship between area per 1000 households and population density (p < 0.01), and average posted land prices (p < 0.01). There is a positive relationship between area per 1000 households and area ratio of productive green land (p < 0.01).

We eliminated five insignificant potential independent variables: (1) tax levied per household; (2) the percentage of households living in rented houses; (3) the percentage of families living in apartments; (4) the population ratio of farm households; (5) the area ratio of agricultural land. This suggested that the income, housing conditions, and urban agriculture scale are not correlated with the provision of AGs.

5. Discussion

5.1. Differences in the Provision of AGs

The results confirmed that the vast majority of the AGs were located in UPA. We believe that several factors decide the spatial distribution of AGs in different land-use contexts: (1) UPA was designed to be an urbanized area within 10 years, encouraging development related to urban planning and land readjustment; however, the UCA limits development in principle to rural conservation [69]; (2) Farmland in the urbanization promotion area of Tokyo has been restricted from development for conservation purposes, but various methods have been taken with the change of the times [90]. The starting point of the AG system is the institutional framework for the lease of farmland that connects urban residents and farmland owners; (3) According to the outline of AG use issued by MAFF, AGs are places where urban residents continue to carry out agricultural activities, and their target population is urban residents. Therefore, location in urban areas is more beneficial to providing services to the residents; (4) Most residents living in agricultural promotion areas are farmers with farmland who do not need AGs.

Although the original intention is to provide a place for horticultural activities and social welfare for urban residents, the geographical location of AGs in the city, high land price, and high population density make them a luxury, especially within 10 km from the center of Tokyo. We can think of the 10–20 km distance zone as the edge of the 23 wards. As Yoshida (2020) [91] highlighted, proximity to urban centers positively impacts diversified agricultural activities. AGs are also a reflection of the diversity of agricultural activities. For individual municipalities, it is easy to understand that the development of municipalities should be guided by a holistic view while meeting the needs of residents for diversified “farming” activities; it follows that AGs be built close to the boundary. However, there are some drawbacks to this pattern, as some residents in the city need more time and high transportation costs if they are to lease AGs, and the uneven development could affect the sustainability of AGs.

5.2. Factors Affecting Provision at the Municipal Level

We identified three factors that significantly influence the supply of AGs: population density, average posted land value (residential land), and the ratio of productive green spaces. Population density is inversely related to the supply of AGs, because of the lack of previous research literature on AGs, and we compared research findings on urban green spaces. Studies have shown that providing urban green spaces in a compact urban environment is a significant challenge [92,93,94]. The more densely populated cities have less urban green space per capita [95]. Furthermore, high urban land prices have made it more challenging to acquire AGs. Despite growing evidence of the benefits of AGs, the need for housing was seen as more pressing than the need for AGs [96,97], which are under threat in many cities across the world [19]. The ratio of productive green space is significantly positively correlated with the supply of AGs because AGs in Japan are mainly opened on farmland by municipal authorities, Japan Agricultural Co-operatives, farmers, or NPOs. AGs are consistent with the original intent of productive green space to protect urban farmland [68].

5.3. Towards Solutions from Practical Experience

Japan has pursued a highly protective agricultural strategy for decades. Still, the acute scarcity of public land has allowed local authorities to play only a minor role in directly providing land for AGs [63]. We try to find possible solutions from international experience.

In many countries, both at the national and local levels, urban and peri-urban land no longer prioritizes agricultural production [98]. Urban AGs are regarded as land reserves for housing construction and other urban developments [99]. The possible reason for this is that the benefits of AGs are wrongly perceived to be relevant only to the fortunate few who rent plots, and the role of AGs is often underestimated. Many studies have illustrated that AGs contribute positively to ecosystems by mitigating urban climate change, improving air quality, enriching plant diversity, safeguarding food security [100,101,102]. A comprehensive assessment of the sustainability of the implementation of AGs in Murcia, Spain, demonstrates their social, economic, and environmental viability [43]. In Leicester, UK, urban food security could be improved by providing more allotments [103]. Soga et al. (2017) [71] demonstrate the benefits of urban allotment gardening for improving physical and mental health and social integration. Camps-Calvet et al. (2015) [104] identify and characterize ecosystem services provided by AGs in Barcelona. Plans to provide AGs should aim to support and enhance the value of ecosystem services. However, direct land planning will achieve better results than payment for ecosystem services [105]. In Tokyo, local authorities have negotiated with individual landowners to increase the provision of AGs and, in return, have significantly reduced the annual property tax [63], but without a stable funding stream for land maintenance and regulatory services. Building partnerships between governmental and non-governmental actors may be suited to the current situation in Tokyo, just as AGs in Glasgow are supported by grassroots organizations, local authorities, and national funding [106]. It is not easy to acquire land for urban gardening projects in areas with high population density and high land prices.

AGs are officially recognized and embedded in statutory urban plans [107]. The inclusiveness with urban system [45] makes AGs valuable resources for urban regeneration, which can be used in areas that are not yet aligned with the development objectives of urban policy. Ferretti et al. (2015) [108] demonstrate that urban gardens are successful urban regeneration strategies. AGs, as restorative urban spaces and sites for social–ecological experiments [101], have been praised for solving many problems associated with abandoned land [109,110,111]. Many urban communities have attempted to use vacant land for functional greening [112]. Switzerland’s current national policy is to install tiny gardens in urban gaps [35], with fragmented patchwork gardens filling in the gaps between high-density buildings and increasing the provision of gardens. AGs can also come from projects that were never realized or only partially realized; for example, one municipal garden in the València Metropolitan Area was launched after an urban development project failed to build skyscrapers [46]. Green roofs have also been proposed to green compact urban environments [113], which can fulfill function as food production [94] with hydroponics, aeroponics or air-dynaponics systems, or container gardens techniques [114]. In London, container-grown rooftop food gardens were designed for sustainable agriculture [115]. In Barcelona, allotment roofs were explicitly intended for rooftop agriculture [116]. In any case, it is fundamental to assess the land use compatibility, and planners should avoid conflicts between functions to get the best desired urban space. Soil decontamination techniques and soil-less cultivation methods alleviate AGs’ food production constraints from urban soil-related factors. In addition, the interventions of AGs are reversible, as small-scale urban food production can proceed without soil degradation and thus preserve regulating ecosystem services [117]. Our study found that some wards in central Tokyo also create “farming” spaces for citizens. One AG run by the municipal authority was opened in a children’s park [118], and private enterprises developed AGs on rooftops [119]. However, these practical projects are scarce in Tokyo and require flexible urban regulations that empower local governments to identify suitable sites and propose land use plans.

Furthermore, the number of individual plots in AGs is limited, and the space designers have experimented with sustainable planning strategies to provide AGs for more urban households access to gardening. Decreasing the size of individual plots could meet the rapid growth in demand for AGs [52]. The setting of share plots allows more households to develop horticultural cooperation, avoiding the temporary hegemony of individual plots [99] and reducing the pressure to acquire land [120]. The same planning strategy may have different impacts in different contexts. A recent study shows that more than half of people prefer to rent an individual plot with their family rather than a shared plot where they work with other gardeners and share the harvest [14]. The experience from Paris [121] can be drawn upon before applying these solutions, with a questionnaire to understand the knowledge and perceptions of urban residents to make scientific judgments.

5.4. Study Limitations

The empirical results reported herein should be considered in light of some limitations. First, the increased interest in AGs over the past few years has led to the emergence of private AGs; however, data of these private AGs not available; our data only cover AGs based on open source government data in Tokyo. Second, we conducted OLS regression analysis based on the cross-sectional data of 49 municipalities, and the sample size is restricted, which may lead to the un-robustness of the regression coefficient. Finally, Tokyo is not entirely representative, and there is no doubt that relevant debates in more cities will be needed in the future. Despite these limitations, the official lists of AGs are still the most widespread and accessible, and the observation is the smallest geographical area for which socioeconomic data is available in Tokyo. Moreover, clarifying the provision of AGs in one of the world’s largest cities, Tokyo, provides important implications for considering future policies and planning for AGs in large cities.

6. Conclusions

Based on the GIS methods, we revealed the geospatial characteristics of AGs’ provision in Tokyo, in particular, the mappings of four dimensions: in the context of land use, at the municipal level, in the multi-distance zones from the center of Tokyo, and spatial clustering, while extending our geographical understanding of which areas of residents can not benefit from AGs. In the context of increasing urbanization, revealing the distribution of available resources is essential for policymakers and planners, as it increases the suitability and rationality of subsequent planning and contributes to the livability and sustainability of cities.

In this study, OLS models are employed to address a fundamental yet crucial question of what socioeconomic drivers for AGs’ provision at the municipal level. The higher the population density of municipalities, the more challenging it is to access the AG space. The high residential land prices have had a significant negative impact on the AGs’ provision. On the one hand, the acute scarcity of land under public ownership and the limited availability of land have increased land-use competition. On the other hand, the provision of allotments has been reduced by the potentially huge profits from land transference. However, Japan’s productive green spaces provide a degree of protection for AGs land. Our results can inform policymakers that the interests of urban farmland owners need to be safeguarded in the pursuit of sustainable development to benefit urban dwellers and enhance social welfare.

Last but not least, we believe that additional research is needed to identify the ecosystem services provided by Tokyo AGs and explicitly characterize these services in the related planning and strategy documents. Future research could focus on developing flexible and tolerant policies in the context of urban regeneration, tracking the stock of vacant and underutilized land, and exploiting its potential to increase the provision of urban AGs. Beyond this, the concern for land should be extended to people. Allotment gardeners need to make more neighbors profit from allotments rather than be complacent about occupying private plots. Households with urban gardening intentions should accurately convey their demands rather than being deterred by difficult access. Therefore, an in-depth discussion with urban gardeners and non-gardeners is needed.

To the best of our knowledge, this is the first study related to the provision of AGs in Tokyo. European and American researchers dominate current publications on urban gardening, and the case from a non-English country promises to contribute to the ongoing debate on the provision of AGs on a global scale.