Pro-environmental behavior (PEB) is a multidimensional construct delineating actions that benefit the natural environment, enhance environmental quality, or minimize harm to the environment (Steg and Vlek, 2009). For instance, Larson et al.’s (2015) study on rural residents of New York, United States, identified various PEB domains such as conservation lifestyle, environmental citizenship, and land stewardship. Furthermore, in the study of Taiwanese island tourists’ intention to engage in environmentally responsible behaviors, Cheng and Wu (2015) considered an environmentally responsible behavior from two dimensions: general behavior that occurs outside the recreational site (off-site) such as reading books about environmental issues related to a particular area, and specific behavior is actions happening inside the site (on-site) such as participating in activities to clean the area. Similarly, Lange and Dewitte (2019) measured PEB in two similar categories, domain-general and domain-specific behaviors, and Larson et al. (2018b) conceptualized PEB in two dimensions, low-effort (mostly on-site) and high-effort (mostly off-site) behavior. In each case, researchers distinguished between behaviors that occur in a park and might have a direct impact on park resources and management priorities, and behaviors that occurred outside of a park that might have a broader impact on global conservation efforts. In our study, we mirrored this approach and operationalized PEB as a two-dimensional construct (on-site and off-site), to examine visitors’ intent to participate in PEBs within the boundaries of the park itself and out in their broader lives.

Environmental knowledge is a cognitive component of environmental attitudes which refers to an individual’s level of understanding about the environment and human impacts on the environment (Ramsey and Rickson, 1976; Simmons and Widmar, 1990). Several researchers have noted that environmental knowledge is a critical precursor to PEB. For example, Ramdas and Mohamed (2014) discovered that people who possessed elevated environmental knowledge were more likely to purchase green products. Similarly, Flamm (2009) noted a positive relationship between consumers’ environmental knowledge and PEBs. In addition, Lawhon et al. (2017) documented that self-reported environmental knowledge influenced the behavioral intention of engaging in leave-no-trace actions in three U.S. state parks. In their study of Chinese residents, Liu et al. (2020) noted an association between environmental knowledge and PEB, but this relationship was indirect and mediated by a variety of other factors.

Environmental sensitivity is an affective aspect of environmental attitudes that refers to “a set of affective attributes which result in an individual viewing the environment from an empathetic perspective” (Perterson, 1982, p. 5). Although environmental sensitivity is a fundamental element of environmental citizenship and PEB that has been emphasized in the environmental education literature (Hungerford and Volk, 1990; Chawla, 1998), the concept has received less attention. However, in a nature-based tourism context, Cheng and Wu (2015) found that the strongest predictors of PEB were environmental sensitivity and place attachment. Bustam et al.’s (2003) also studied the influence of environmental sensitivity on recreation site preference. Other studies have broadly explored links between related concepts, such as environmental concern and outdoor recreation participation, often yielding mixed results (Berns and Simpson, 2009; Ghazvini et al., 2020). Ultimately, although few researchers have focused explicitly on environmental knowledge and sensitivity and precursors to PEB, existing research suggests that both concepts could significantly influence the PEB of park visitors and other outdoor recreationists.

Place attachment might have a significant impact on PEB within (on-site) and outside of (off-site) the park area (Vaske and Kobrin, 2001; Halpenny, 2010). Place attachment refers to the emotional bond between humans and environment (Low and Altman, 1992). Existing literature supports the idea that place attachment is a multidimensional construct (Kyle et al., 2005; Tumanan and Lansangan, 2012). Two of the most commonly-described and widely recognized dimensions of place attachment are place identity (one’s identity expressed within the physical environment; Proshansky, 1978) and place dependence (the functional utility of a setting in satisfying one’s goal; Stokols and Shumaker, 1981).

Some research has demonstrated links between environmental attitudes and place attachment, especially among outdoor recreation enthusiasts (Brehm et al., 2006; Budruk et al., 2009). Place attachment may be particularly important in nature-based recreation contexts where quality experiences depend on the resource itself (Larson et al., 2018a). Jorgensen and Stedman’s (2006) investigation of predictors of sense of place documented that lakeshore property owners’ perceptions on environmental features (e.g., native vegetation and shoreline development) were the strongest predictors of their place identity, attachment, and dependence. Place attachment can also be associated with recreationists’ perceptions regarding setting conditions (Kyle G. et al., 2004) and the impacts of recreation activities on the environment (White et al., 2008; Eder and Arnberger, 2012). Thus, place attachment can influence the way recreationists think about a natural resource and impacts to that resource, potentially influencing their intent to engage in behaviors that protect the resource.

Many researchers have studied the relationship between place attachment and PEB (Stedman, 2002; Masterson et al., 2017; Larson et al., 2018a). Høyem (2020) found that developing a meaningful connection with natural environment was crucial in terms of increasing PEBs. Similarly, Clayton (2003) found that people who strongly identified with the natural environment demonstrated higher engagement in environmentally sustainable actions than individuals who displayed a weaker identification. Kaiser and Wilson (2000) also showed that attachment to a natural environment significantly predicted general ecological behavior such as ecological garbage removal, water/power conservation, and ecological automobile use when controlling for residential area, length of residence, education, age, and gender. Furthermore, Halpenny (2010) found that place attachment to national parks in Canada positively predicted both place-specific (on-site) and general (off-site) pro-environmental behavioral intentions.

Despite the wealth of research on these topics, few studies have simultaneously considered associations among PEB, cognitive and affective components of environmental attitudes, and multiple dimensions of place attachment in an outdoor recreation setting. To expand this growing body of research, we specifically tested whether the two dimensions of place attachment (i.e., place identity and place dependence) mediated the relationship between the two types of environmental attitudes (i.e., environmental knowledge and environmental sensitivity) and off-site and on-site PEBs in a popular Korean national park. As illustrated in Figure 1, we tested the following hypotheses:

National parks are most visited outdoor recreation sites in South Korea and contain natural environmental of national significance (Oh and Kim, 2013). To date, outdoor recreation research has been mainly conducted in the Western countries, often excluding East Asian context (Walker et al., 2001). Bukhansan Dulle-gil, one of the most visited outdoor recreation sites in South Korea, was selected to address this research gap.

Mt. Bukhan Dullegil (see Figure 2) is a hiking trail along the foothills of Mt. Bukhan National Park in the northern part of Seoul, South Korea. In total the trail spans 71.8 km, comprising the Mt. Bukhan section (45.7 km) opened in September 2010 and the Mt. Dobong section (26.1 km) opened in June 2011. The number of visitors to Mt. Bukhan National Park is approximately 5.5 million per year (Korean Statistical Information Service, 2020). The park is located near the center of Seoul and is easily accessible by public transport, making it one of the most popular and most heavily impacted national parks in the country (Seo and Ko, 2013). The unique natural environment and popularity of the park makes it an ideal research context to understand the relationship between environmental attitude, place attachment, and PEBs.

Six research assistants conducted intercept surveys at four trails on the study site: Uiryeong Trail, Pine Forest Trail, Old castle (Yetseong) Trail, and Devoted Son (Hyoja) Trail. These four locations represent the park’s main highlights and attract the most visitors throughout the year (see Table 1). The data were collected during the second weekend of October 2018, one of the peak seasons. Research assistants randomly approached potential respondents at the entrance/exit points of the hiking trails. In each encounter with a visitor, the research assistants first explained the purpose of the survey and requested voluntary participation. A total of 550 copies of the questionnaire were distributed to potential respondents, and 432 copies were received by the assistants (response rate: 78.56%). After excluding 29 surveys that contained incomplete/insincere responses, 403 surveys were used in the final analysis.

The specific questions we used for measuring each construct are reported in Table 2. We measured environmental knowledge using four items from the scale developed by Haron et al. (2005) to assess individuals’ understanding of environmental problems stemming from recreation behavior (e.g., I know that excessive recreational activities will damage the natural environment of the site). We measured environmental sensitivity using four items adapted from Sivek’s (2002) scale. The four items captured general affection toward the natural environment (e.g., I appreciate the natural environment of Mt. Bukhan Dullegil) as well as specific affective responses to the study site (e.g., I care about the natural environment of the site). We measured place attachment with the scale originally developed by Williams and Roggenbuck (1989) and later adopted by outdoor recreation researchers (Kyle G. T. et al., 2004). Four items were used to capture place identity (e.g., I have a special feeling for the trail) and another four items measured place dependence (e.g., It is more important to visit this trail than other similar destinations). We measured PEB with six items used in Smith-Sebasto and D’Costa’s (1995) study of the university students’ PEB. These consisted of four items about off-site PEB that might occur outside park boundaries in people’s broader lives (e.g., I discuss the environmental protection of Mt. Bukhan Dullegil with others) and two items about on-site PEB that might occur at the park itself (e.g., I pick up trash and branches when I see them on the trail). All items were rated on a 5-point Likert-type scale (1 = strongly disagree - 5 = strongly agree). Additionally, the survey included questions about the respondents’ sociodemographic information such as age, gender, education level, and place of residence. Table 2 lists all of the items and their corresponding dimensions.

We tested a hypothesized model (Figure 1) using a two-step approach for latent variable modeling commonly employed in environmental psychology research (Anderson and Gerbing, 1988; Zhang et al., 2018). First, we used confirmatory factor analysis (CFA) to validate the theorized factor structure of our hypothesized model for all six constructs (two dimensions each for environmental attitudes, place attachment, and PEB). We assessed internal consistency using Cronbach’s alpha coefficients, aiming for values greater than 0.80 (Hair et al., 2009). Our threshold for acceptable factor loadings was 0.60, exceeding the standard suggested by previous studies (Bagozzi and Yi, 1988; Brown and Moore, 2012). We also evaluated the composite reliability (CR) of all dimensions and the average variance extracted (AVE), aiming for values above 0.80 and 0.50, respectively (Fornell and Larcker, 1981; Sahoo, 2019).

Variables in the hypothesized model were assessed for normality by examining skewness and kurtosis. Especially in studies with samples greater than 200 cases, variables with statistically significant skewness or kurtosis typically do not deviate enough from normality to influence the analysis (Tabachnick and Fidell, 2000; White, 2008). Thus, if the sample is large enough, it is better to check the shape of the distribution, instead of using formal inference tests. Considering that the “standard errors for skewness and kurtosis decrease with large N, the null hypothesis will likely be rejected with significant cases when there are only minor deviations from normality” (Tabachnick and Fidell, 2000, p. 74). To examine the shape of the variable distributions, we checked the frequency histograms with superimposed normal distribution and expected standard probability plots. By assessing the plots, we concluded that all variables were considered normally distributed and maintained for examination using CFA.

We used structural equation modeling (SEM) to test the hypothesized structural relationships (H1 and H2) by examining regression coefficients among the constructs of interest for the hikers surveyed on Mt. Bukhan Dullegil. In addition, to test the mediating effect of the two dimensions of place attachment (i.e., place identity and place dependence) on the hypothesized relationship, we used the bootstrap method (Liu et al., 2020) to measure mediating effects (H3). The hypothesized model was assessed using the following goodness-of-fit indices: root mean square error of approximation (RMSEA under 0.10; MacCallum et al., 1996, Savalei, 2018), comparative fit indices (CFI greater than 0.95; Hu and Bentler, 1998, Savalei, 2018), the goodness of fit index (GFI greater than 0.90; Joreskog and Sorbom, 1986, Sahoo, 2019), and non-normed fit indices (NNFI greater than 0.90; Hu and Bentler, 1998, Hoe, 2008). All analyses were conducted using LISREL 8.7 and Mplus 7.0.

Our sample included a relatively diverse sample of outdoor recreationists. As shown in Table 3, about half of the 403 respondents were female (51.3%). The respondents’ mean age was 47.3 years, ranging from 17.0 to 84.0 (SD = 15.6). Their most frequent categories of monthly income were “below $1,800” (32.0%) and “$1,800 to $2,799” (22.1%). About half of the respondents were college graduates (53.7%). Most of them had visited the trails before (70.3%), while 29.7% were first-time visitors.

The measurement model was tested in our first step of analysis. The modification indices from the initial analysis suggested that fit would be improved by covarying error terms for item 2 with 1 and item 9 with 3. After reviewing the original scale for these items, we affirmed the decision to modify the model slightly based on similarities between items. For example, item 2 read “I enjoy natural environments” and item 1 read “I am interested in ecological preservation in Mt. Bukhan N. P.” Both items belonged to the same dimension (environmental sensitivity) and captured the affective aspect of an environmental attitude. Also, item 9 read “Visiting Mt. Bukhan N. P. has a deep meaning for me” and item 3 read “I appreciate the natural environment of Mt. Bukhan N. P.” Both items measured the perceived meaning of Mt. Bukhan N. P. and captured the value of visiting Mt. Bukhan. After allowing covariations between the error terms of those items, the final measurement model showed an acceptable fit (χ² = 1092.168, df = 204, RMSEA = 0.060, NNFI = 0.953, CFI = 0.958; GFI = 0.894) (see Table 2). All other measurement model metrics were also satisfactory. Cronbach’s alpha coefficients for subscales were all greater than 0.80. Factor loadings ranged from 0.690 to 0.924. The composite reliability (CR) of all dimensions ranged from 0.840 to 0.935, showing good reliability across the constructs, and the AVE extracted was above 0.50 (Table 2).

After establishing a valid measurement model, we tested the structural model (see Figure 1). We hypothesized that the two environmental attitude variables (i.e., environmental knowledge and sensitivity) would positively predict both dimensions of place attachment (i.e., place identity and place dependence). We hypothesized that these place attachment dimensions would, in turn, positively influence two types of environmental behaviors (i.e., off-site and on-site PEBs). The final structural model indicated a satisfactory model fit (χ² = 547.403, df = 195, RMSEA = 0.064, NNFI = 0.980, CFI = 0.983; GFI = 0.914).

Table 4 and Figure 3 show the results of the structural equation model (SEM), including links between environmental attitudes, place attachment, and PEB. First, environmental knowledge positively influenced place identity (H1-1; β = 0.166, t = 2.627, p < 0.01) and place dependence (H1-2; β = 0.236, t = 3.877, p < 0.001). Second, environmental sensitivity positively influenced both place identity (H1-3; β = 0.376, t = 5.845, p < 0.001) and place dependence (H1-4; β = 0.387, t = 6.306, p < 0.001). Environmental knowledge and environmental sensitivity accounted for 23.7% of the variance of place identity and 30.4% of the variance of place dependence. Thus, all components of H1 were supported by the analysis. Third, place identity positively influenced off-site PEB (H2-1; β = 0.216, t = 2.059, p < 0.05) but did not influence on-site PEB (H2-2; β = 0.135, t = 1.143, p > 0.05). Lastly, place dependence positively affected both off-site PEB (H2-3; β = 0.571, t = 5.518, p < 0.001) and on-site PEB (H2-4; β = 0.598, t = 4.786, p < 0.001). Therefore, H2 was partially supported by the analysis. Place attachment dimensions accounted for 59.4% of the variance in off-site PEB and 52.0% of the variance in on-site PEB, demonstrating high levels of predictive power.

We further analyzed the indirect effects to examine whether place attachment was a significant mediator of the relationship between the two types of environmental attitudes and off-site and on-site PEB. Specifically, bootstrapping results provided support for the causal relationships among the variables (see Table 5). None of the confidence intervals for the indirect effects included zero, indicating that there were significant indirect effects for all six paths in the hypothesized model. These analyses empirically demonstrated that environmental knowledge had a positive indirect effect on off-site PEB through place identity (path 1: indirect effect = 0.083, 95% CI = [0.032, 0.163], p < 0.01) and place dependence (path 2: indirect effect =0.209, 95% CI = [0.159, 0.330], p < 0.001). In addition, the indirect effects of environmental knowledge on on-site PEB were positively mediated by place dependence (path 3: indirect effect = 0.238, 95% CI = [0.163 0.325], p < 0.001). Environmental sensitivity also had a positive indirect effect on off-site PEB via place identity (path 4: indirect effect = 0.037, 95% CI = [0.002, 0.076], p < 0.05) and place dependence (path 5: indirect effect =0.116, 95% CI = [0.049, 0.194], p < 0.01). Lastly, the indirect effects of environmental sensitivity on on-site PEB were positively mediated by place dependence (path 6: indirect effect =0.131, 95% CI = [0.052 0.190], p < 0.001). Therefore, H3 was supported by the analysis.

This study has several limitations. First, this study collected the data from one national park in South Korea that is located in the vicinity of the Seoul metropolitan area. Research suggests that place attachment varies across the recreation opportunity spectrum (Wynveen et al., 2020), and different recreation contexts would likely yield different levels of visitor attitudes and attachment. Additionally, since we operationalized environmental attitude as a two-dimensional construct, investigating other components of environmental attitudes in more diverse outdoor recreation contexts could be helpful to estimate the generalizability of current study findings. Second, to deepen our understanding in the predictors of PEBs, other factors not incorporated in the current study, such as social norms (Farrow et al., 2017), perceived behavioral control (Alshurideh et al., 2019), and cultural and symbolic interpretation/understanding of nature (Cordano et al., 2010), could be included in future models exploring the relationship between pro-environmental attitudes and behaviors. Third, future studies that intend to examine the role of place to predict PEBs could differentiate place meanings (see Brehm et al., 2013; Larson et al., 2018a) and place attachment (Raymond et al., 2010), considering both the level of attachment that exists and the diverse mechanism (environmental, social, etc.) through which those attachments develop. Fourth, specific questions about other types of behaviors that might occur within either the on-site context, such as Leave No Trace practices (Lawhon et al., 2017), or off-site practices, such as social environmentalism (Larson et al., 2015), could shed more light on PEB engagement patterns.

This study has several implications for nature-based recreation and environmental research and practice. First, our finding that environmental knowledge positively predicted place attachment underscores the importance of education and interpretation in park settings. By providing information about the natural environment through various mediums (websites, flyers, guided tours, etc.) that increase visitors’ knowledge and awareness of key resources and recreation impacts, managers can also increase the affective connection to a recreation site (Hwang et al., 2005; Lawhon et al., 2017). Second, managers should also consider visitors’ emotional and affective responses to the environmental conditions at places such as Mt. Bukhan National Park because environmental sensitivity is associated with both place identity and place dependence. As other studies have shown, efforts to build connection to nature and create emotionally meaningful experiences for visitors to nature-based tourism destinations are likely to foster PEB (Folmer et al., 2013; Cajiao et al., 2022). Third, park managers hoping to achieve stewardship goals should seek ways to cultivate and strengthen visitors’ attachment to place. As place attachment can be fostered through additional time spent in a recreational setting (Smaldone, 2007), it is important to encourage return trips from first-time visitors, providing more opportunities for attachment to develop (Ramkissoon et al., 2018). Moreover, managers might pursue innovative ways of attracting new visitors to parks, such as outdoor educational programs or unique initiatives that draw in new users (e.g., the First Day Hike program on New Year’s Day in U.S. state parks, Wilcer et al., 2018). Collectively, these approaches can be used to facilitate positive visitor experiences while achieving sustainable management goals at popular destinations such as Mt. Bukhan, South Korea, and other heavily visited parks and protected areas worldwide.