Chinese Residents’ Perceived Ecosystem Services and Disservices Impacts Behavioral Intention for Urban Community Garden: An Extension of the Theory of Planned Behavior

Abstract

Urban community gardens (UCGs), greenspace cultivated and managed for vegetables by local communities, provide substantial ecosystem services (ES) and are warmly welcomed by residents. However, they also have many ecosystem disservices (EDS) and are almost always refused by the decision-makers of the government, especially in China. Better understanding the residents’ perceived ES and EDS and the impact on the behavioral intention (BI) toward UCGs is of great value to solve the conflicts between residents and the government concerning UCGs and to develop sustainable UCGs. Following the theory of planned behavior (TPB), we measured perceived ES/EDS, attitudes (ATT), perceived behavioral control (PBC), subjective norm (SN), and BI of 1142 residents in Changsha, China, and investigated their direct and indirect causal relationships using structural equation modeling (SEM). The results showed that: (1) ATT, PBC, and SN significantly and positively impact the BI of UCGs and together explained 54% of the variation of BI. (2) The extended TPB model with additional components of perceived ED/EDS improved the explanatory ability of the model, explaining 65% of the variance of BI. Perceived ES and perceived EDS showed significant direct positive and negative impacts on UCGs, respectively. They also indirectly impacted BI by influencing ATT, PBC, and SN. The findings of this study can extend our understanding of residents’ attitudes, behavior, and driving mechanism toward UCGs, and can help decision makers to design better policies for UCG planning and management.

1. Introduction

According to the American Community Garden Association (ACGA), an urban community garden (UCG) is any piece of land managed and cultivated by local communities to grow vegetables or flowers [1,2]. UCGs are hybrid parts of the city belonging to both the built environment and the green infrastructure, the public and the private, and the planned and the unplanned [3]. They are productive landscapes where people and place, mind and body, social and physical, and past and present intermingle [4]. UCGs offer many ecosystem services (ES) [5] such as the provision of food and medicinal plants [6], local climate regulation [7], biodiversity [8], habitat for species [9], the facilitation of active and healthy lifestyles [10], neighborhood relationships [11], opportunities for relaxation and recreation [12], increased social cohesion [13], and environmental education [14]. However, UCGs are also associated with various ecosystem disservices (EDS) [15] such as sheltering harmful animals and vectors of diseases, contaminating soil, destroying landscape aesthetics, and increasing interpersonal tension [3,16]. The current theoretical research mainly focuses on the direct or indirect social and environmental impacts of urban community gardens [1,5], where the understanding of ES and EDS focuses on assessing its level of delivery benefits and value [2,6]. However, to our knowledge, few studies investigate whether these services and impairments affect personal and social cognitive factors such as perception and attitudes, which affect their behavioral intentions or practices.

In the 19th century, the UCG emerged in the European and American urban areas in response to the lack of fresh food during the Industrial Revolution, and became an effective measure to prevent social unrest during war periods and the Great Depression [17]. The UCG is currently expanding worldwide and is seen as one strategy for responding to the dietary, social, and environmental problems caused by rapid urbanization in developed and developing countries. UCGs have attracted increasing attention from scholars and practitioners [18], with examples of UCGs including the Verge Garden in Australia [19], the Community in Bloom in Singapore [20], the P-Patch program in Seattle [21], and the Empty-Spaces Plan in Barcelona [22]. At the same time, many countries see UCGs as illegal and informal spaces; gardeners grow land in community public spaces without permission and consultation almost around the world [23], such as in Zimbabwe, Kenya, and South Africa [24].

Before 2010, there were no formal UCGs in Chinese cities [25]; residents spontaneously occupied open public space, usually green space, by creating self-claimed vegetable lots in many corners of cities [26]. Like many illegal UCG practices abroad, this spontaneously formed open gardening space [15] constitutes an infringement of land use rights without consultation with stakeholders such as non-gardeners and community officials. These spontaneous practices lead to disputes and conflicts, posing challenges to urban community governance in China [25,26]. In recent years, the Chinese government has increasingly supported the intensive development and ecological utilization of idle urban land. Within this context, the UCG is also seen as an effective measure for many urban regeneration policies, such as “Stock Regeneration” and “City Betterment and Ecological Restoration” [25]. Furthermore, since the UCG is regarded as a beneficial way to meet the well-being and needs of contemporary urban people, many local government agencies began to legally establish and maintain UCGs [27]. As of 2020, the number of legal UCGs has reached more than 300 in Shanghai and 30 in Changsha [28]. Most UCGs in China are initiated and created by residents or community grassroots organizations; in contrast, UCGs in developed countries are more often planned, developed, and managed by governmental or commercial agencies [25]. Therefore, to facilitate the establishment of comprehensive UCG planning, construction, and management policies in China, and to ensure that UCG projects remain vital and relevant, a deep understanding of urban residents’ community gardening behavior is necessary.

To develop more effective community gardening interventions for urban residents, it is important to conduct research based on theoretical models that adequately explain and predict gardening behavioral intention. A primary framework for predicting and explaining behavioral intention is the theory of planned behavior (TPB) [29,30]. The theory emphasizes that the behavioral intention (BI) of individuals is influenced by their attitude (ATT), subjective norms (SN), and perceived behavioral control (PBC) [29]. Due to different research focuses, previous TPB-based research did not examine the interaction of residents’ perceived ecosystem services and disservices associated with the formation of UCG behavioral intention. Therefore, assessing the perceptions of ecosystem services and disservices is pivotal to developing guidelines and policies that improve UCG planning and management [31].

Given the lack of a comprehensive theoretical framework, the present study aimed to create a structural equation modeling (SEM) of the extended TPB model, including residents’ perceived UCG ecosystem services and disservices, as a means of predicting the community gardening behavioral intention of Chinese urban residents. Our main objectives were to (i) investigate the perception of UCGs’ ecosystem services and disservices by urban residents in inland China and (ii) examine the possible causal relationships among residents’ perception of UCG ecosystem services and disservices and individual and social cognitive factors. SEM was applied to examine the hypothetical causal relationships driving the behavioral intention. The analysis in this paper focuses on the impact of perceived UCG ecosystem services/disservices on residents’ individual cognitive factors (attitudes and perceived behavioral control) and social cognitive factors (subjective norms), as well as the significance and magnitude of impact on the gardening behavior intention. Therefore, the study findings contribute to the literature by revealing the mechanism for residents’ perceived ecosystem services/disservices, attitude, self-efficacy, social stress, and UCG behavioral intention.

This study’s first original contribution was novel evidence for the cognition of ecosystem services associated with UCG. The second original contribution was that we adopted an integrated view to jointly consider UCGs’ abilities to provide ecosystem services and disservices. The third original contribution was that by adding the structure of perceived ecosystem services and disservices, we constructed an extended theoretical framework based on classical TPB theory. Our findings could help urban decision-makers and managers to better understand the ways UCGs are perceived and predict the possible impacts of changes in UCG ecosystem services and disservices on urban residents’ gardening participation. Increasing residents’ awareness of benefits that the UCG could provide (perception of ecosystem services) can motivate their continued behavioral intention. Reducing the various obstacles associated with UCGs and thus reducing residents’ perceived ecosystem disservices will help provide useful information for future scientific regulations, planning concepts, construction models, and management policies related to UCG practice, giving community gardens sustained long-term vitality.

The rest of this article is organized as follows. In Section 2, we review the relevant literature and develop hypotheses. Then, the research design and methodologies are described in Section 3. Furthermore, in Section 4, we analyzed the results and examined the influential mechanism for residents’ perceived ES/EDS, individual/social cognition factors, and behavioral intention. Finally, the discussion and conclusions are given in Section 5.

2. Research Background and Hypotheses Development

To develop more effective community gardening interventions for general urban residents, it is important to conduct research based on theoretical models that adequately explain and predict gardening behavioral intention. The theory of planned behavior (TPB) is a socio-psychological model for predicting and explaining individual motivation and behaviors, especially pro-environmental behaviors [29,30]. From the psychological perspective, TPB theory is a useful theoretical framework that can shed light on the complex psychological processes underlying individual behavior, which makes it an important tool to apply while clarifying the role of human volitional behaviors [32,33]. The theory emphasizes that the behavioral intention (BI) of individuals is influenced by their attitude (ATT), subjective norms (SN), and perceived behavioral control (PBC) [29]. Attitude is the first determinant and refers to the degree of a person’s positive/negative [34,35], support/non-support [36], and/or favorable/unfavorable [37] evaluation or appraisal for performing a behavior [38]. Perceived behavioral control (PBC) is the second determinant and refers to the personal perceptions that make it difficult or easy to conduct a certain action and the extent to which performing a particular act is under the individual’s volitional control [39]. Subjective norm as a social predictor is the third determinant of behavioral intention and refers to an individual’s perception of the degree to which social pressure prompts them to perform or avoid a particular behavior and whether important others support or do not support the activity [40]. Current studies indicate that an individual’s intention to participate in a behavior is stronger when they have positive attitudes towards that behavior, a greater sense of self-efficacy, or strong social support [35,41]. Behavioral intention is at the core of TPB and is the key link between psychological perception and behavior. PBC, ATT, and SN all have an impact on BI, but in different ways and to various extents [41]. The stronger an individual’s intention to undertake a behavior, the more likely the behavior will be performed [42,43]. If a person has a favorable appraisal and a better perceived behavior control ability of the UCG, he may also expect his important referent individuals (or groups) to endorse his behavior. The higher a person’s evaluation and confidence in his participation in UCG behavior the more likely he is to perceive less pressure from important representatives. TPB has been successfully applied to many domains of pro-environmental behavior, including sustainable housing purchase intention [44], energy-saving and emission reduction [45], recycling behaviors [46], environmental protection [47], and participatory natural resource management [48].

Due to different priorities, previous researchers have investigated the determinants of a range of pro-environmental behaviors among citizens. Nevertheless, in the context of the UCG, few studies have used the TPB to investigate the relationship between residents’ individual or social psychological factors, and their impact and extent on behavioral intention [39,40]. These lead to the following hypotheses:

Hypothesis 1 (H1).

Attitudes are positively associated with UCG behavioral intention.

Hypothesis 2 (H2).

Subjective norms are positively associated with UCG behavioral intention.

Hypothesis 3 (H3).

Perceived behavioral control is positively associated with UCG.

These lead to the following hypotheses:

Hypothesis 4 (H4).

Attitudes are positively associated with subjective norms.

Hypothesis 5 (H5).

Perceived behavioral control is positively associated with subjective norms.

The TPB is a very parsimonious model that allows researchers to include additional predictors associated with a particular behavior [42]. Recent interpretations of the theory suggest that the original TPB model is extended through new factors if behavioral intention or practice achieve a significant amount of the variance by including these predictors [49]. Researchers have explained that UCG ecosystem services (or disservices) are a series of beneficial (or harmful) consequences of promoting (or hindering) social and ecological sustainability produced through gardening behavior. Therefore, the perception of ecosystem services (or disservices) is rooted in an awareness of beneficial (or harmful) consequences [50,51]. Although it is rare to intervene in perceived ecosystem services as an extension factor in the TPB theory to explain behavioral intention and actual action, some researchers have studied residents’ perceptions of ecosystem services and disservices for urban green spaces as predictors of behavioral intention and suggest that the decision-making process is related to protective behavior [52]. Moreover, in a meta-analysis of psychological–social determinants of pro-environmental behavior, it was found that the awareness of consequences has important and direct effects on individuals, social elements, and intentions in the TPB and indirectly affects behavior [53]. These results suggest that perceived detrimental consequences affect behavioral intention [33,34,54,55] and that a high level of awareness of positive consequences will help foster a more favorable attitude [32]. In addition to the relationship between awareness of consequences and attitude, the fact that individuals recognizing specific behavioral consequences will better understand the impact of the behavior on others and the environment [56] implies that people who are highly aware of consequences have a strong sense of social expectation [57]. Consequently, some studies have provided empirical evidence for the direct impact of awareness of consequences on subjective norms [53,57,58]. Moreover, research has shown that an awareness of consequences precedes attitudes, perceived behavioral control, and subjective norms [58]. These results suggest that an awareness of beneficial (or harmful) consequences may influence behavioral intention and real action through individual and social variables.

However, few studies have included residents’ perception of ecosystem services and disservices in the UCG as additional factors in TPB models and explored the driving effects of perception on residents’ attitudes, self-efficacy, and social stress. These lead to the following hypotheses:

Hypothesis 6 (H6).

People who perceive higher ES will have more positive attitudes toward UCGs.

Hypothesis 7 (H7).

People who perceive higher ES will have more positive SN toward UCGs.

Hypothesis 8 (H8).

People who perceive higher ES will have more positive PBC toward UCGs.

Hypothesis 9 (H9).

People who perceive lower EDS will have more positive attitudes toward UCGs.

Hypothesis 10 (H10).

People who perceive lower EDS will have more positive toward UCGs.

Hypothesis 11 (H11).

People who perceive lower EDS will have more positive PB toward UCGs.

Moreover, ecosystem services and disservices are often formed based on the same set of ecosystem characteristics, ecological functions, or species groups [59]. This can easily make people conscious of the connection between the perceived ecosystem services and disservices. These lead to the following hypothesis:

Hypothesis 12 (H12).

Perceived ecosystem services are negatively correlated with perceived ecosystem disservices.

Recently, increasing attention has been given to the importance of ecosystem services and the application of ecological models to analyze the multi-level effects of pro-environmental behavior [19,39,40]. The concept of ES has been mainstreamed as an interdisciplinary guiding framework in urban sustainability and resilience agendas [60]. The psychological theory of motivational functionalism states that motivation is the process that initiates, guides, and maintains goal-oriented behaviors and involves biological, emotional, social, and cognitive forces [61]. Motivation is the need to drive people in a specific direction to a certain purpose and to conduct a certain behavior. However, as ecosystem services help shed light on the relationship between the ecosystem and human behavior [62], ES can not only motivate the development of personal and social processes that initiate, direct, and sustain human action [38,63], but also stimulate changes in attitudes and behaviors regarding some symbol or object or aspect. Thus, the psychological theory of motivational functionalism defines ecosystem services as motivation [38,62]. The ecosystem service motivations of UCGs include those related to food supply, biodiversity preservation, soil fertility maintenance, air purification, climate regulation, physical health, recreation, interpersonal communication, and cultural education [39,51,61,64,65,66,67]. While fewer existing studies have focused on the impact of residents’ perceived UCG ecosystem services on their gardening behavior, many studies confirm a significant association between resident or gardener motivation and urban gardening intention or behavior, focusing on the causes of motivation and how motivation affects behavioral intention or practice. The literature relevant to motivation-related urban gardening behavioral intention or practice studies is summarized in

Table 1. Motivations related to urban gardening behavioral intention or behavior studies.

Reference	Garden Type	Sample Size	Data Collection	Interviewee Type	Main Finding
[39]	UCG/UAG	180	Questionnaire survey	Gardener	Used TPB theory; the gardener’s behavioral intention to participate is influenced by functional motivation, emotional motivation, and conditional motivation.
[67]	UCG/UAG	300	Questionnaire survey	Resident	Discussed the associations between personal characteristics and perceptions, needs, and motivations of potential future gardeners.
[15]	UCG	300	Questionnaire survey	Resident Gardener	Discussed respondents’ corresponding motivation, the existing barriers and challenges of their behavior, and the behavioral intention to participate in paid gardening.
[65]	UAG	873	Online survey	Resident	Identified the main characteristics and motivation of potential urban gardeners and determined the mechanism of the influence of these features on motivation.
[61]	UAG	141	Questionnaire survey	Resident	Explored the correspondence between the six resident motivations and the eight gardening needs.
[68]	UHG	126	Questionnaire survey	Gardener	Divided the motivation for home gardening into two categories, one involving social interests and function and the other inherent in nature.
[69]	UCG	23	Interview	Gardener	Explored six functional motivations and three conditional motivations that influence community garden behavior.
[51]	UAG/UHG	23	Interview	Gardener	Indicates gardeners with different cultural backgrounds have different motivations for participating in gardening.
[18]	UG	60	Questionnaire survey	Gardener	Gardener’s gardening motivation is divided into two types: a clear preference for gardening as a means of physical and mental health and learning new skills, and another mainly a yield-based motivation.
[64]	UCG/UAG	40	Interview	Gardener	Gardener functional motivation has direct relevance to their gardening practice; the connection between gardeners and garden occurs in gardening behavior.

Note: UG stands for urban gardens, UCG stands for urban community gardens, UAG stands for urban allotment gardens, URG stands for urban roof gardens, and UHG stands for urban home gardens.

.

Furthermore, motivation and barriers always appear simultaneously in many research articles. There is a range of motivations to promote gardening behavior and various barriers to the behavioral development have also been studied. Barriers regarding UCGs are related to finances, external damage, space, water, soil, organizational structure, communication, interpersonal relationships, and participation [70,71,72]. In many articles, ecosystem disservices have been described as harmful consequences of ecological change or as deficient ecosystem services [73,74]. Ecosystem disservices belong mainly to four fields [73]. To begin with, ecosystem structure, processes, and services provided are negatively affected by population and development systems [75]. Then, (socio-) economic structures and processes are negatively affected by ecosystem disservices [76,77]. Furthermore, human health is negatively affected by ecosystem disservices [52]. Finally, people’s quality of life is reduced by ecosystem disservices, which in turn leads to negative emotions such as anxiety and depression [52]. Studies have pointed out that many of the barriers to the occurrence and sustainability of behavior are due to ecosystem disservices [74]. Recent studies have reported that UCGs’ ecosystem disservices obstruct the development of UCGs and negatively impact gardening behavior. Studies have also found that residents’ motivation to grow organic or healthier food, maintain personal health, and improve quality of life [39] were positively associated with their behavioral intention and actual gardening behavior, whereas environmental impacts and health threats were negatively associated with horticultural participation and behavior [68]. Notably, people of different social backgrounds have different motivations or obstructions, but all factors have positive or negative effects on gardening behavior [66]. However, few articles have talked about the perception of UCG ecosystem services and services affecting horticultural behavioral intention or practice. The more residents believe in high ecosystem services, the greater their motivation, and the more forward their behavioral intention. In contrast, the more they think that UCGs produce various ecosystem disservices and hinder the development of gardening behavior, the more cautious and even less intent they have to carry out this behavior [70]. In addition, studies that rely on survey responses while examining the perception of motivation or obstruction have focused on gardeners in Western countries and have rarely centered on urban residents in Asian environments [15]. These lead to the following hypotheses:

Hypothesis 13 (H13).

Perceived ecosystem services are positively associated with UCG behavioral intention.

Hypothesis 14 (H14).

Perceived ecosystem disservices are negatively associated with UCG behavioral intention.

In conclusion, all previous studies have pointed to a relationship between perceived ecosystem services or disservices and gardening behavior. However, a deeper interpretation of the interactions between individual, social, and perceived ecosystems should explore a theoretical framework. Given the absence of an integrated theoretical framework to predict the gardening behavioral intention of Chinese urban residents, the present study adopted an extended TPB model. Perceived ecosystem services and disservices variables were hypothesized as common precursors of attitude, perceived behavioral control, and subjective norms as determinants of gardening behavioral intention among urban residents (Figure 1).

3. Methodology

3.1. Participants and Data Collection

This paper gathered data in Changsha, China. The city is situated in the subtropical monsoon climate and has an urbanization rate of 79.6%, absorbing 2 million permanent migrants (23.8% of the permanent population). Since 2017, the government has launched the UCG project in 30 communities within the city. This survey went through three phases over a period of two years—two pre-surveys and a formal survey stage. First, between January and May 2019, we conducted the first pre-survey. This consisted of semi-structured face-to-face interviews with 84 gardeners living in six adjacent communities in the Furong District of Changsha to identify and characterize several of the most important ecosystem services in the area. Respondents could choose among eighteen ecosystem services available for urban gardens with content derived from previous research [22]. All respondents were audio recorded with their permission and each session lasted between 30 and 75 min. Twelve ecosystem service categories that were easily understood and perceived by gardeners were finalized. Subsequently, we conducted a second pre-survey. Between May and June 2019, we conducted a household survey of residents living in the same community as the previously interviewed gardeners, with 778 residents completing a structured questionnaire. We examined which of the twelve ecosystem services proposed by gardeners are more easily perceived by residents. At the same time, residents’ perceptions of ten ecosystem disservices from UCGs were investigated. Variable preparation for EDS was derived from previous studies [70,73]. We placed the six ecosystem service functions that residents can easily perceive under the four major categories covered in established ES classifications: provisioning, regulating, habitat/supporting, and cultural services [78]; four EDS categories that were easily understood and perceived by residents were finalized.

Through the above two pre-surveys, the focused content of the more easily perceived ES and EDS by the urban residents of the surveyed area was identified. Furthermore, the individual/social cognitive factors and behavioral intention of UCGs were investigated. To obtain more representative data, with the help of the Changsha City Administration, the community managers of the five main urban areas became key informants. Using a “snowball technique,” interviewed community managers were asked to provide connections to new informants. An online survey was conducted in July 2020. In cooperation with Changsha Ranxing Information Technology Co., Ltd., a professional online survey service provider, we designed and collected the questionnaire. The company’s users have covered more than 90% of the universities and research institutes in China; their questionnaire survey is a well-known brand trusted by leading enterprises in various industries. The company’s strict privacy regulations for personal information and incentives to stimulate participation help to reduce potential bias. This survey received 1692 replies with a valid response rate of 67.5%. Statistical analysis was performed on the data of 1142 valid collected questionnaires.

The respondents’ socioeconomic status is shown in

Table 2. Distribution of the socioeconomic status characteristics of the respondents.

Socioeconomic Status	Items	Frequency (n)	Proportion (%)
Gender	Male	465	40.72
	Female	677	59.28
Age	18–35	326	28.55
	36–44	342	29.95
	45–64	265	23.20
	>65	209	18.32
Education	Junior middle school and below	267	23.38
	Senior middle school	289	25.31
	College	449	39.32
	Master’s degree and above	137	12.00
Income (CNY per month)	<1000	195	17.08
	1000–2999	272	23.82
	3000–4999	225	19.70
	5000–6999	207	18.13
	7000–9999	126	11.03
	>10,000	117	10.25

. Respondents consisted of 465 men (40.72%) and 677 women (59.28%). The demographic and social attributes of the respondents were surveyed in the research, including their gender, age, education level, and family monthly income. There were 326 individuals aged between 18 and 35 (28.55%), 342 aged between 36 and 44 (29.95%), 265 aged between 45 and 64 (23.2%), and 209 people aged over 65 (18.32%). A total of 12% of the respondents had a master’s degree and above, and 39.32% had a college diploma. People with senior middle school education accounted for 25.31% of the total population, and 23.38% population had completed junior middle school and below. With respect to income level, 17.08% of the respondents had a monthly income below CNY 1000, 10.25% had a monthly income over CNY 10,000, and 23.82% had a monthly income of CNY 1000–2999.

3.2. Measures

To compile a reliable set of items to measure the constructs and test the proposed model, we conducted a comprehensive literature review.

Table 3. Measurement items for the personal, social, and perceived ES/EDS constructs.

Constructs	Measurement Items
Personal and Social Factors
Behavioral Intention	Do you intend to do the following behavior in the community garden? (BI1) Gardening activities, such as planting, watering, digging, and weeding; (BI2) mental recovery activities, such as meditation and relaxation; (BI3) and recreational communicative activities, such as chatting and gathering.
Attitude	What is your attitude towards UCG as follows? (ATT1) health, education, communication, and other functions; (ATT2) community gardening space and fields; (ATT3) a variety of planting, leisure, and communicative behaviors; and (ATT4) fruit and vegetable products.
Perceived behavioral control	How do you view your behavioral control capabilities associated with UCG? (PBC1) Gardening experience, (PBC2) gardening skills, (PBC3) time available for gardening, and (PBC4) gardening information possession.
Subjective Norms	To what extent are you willing to obey the attitudes and expectations of these people? (SN1) Family members, (SN2) friends, (SN3) neighbors, and (SN4) municipal authorities.
Perceived ecosystem services and disservices
perceived ecosystem services	To what extent do you think UCGs could bring the following benefits? (P-ES1) Vegetable food supply, (P-ES2) physical recreation/leisure, (P-ES3) social cohesion/integration, (P-ES4), maintained soil fertility/purified air/regulated climate, (P-ES5) biodiversity conservation/habitat maintenance, and (P-ES6) maintenance of agricultural or horticultural cultural heritage/education.
perceived ecosystem disservices	To what extent do you think UCGs could bring the following four damages? (P-EDS1) Volatile organic compounds and greenhouse gas emissions pollute the community environment, (P-EDS2) urban infrastructure is damaged by growing plants and microbes, (P-EDS3) vector-borne disease lurking in urban wetlands affecting residents’ health, and (P-EDS4) the messy and dense growth of vegetation or crops makes people irritable.

Note: P-ES = perceptions of ecosystem service; BI = behavioral intention; P-EDS = perceptions of ecosystem disservice; ATT = attitude; PBC = perceived behavioral control; SN = subjective norm.

provides a summary of the items with their corresponding constructs.

A validated TPB questionnaire [49] was translated into Chinese. The questionnaire was reworded so the questions and answers were more in line with our national conditions, making it more easily understood by Chinese people. The revised questionnaire, used to measure individual and social structure, was hypothesized to reflect gardening behavioral intention in UCGs. We used the questionnaire to conduct face-to-face interviews with 84 gardeners and 778 residents.

In the theory of planned behavior, behavioral intention (BI) is the most direct factor affecting behavior, which is the individual volitional intention to perform (or not to perform) a behavior [49]. This paper defines BI as the volitional intention of residents to carry out or not carry out community gardening behavior. Residents’ BI toward UCG was measured with three items: “Do you intend to do the following behavior in the community garden? Gardening activities, such as planting/watering/digging/weeding [79], mental recovery activities, such as meditation and relaxation [48]; and recreational communicative activities, such as chatting and gathering [44].” Participants were asked to rate each item described above on a 5-point scale ranging from 1 (never) to 5 (always). The BI scale internal consistency (Cronbach’s alpha) was 0.913.

Attitude (ATT) is defined as a psychological tendency that is expressed by evaluating a particular entity with some degree of favor or disfavor. This study defines attitude as the evaluative response of residents to the community gardening behavior itself and its target products. Residents’ attitude toward UCG was measured with four items: “What is your attitude towards UCG as follows? Health, education, communication, and other functions [80]; community gardening space and fields [77]; a variety of planting/leisure and communicative behaviors [81]; and fruit/vegetable products [82].” Participants were asked to rate each item described above on a 6-point scale ranging from 1 (strongly dislike) to 6 (strongly like). The ATT scale internal consistency (Cronbach’s alpha) was 0.963.

Perceived behavioral control (PBC) is the degree of ease or difficulty with which an individual thinks he can control and perform a behavior [49]. PBC is influenced by two factors: controlled belief and perceived power [43]. This study defines the perceived level of four important factors: information, skills, experience, and abilities related to the development of community gardening behavior [44]. PBC was measured by confidence under four conditions in response to the question: “How do you view your behavioral control capabilities associated with UCG? Gardening experience [43], gardening skills [46], time available for gardening [51], and gardening information possession [44]”. Participants were asked to rate each item described above on a 6-point scale ranging from 1 (not confident at all) to 6 (very confident). The internal consistency (Cronbach’s alpha) was 0.940.

Subjective norm (SN) is a person’s perception of social pressures that enables them to perform or not perform the behavior [49]. SN is influenced by normative beliefs and submissive motivations. This paper defines SN as the degree to which residents follow the expectations of a “significant other” for community gardening behavior. Subjective norms were measured by asking about the following four “significant others”: To what extent are you willing to obey the attitudes and expectations of these people? Family members, friends, neighbors and municipal authorities [47,54].” Participants were asked to rate each item described above on a 6-point scale ranging from 1 (strongly unwilling) to 6 (strongly willing). The internal consistency (Cronbach’s alpha) was 0.903.

Before measuring the perceived UCG’s ecosystem service, we first evaluated the ecosystem services and disservices for gardeners. Eighteen ES items (food supply, biodiversity, maintenance of soil fertility, local climate regulation, air purification, global climate regulation, pollination, entertainment and leisure, quality of food, nature and spiritual experiences, exercise and physical recreation, relaxation and stress reduction, aesthetic information, place-making, social cohesion and integration, biophilia, learning and education, and maintenance of cultural heritage) were extracted using items from the studies of Marta Camps-Calvet et al. [22]. Six EDS items (volatile organic compounds and greenhouse gas emissions, excessive use of pesticides and chemical fertilizers, urban infrastructure damaged by growing plants and microbes, vector-borne disease lurking in urban wetlands, disgust about animal feces or plant waste, and irritability about the messy and dense growth of vegetation or crops) were also extracted from the studies of Von Dohren [73] and Becker [70]. We visited the gardeners’ UCGs and 178 gardeners were asked to score the above 18 ES items and 6 EDS items on a scale from 1 (strongly disagree) to 6 (strongly agree). The reliability and validity of each item was reported. This step shows us that in selected regions the UCG is relatively significant in the provision of nine ES and four EDS; this result is similar to the previous article [15]. However, several recent research articles have identified differences between the actual delivery of ecosystem services and human perception [83]. Therefore, before the formal survey we surveyed 778 residents, targeting the nine perceived ES and four perceived EDS indicators described above, to examine whether their perceived ES and EDS were significant for measured items.

Finally, in the formal questionnaire, perceived ES was measured by the following six items: “To what extent do you think UCGs could bring the following benefits? Vegetable food supply [84], physical recreation/leisure [85], social cohesion/integration [86], maintained soil fertility/purified air/regulated climate [87], biodiversity conservation/habitat maintenance [88], and maintenance of agricultural or horticultural cultural heritage/education [89]”. Additionally, perceived EDS was measured by the following four items: “To what extent do you think UCGs could bring the following four damages? Volatile organic compounds and greenhouse gas emissions pollute the community environment [90], urban infrastructure is damaged by growing plants and microbes [91], vector-borne disease lurking in urban wetlands affects residents’ health [73], and the messy and dense growth of vegetation or crops makes people irritable [92]”. The internal consistency (Cronbach’s alpha) was as follows: perceived ES scale was 0.928 and perceived EDS was 0.848.

3.3. Statistical Analysis

SEM is a combination of two statistical methods: confirmatory factor analysis and path analysis. Confirmatory factor analysis (CFA), which originated in psychometrics, estimates latent psychological traits [93,94]. Path analysis, on the other hand, had its beginning in biometrics and finds the causal relationship among variables by creating a path diagram [95]. This article follows the two-step approach for the practical structural equation model (SEM) recommended by Anderson and Gerbing [96].

First, the CFA was conducted with IBM SPSS AMOS to test data reliability by calculating the Cronbach’ s alpha of all observed variables [97]. This step tested whether measurement models of residents’ perceptions of ecosystem service/disservice, attitude, subjective norms, perceived behavioral control, and behavioral intention towards the UCG were consistent with our hypothesis. Before data analyses were carried out, we completed data descriptive analysis for socioeconomic status (SES), which was measured as a construct of education and income.

Second, SEM was performed. SEM is a special form of multivariate analysis and was used to examine the hypothetical causality among multiple variables and how their inter-relationships may play a role in determining a particular outcome in this study. Hypothetical causal relationships were illustrated using a path diagram and analyzed for the standardized partial regression coefficients, which can be interpreted as the magnitude of direct causal influence [98]. The maximum likelihood method was used during both steps to estimate parameters. The tests involved the two models: the TPB-only model (M0) and the TPB plus perceived ES/EDS model (M1), which integrated the perceived ecosystem service and disservice constructs. Model fit was assessed using the chi-square test (X²), likelihood ratio (X²/df), comparative fit index (CFI), and root mean square error of approximation (RMSEA) following the best practices in SEM as suggested by Mueller and Hancock [99]. All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) 22.0 for Windows and Analysis of Moment Structures (AMOS) 23.0.

4. Results

4.1. Descriptive Statistics

The descriptive statistics for the measured variables are shown in

Table 4. Standardized estimates and psychometric properties of the parameters of the measurement model.

Measurement Item	Mean (S.D.)	Standard Deviation	Standard Error	Factor loadings	Squared Multiple Correlations	Average Variance Extracted	Composite Reliability
BI	2.881	1.451				0.781	0.914
BI1	3.045	1.42	0.042	0.686	0.712
BI2	2.824	1.463	0.043	0.772	0.757
BI3	2.774	1.469	0.043	0.773	0.699
P-ES	4.048	1.403				0.682	0.928
P-ES1	4.305	1.503	0.044	0.675	0.634
P-ES2	4.566	1.406	0.042	0.714	0.677
P-ES3	3.884	1.366	0.04	0.854	0.719
P-ES4	3.665	1.357	0.04	0.849	0.703
P-ES5	3.848	1.385	0.041	0.798	0.687
P-ES6	4.022	1.401	0.041	0.782	0.714
P-EDS	3.452	1.232				0.598	0.853
P-EDS1	4.147	1.119	0.033	−0.705	0.538
P-EDS2	3.165	1.275	0.038	−0.827	0.59
P-EDS3	3.835	1.311	0.039	−0.798	0.652
P-EDS4	2.659	1.222	0.036	−0.735	0.53
ATT	4.077	1.526				0.869	0.964
ATT1	4.306	1.426	0.042	0.773	0.828
ATT2	3.887	1.614	0.048	0.818	0.844
ATT3	3.963	1.517	0.045	0.806	0.855
ATT4	4.153	1.545	0.046	0.803	0.847
PBC	3.197	1.557				0.806	0.943
PBC1	3.173	1.614	0.048	0.883	0.829
PBC2	3.131	1.598	0.047	0.901	0.861
PBC3	3.071	1.539	0.046	0.883	0.782
PBC4	3.411	1.478	0.044	0.786	0.643
SN	2.921	1.347				0.701	0.903
SN1	2.984	1.362	0.04	0.842	0.683
SN2	2.757	1.307	0.039	0.82	0.694
SN3	2.934	1.329	0.039	0.83	0.685
SN4	3.007	1.388	0.041	0.789	0.583

Note: S.D.: standard deviation; P-ES = perceptions of ecosystem service; BI = behavioral intention; P-EDS = perceptions of ecosystem disservice; ATT = attitude; PBC = perceived behavioral control; SN = subjective norm.

. Overall, participants had a positive attitude (M = 4.077, SD = 1.526), and a moderate level of confidence in perceived behavioral control associated regarding UCG (M = 3.197, SD = 1.557). Furthermore, residents’ behavioral intention to participate in the UCG, which was measured by using a five-point scale, had a moderately high score (M = 2.881, SD = 1.451) In addition, means of 4.048 (SD = 1.403) for the perceived ecosystem services indicated that the participants had a high level of evaluation with the UCG ecosystem services. The mean score of 3.452 (SD = 1.232) for perceived ecosystem disservices indicated that residents favored a certain amount of the possible damages caused by UCGs. However, social support for community gardening from “important others” around them is inadequate (M = 2.921, SD = 1.347).

4.1.1. Behavioral Intention for UCG

Respondents reported a high degree of behavioral intention for UCGs. Residents were asked to choose based on the five-point scale, with higher scores representing stronger behavioral intention. Residents scored the highest enthusiasm for PI1-Gardening activities, such as planting, watering, digging, and weeding (3.05). They also participated in community gardening to undergo spiritual relaxation and recovery; therefore, the score for PI2-Mental recovery activities, such as meditation and relaxation (2.82), was also high. In addition, residents were interested in the social interactions that took place in community gardening, which was reflected in the score for PI3-Recreational communicative activities (e.g., chatting and gathering) (2.77).

4.1.2. Attitude, Perceived Behavioral Control and Subjective Norm

The study investigated the respondents’ attitudes towards four types of the behavioral objects for the UCG, which all received high support. ATT1-Attitude towards health, education, communication, and other functions had the highest score (4.31), followed by ATT4-Attitude towards fruit and vegetable products (4.15) and ATT3-Attitude towards a variety of planting, leisure, and communicative behaviors (3.96). The lowest score was for ATT2-Attitude towards community gardening space and fields (3.89).

Participants had a certain high level of perceived behavioral control for UCG. The highest score was PBC4-Perceived level of gardening information possession (3.41), which may stem from the fact that we live in an informational age and people have easy access to knowledge about urban gardening and agriculture. PBC1-Perceived level of gardening experience (3.17) and PBC2-Percieved level of gardening skills (3.13) had similar scores, which indicates a strong association between these two factors. Respondents with long-term agricultural experience also had higher levels of perceived skills in the areas of agriculture and gardening; however, respondents who feel weaker behavioral control over PBC3 (perceived level of time available for gardening) reported lower scores (3.07), which indicates that they were uncertain about whether they could dedicate the time necessary to maintain an urban community garden or participate in those types of activities.

To a certain extent, residents were willing to follow the expectations of “significant others” or be influenced by pressure from individuals or institutions while making a decision about whether to support or oppose community gardens. Under the current urban management system, respondents had the highest compliance with urban and community managers, hence the high score for SN4-Municipal authorities (3.00). Their level of compliance towards family and neighbors was relatively neutral, so SN1-Family members (2.98) and SN3-Neighbors (2.93) received similar scores. Finally, they showed the lowest compliance toward their friends, which is reflected in the score for SN2-Friends (2.76).

4.1.3. Perceptions of Ecosystem Service and Disservice for UCG

The survey listed six types of common ecosystem services and allowed respondents to score them based on their perceptions of UCGs. Using statistical analysis, the mean value of each term was obtained. Residents had diverse sensitivity to the different types of ES provided by the UCG, which showed high perception in general. Cultural services, such as PES2-Physical recreation and Leisure services, were the most easily perceived by the respondents and had the highest score (4.57) followed by the product offering functions of Community gardening, such as PES1-Vegetable food supply, which were also highly recognized (4.31). Factors of cultural services related to human well-being, such as PES6-Maintenance of agricultural or horticultural cultural heritage/education (4.02) and PES3-Social cohesion/integration (3.88), were also favored by the respondents. Compared to several other ecosystem services, the scores for the two regulatory services related to ecological environment declined dramatically; PES5-Biodiversity conservation/habitat maintenance received a score of 3.85, and PES4-Maintain soil fertility/purified air/regulate climate received a score of 3.67.

Generally, in these negative statements on ecological, economic, and physical health impacts, more respondents agreed or strongly agreed that urban community gardens led to these ecosystem disservices. Respondents believed that UCGs contaminated the community environment due to sewage, toxic or irritating gas, and rubbish; therefore, the P-EDS1-Volatile organic compounds and greenhouse gas emissions pollute the community environment had the most negative impact with the highest score (4.15), followed by P-EDS3-Vector-borne disease lurking in urban wetlands affecting residents health (3.84), which includes plant allergies, animal attacks, and vector-borne diseases hidden in urban wetlands. The residents expressed a neutral attitude towards the factors related to economic damage, such as P-EDS2-Urban infrastructure is damaged by growing plants and microbes (3.17), and they were not sensitive to those damages. Overall, respondents did not believe that urban community gardening would have a significant negative impact on human psychology; thus, the opinion about P-EDS4-The messy and dense growth of vegetation or crops makes people irritable, has the lowest score (2.66).

4.2. Measurement Model

This study first conducted a confirmatory factor analysis of the measurement model. The results showed that the measurement model was significant, and all the other measurement loads on the latent variables were also significant, as shown in Table 4. The collected data and measurement models reflected a good fit because all model fits exceeded the acceptable levels proposed by previous researchers. The fit indices for the TPB model were as follows: CMIN = 813.88, df = 245, CMIN/df = 3.322, GFI = 0.942, CFI = 0.978, AGFI = 0.924, NFI = 0.969, RFI = 0.962, IFI = 0.978, TLI = 0.973 and RMSEA = 0.045. The initial model M0 is a simple TPB model, and model M1 is an extended TPB plus perceived ES/EDS model. The composite reliability values for the constructs all exceeded the 0.80 guideline. Previous studies state that the average variance extracted (AVE) should not be below the recommended 0.50 level, and the AVEs in this study were between 0.64 to 0.97, fulfilling the requirements [100]. Additionally, the CFA detected that the factor composite score was greater than 0.50, which indicates that the convergent validity is good [101]. In this study, the AVE of an individual construct was compared with the squared shared variances between constructs and was used to examine discriminant validity. Additionally, as shown in

Table 5. Discriminant validity: average variance extracted (AVE) and shared variances.

Variance	1	2	3	4	5	6
1 BI	0.884
2 P-ES	0.545	0.827
3 P-EDS	−0.485	−0.411	0.773
4 ATT	0.599	0.640	−0.546	0.932
5 PBC	0.521	0.429	−0.413	0.435	0.898
6 SN	0.565	0.466	−0.322	0.493	0.352	0.837

(1) Off-diagonals are shared variance, and diagonals are AVEs (bolded). (2) All were significant at 0.001 levels; P-ES = perceptions of ecosystem service; BI = behavioral intention; P-EDS = perceptions of ecosystem disservice; ATT = attitude; PBC = perceived behavioral control; SN = subjective norm.

, the AVE of each structure of the factor correlation matrix is greater than the square correlation coefficients between constructs and thus has discriminant validity [100]. Overall, the model performs well in terms of reliability, convergent validity, and discriminant validity. The result showed that the Cronbach’ s alpha coefficient is 0.902. The KMO value is 0.937, and the P value of Bartlett’s test of sphericity is significant [102].

4.3. Possible Causal Relations Influencing Behavioral Intention

4.3.1. The Impact of ATT, PBC and SN

After fitting the measurement model, this study tested the M0 structural model and assessed the fitted data. Furthermore, to explore whether the extended TPB model (M1 model) increases the interpretation strength of the variance of behavioral intention, Figure 2 presents the standardized path coefficients and explanatory variance (R²) for the M0 structural model. The M0 model shows good goodness-of-fit measures (GOFs) to the data in

Table 6. Goodness-of-fit measures (GOFs) for the structural equation models.

Goodness-of-Fit Measures (GOFs)	Adequate Level	Recommended Level	M0	M1
Root mean sq. error of approx. (RMSEA)	<0.1	<0.05	0.062	0.045
Goodness-of-fit index (GFI)	>0.9	>0.95	0.955	0.942
Adj. goodness-of-fit index (AGFI)	>0.8	>0.9	0.932	0.924
Comparative fit index (CFI)	>0.9	>0.95	0.980	0.978
Tucker–Lewis index (TLI)	>0.9	>0.95	0.973	0.973
Normal fit index (NFI)	>0.9	>0.95	0.975	0.969
Incremental fit index (IFI)	>0.9	>0.95	0.980	0.978
Relative fit index (RFI)	>0.9	>0.95	0.967	0.962

. The three variables of TPB had significant positive effects on behavioral intention, indicating that residents had a more positive attitude towards the UCG (β = 0.38, p < 0.001) and greater subjective norms (β = 0.38, p < 0.001); stronger perceived behavioral control (β = 0.30, p < 0.001) was also associated with greater behavioral intention. Respectively, attitudes (β = 0.47, p < 0.001) and perceived behavioral control (β = 0.17, p < 0.001) showed significant correlations for subjective norms. In the M0 model, the TPB variables accounted for 54% of the variance in behavioral intention. Therefore, the results of the M0 model tested the theoretical principles of TPB and supported the study’s first hypothesis.

4.3.2. The Effects of Perceived Ecosystem Services and Disservices

The M1 model is an extended TPB model that includes factors of perceived ecosystem services and disservices, and it shows a good GOF in Table 5. The M1 structure model explained 65.0% of the variance in behavioral intention. By comparing the results of the M1 model to the M0 model, the research found an 11% increase in the explained variances of behavioral intention. Of all the 14 path coefficients, 13 were statistically significant (p < 0.01). Most of the results are the same as the ones that we initially hypothesized, with the perception of ecosystem service having significant and direct positive effects on attitudes (β = 0.56, p < 0.001), perceived behavioral control (β = 0.33, p < 0.001), and subjective norms (β = 0.26, p < 0.001). Additionally, the perception of ecosystem disservice had a significant direct negative effect on attitudes (β = −0.34, p < 0.001) and perceived behavioral control (β = −0.29, p < 0.001), but the direct effect on the subjective norm was not statistically significant. Standardized parameter estimates indicated that perceived ecosystem services and disservices were associated the most strongly with attitudes. Subsequently, attitudes (β = 0.13, p < 0.001), subjective norms (β = 0.31, p < 0.001), and perceived behavioral control (β = 0.20, p < 0.01) significantly predicted behavioral intention. Subjective norms are the most predictive of behavioral intention, as they are influenced by attitudes and perceived behavioral control. As shown in Figure 3, the path diagram has completely standardized parameter estimates. In support of the second hypothesis, the inclusion of the perceived ES and EDS variables based on the M0 model led to an increase in the explained variances of behavioral intention.

5. Discussion

UCG development not only conforms to China’s national concept of ecological construction, promoting the intensive development and ecological utilization of idle urban land, but is also an effective way to meet the needs of urban people, such as allowing close access to nature, improving lifestyle quality, and advancing well-being. This study used SEM to incorporate ecosystem services and disservices into the TPB model to explore the relationship between perceived ecosystem services by ordinary urban residents and community gardening behavioral intentions. It added novel evidence for cognition of the ecosystem services associated with UCG and adopted an integrated view to jointly incorporate UCGs’ abilities to provide ecosystem services and deliver ecosystem disservices into an extended theoretical framework based on TPB theory. Overall, this study provides a better understanding of the relationships between individual, social, and perceived ecosystems by revealing that the perception of ecosystem services and disservices has important effects on predicting the behavioral intention of ordinary residents through individual and social factors. Furthermore, this study provides some support for the effectiveness and utility of the extended theory of planned behavior model in predicting the gardening behavioral intention of Chinese urban residents.

We reveal the mechanism of how residents’ perceptions jointly influence their UCG behavioral intention. The first hypothesis of this study holds that behavioral intention is directly and positively influenced by attitude, perceived behavioral control, and subjective norms, which are positively influenced by attitude and perceived behavioral control. Consistent with the TPB theory, the study concluded that residents’ attitudes towards UCGs have the greatest influence on their behavioral intention. Attitude or emotional motivation is an important determinant of all horticulture behavioral intention [15,19,67]. A study of gardeners’ participation in community (allocation) gardens found that the positive emotional connection they established with the garden strengthened their intention to participate [39].

Second, our study showed that residents’ subjective norms of UCG had almost the same effect as attitudes on behavioral intention; however, subjective norms are largely influenced by attitudes; therefore, attitudes indirectly influence behavioral intention through subjective norms. Residents’ subjective norms of UCGs affect behavioral intention to almost the same degree as attitudes. As subjective norms are largely influenced by attitudes, and attitudes are a greater influence on behavioral intent than subjective norms. This study pointed out that subjective norms are the most important cognitive structure that affects residents’ gardening behavior. Compared with gardeners who have participated in a UCG, residents who were not involved in urban gardening thought that important others or institutions around them, especially municipal authorities, would not approve of the existence of community gardening in the city. Existing research has already proposed that municipal authorities could promote gardening participation by readjusting policies to change people’s beliefs in subjective norms, which is a suggestion that our study confirms [19]. Additionally, this study found that the effect of perceived behavioral control or the self-efficacy of overcoming barriers on behavioral intention was the weakest of the TPB structures, which aligns with previous findings [3,71]. We thus confirmed that in order to consider the effects of attitude, perceived behavioral control, and subjective norms on behavioral intention, future interventions should adopt a multi-level research approach.

We simultaneously considered two perception constructs, i.e., perceived ecosystem services and disservices. We found that the two perceptions are negatively correlated with each other. These perceptions can explain and predict people’s personal and social cognition towards UCG, thus playing a major role in influencing gardening participation. Among them, perceived ecosystem services become a non-negligible factor that can help them for residents to develop continuous gardening intentions by forming more positive attitudes, greater confidence, and less social pressure towards UCG. As we initially predicted, the inclusion of perceived ecosystem services and disservices in the TPB model led to an increase in the explanatory variance of residents’ behavioral intention toward the UCG. In the initial M0 model, the explanatory variance of behavioral intention influenced by individual and social factors of TPB was 54%; however, with the extended M1 model, comparing the initial model led to an increase in the explanatory variance of behavioral intention from 54% to 65%, which is higher than the results from previous studies [39]. This finding provides support for the predictive validity of the extended TPB model. Our study did not incorporate residents’ actual gardening behavior into the TPB model because many studies have shown that behavioral intention, though the closest and most direct prediction factor of behavior, does not necessarily explain real behavioral changes [42]. To narrow the gap between behavioral intention and real action and increase the explanatory variance of actual gardening behavior, future studies could delve deeper into the different roles of perceived ecosystem services and disservices.

The results of this study show that perceived ecosystem services (or disservices) not only have an indirect impact on gardening behavioral intention through the ATT, PBC, and SN factors in the TPB but also have positive (or negative) direct effects on behavioral intention. This suggests that the three variables in the TPB model do not fully mediate the relationship between perceived ecosystem and behavioral intention and that other important individual and social variables may exist. The present study examined other perceptions associated with ecosystem service functions, with individual and social factors, such as the neighborhood attachment [103], perceptions of environmental impact [68], perceived risks [104], life satisfaction [40], and identity perception [105], and found that they all have an impact on gardening behavioral intention or actions. Numerous studies have shown that a strong sense of community attachment linked to communal gardening encourages people to form connections between physical and social environments that promote interaction with others and spark changes in sustainable gardening behavior [61,103,106,107]. Studies have also shown that perceptions of environmental impact from urban gardening ecosystems affect the promotion of more sustainable horticultural behaviors and practices [68,108]. The literature suggests that perception and identifying risk factors for urban gardening behavior that may have detrimental effects on human well-being are important considerations for achieving the various ecosystem service functions brought about by urban gardens [82,109,110]. In some studies on urban gardening promoting better life satisfaction and eating habits, these good perceptions have a positive effect on gardening behavior [40,68]. Some research highlights that urban gardening can reshape urban citizenship through residents’ identity perception with their “environmental gardening identity” [105]. This identity perception, brought about by the social and cultural functions of ecosystem services, offers far-reaching potential for changes in gardening behavior and practice [111,112,113,114]. Based on our findings, we believe that further research should be conducted that examines the effects of various perceptual factors linked to urban gardening on behavioral intention and action.

Although aspects of urban horticultural ecosystem services and disservices have been explored in this study, gaining deeper insight into other service functions, such as community resilience [115], urban sustainability [116], landscaping [117], treatment function [118], and urban health service potential [69], would help shed light on gardening behavioral intention and practices. Community gardens have been proven to play an important role in the social–ecological resilience of urban ecosystems and offer residents a place where they can reduce stress, share experiences, and gain community support [119,120]. Previous research has shown that community gardens provide a model that can promote sustainable urban living by linking individuals and communities to the food system and promoting the development of a deep reconnection and long-term commitment to sustainable living practices [84,85,121,122]. It has been concluded that the aesthetic value of urban gardens helps enhance the attractiveness of urban areas and that a scientific garden distribution is more likely to contribute to a larger network of ecosystem services across a broader urban landscape [123,124,125]. Current research also indicates that site spatial and design characteristics affect people’s gardening participation [126,127]. For nearly a decade, community gardens, especially at the local level, have been increasingly praised for revealing the relationship processes that connect people, ecology, and health [128] [129,130,131,132]. Notably, the latest research indicates that urban gardens have functioned as a potential health resource during the COVID-19 pandemic by giving gardeners a source of freedom and joy while facing difficult and potentially isolating situations [133,134]. On the other hand, while examining ecosystem disservices, researchers have found that vegetables grown in community gardens pose a health threat arising from their potential exposure to pollutants, such as heavy metals in the environment [104,135]. In conclusion, the metrics of six ES and four EDS are relatively common in this paper, but deepening the understanding of the function and damage of their ecosystem services is meaningful to illustrate community gardening behavioral intentions and practices in different cultural and social contexts.

Our findings provide a strategy adopted on the premise of adapting to the intensive development and ecological utilization of idle urban land and meeting the well-being and needs of contemporary urban people. That is: (1) to improve the level of delivery of the UCG’s ecosystem services by developing and promoting new gardening technologies; (2) to construct and publicize a series of high-quality UCG exemplary cases, to spread its various ecosystem service functions to urban residents, and to improve their perception of UCG benefits; (3) to consider and address the various gardening technical problems and social obstacles, to reduce the UCG’s ecosystem disservices; (4) to publicize effective technical measures and methods to reduce the UCG’s ecosystem disservices, address potential participant concerns about possible obstacles and reduce their perception of UCG damage; and (5) to find and cultivate institutions or individuals as important reference groups, such as management agencies good at organizing and operating UCGs and enthusiastic gardeners who have successfully developed UCGs; their demonstration and drive will help improve the subjective norms of residents and thus promote their gardening behavioral intention. The proposed strategy implies that the roadmap for UCG sustainable development can be implemented by early practitioners to later participants, from megacities to middle-and-small cities.

This study had several limitations that should be considered. First, potential confounding factors, such as demographic factors and UCG location (i.e., whether it was on a roof of a public building, on a communal inland surface, or in a vacant area outside of a given community), were not strongly distinguished. Although this study is based on TPB theory as well as previous studies, variables of the SEM model containing more confounding factors would be beneficial to elucidate our understanding of the perceived ecosystem correlates of gardening behavioral intention. Second, the survey of perceived ecosystem services/disservices, ATT, PBC, and SN measures in this study were self-managed by the authors, and although the reliability and validity of the measurements of these items were derived from previous conclusions, there was not a widely tested maturity scale, and this may have led to potential errors. Using objective techniques would help to reduce biases. Third, because research on UCG perceptual ecosystem services and disservices is in its infancy, further measurement tools need to be developed. Identifying the ecosystem supply characteristics of the different types of UCGs associated with resident gardening behavior will help develop more effective policies and incentives. Finally, the structural equation model that this study proposes is based on previous research and the TPB theories of environmental psychology, but it would also be productive to explore other plausible models to help explain community gardening behavioral intention and the practices of urban residents.

6. Conclusions

The results showed that perceived ES and EDS had significant effects on gardening behavioral intention and that factors of attitude, perceived behavioral control, and subjective norms in the TPB model partly mediated this association. These results have important practical implications for both policymakers and managers. For policymakers, the benefits and positive impact of UCG can be gradually advanced by improving its capacity to deliver ecosystem services and reducing the level of disservices to their ecosystems. Therefore, the findings of this study are valuable for developing guidelines and intervention policies for the planning and management of UCGs, which can promote more scientific and sustainable gardening behavioral intention among Chinese urban residents. Furthermore, the findings of this study also have important implications for community managers or local institutions following the legalization of UCGs. Since perceived ecosystem disservices were important for new participants who focused on innocuous practice, UCG managers should pay more attention to reducing these damages by advancing gardening techniques and optimizing management policies. Future studies could distinguish between the correlation between personal characteristics of potential gardeners in residents and their perceived ecosystem services and disservices. It is necessary to further explore the impact of the perception of ecosystem services and disservices in UCG, and of different spatial and governance types, on gardening behavioral intention and behaviors. To conclude, future studies are necessary to develop evidence-based theoretical models of changes in community gardening behavior among urban residents.