Place attachment originates from attachment theory (Bowlby 1969) and is drawing substantial attention from tourism researchers (McLeod & Busser 2012; Ramkissoon et al. 2013b) who utilise it to discover recreationists’ or tourists’ attachment behaviour and feelings (Hwang et al. 2005). Williams et al. (1992) define place attachment as the emotional bond that is formed between an individual and a specific setting. Jorgensen and Stedman (2001) go further by explaining place attachment as not only an emotional bond, but also a cognitive and functional bond with a location. In recreation and leisure, Hidalgo and Hernandez (2001) are of the opinion that place attachment is personified in the feelings and emotions linked to a recreational setting. Some authors claim that when tourists become attached to tourism destinations, they display affective identification and dependence (Schultz 2000) and grow an inseparable connection with the location (Kals, Schumacher & Montada 1999). Moore and Graefe (1994) further state that while place attachment links people with their natural environment, it also induces identification, gratification and concern for a distinctive area (Harris, Brown & Werner 1996). Gosling and Williams (2010) have found that when people grow attached to a specific location (place), they demonstrate care and concern for the protection of the environment, and become more aware of current matters affecting the environment (Lee 2011). This then increases their commitment to the growth and conservation of natural resources (Scannell & Gifford 2010), while at the same time leading them to exhibit environmentally responsible behaviours, such as willingly picking up litter (Halpenny 2006), recycling as well as conserving water (Vaske & Kobrin 2001) and preventing environmental damage (Stedman 2002).

Cheng, Wu and Huang (2013) assert that most leisure tourism researchers assess place attachment with two constructs: place dependence, which is linked to the usefulness of a location for a leisure pastime, and place identification, which is a symbolic or emotional connection to a location (Kyle Absher & Graefe 2003). Ramkissoon et al. (2013a) conversely see place attachment as a multidimensional construct including place affect (Kals & Maes 2002), place dependence (Stokols & Shumacker 1981), place social bonding (Scannell & Gifford 2010) and place identity (Prohansky 1978), with each construct significantly different from the other (Kyle et al. 2005; Ramkissoon et al. 2012). In our study, we also view place attachment as a multidimensional construct as Ramkissoon et al. (2013a) and Devine-Wright and Clayton (2010) found that construing place attachment, as a singular concept, is rather deceptive. They emphasised the need for future research to see place attachment as a multidimensional construct, as this would aid in developing research questions that stay true to vital theoretical concepts (Stedman 2002).

These four subconstructs of place attachment – place dependence, place identity, place affect and place social bonding – are defined next.

The Kruger National Park (KNP) is the flagship conservation and tourism product offering within the South African National Parks system. Hausmann et al. (2017) found that visitors gave high ratings (between 4 = perceived and 5 = highly perceived) to the SoP dimensions in three parks, namely, KNP, iSimangaliso Wetland Park and Table Mountain National Park. Even so, research into place attachment in the context of specific camps within this park remains scarce. Some studies have been conducted in terms of SoP, with the results suggesting that although SoP is accounted for in national parks and environmental management, it remains an underdeveloped concept denoting a substantial void in the way that we understand the link between park management and visitor experiences (Ament et al. 2017; Barendse et al. 2016; Hausmann et al. 2017). To measure place attachment to camps, the selected camps had to comply with specific criteria. Firstly, the selected camps should have the capacity to accommodate enough visitors so that comparisons between camps could be drawn. Secondly, camps in the Southern Kruger were preferred because this would allow for the collection of more data (both the capacity and occupancy levels of camps in the southern part of the Kruger are higher than camps in the northern part of the Kruger). Thirdly, we wanted to use contrasting camps to assess if place attachment differed between the camps (the camps chosen should not only be different in name, but also in all other aspects). If similar camps were chosen, we would not have been able to attribute the differences to anything. Fourthly, we wanted to include one main camp and one satellite camp. Lastly, attachment to the camp should not be obvious. For example, given the Lower Sabie’s popularity, one would assume high levels of place attachment, and thus no need to measure it. The camps that met all these criteria were Tamboti and Satara. Satara can be described as an older, more established (in the traditional KNP style) camp with permanent chalets, and ample infrastructure and facilities. On the other hand, Tamboti is a newer (more modern) camp, with rustic, semi-structured eco-tents and limited infrastructure and facilities. Satara is open to day visitors, whereas Tamboti does not allow day visitors. Interestingly, in terms of overnight guests’ overall satisfaction scores with these camps (for the period May to June 2017), we see that Satara’s overall satisfaction score is slightly lower than the average satisfaction score of visitors to all camps in the Marula region of the Kruger. In contrast, the overall satisfaction score of visitors to Tamboti is higher than the average satisfaction score of visitors to all camps in the Marula region (SANParks 2019).

A paper-based questionnaire was distributed to a convenience sample drawn from both day and overnight visitors, domestic and international, to Satara and Tamboti camps in the Kruger National Park. The questionnaires were distributed from 23 to 28 July 2017. The first section of the questionnaire asked some demographic questions, including age, gender, education level, nationality, travelling party, wild card membership and frequency of visits. The next section measured place attachment (adapted from Ramkissoon et al. 2013b). For the place attachment construct, four dimensions were included in this study: place dependence (three items), place identity (three items), place affect (three items) and place social bonding (three items). Each item was measured on a five-point Likert scale (1 = strongly disagree, 5 = strongly agree). Tourists were approached in both the accommodation areas as well as the public areas of the two camps. Fieldworkers were instructed to attempt to vary the age, gender and nationality of respondents. A total of 201 responses were obtained (day visitors collected at Satara = 34; overnight in Satara = 124; overnight in Tamboti = 43).

In order to meet the purpose of the study satisfactorily, diverse techniques for data analysis were used. The descriptive methods contributed in describing the data in terms of age groups, gender representations and levels of education, while inferential methods permitted us to draw certain deductions about the larger population of travellers and their place attachment to two specific camps in the Kruger. Because of modifications necessary to customise Ramkissoon et al.’s (2013b) scale to our context, the place attachment scale was subjected to an exploratory factor analysis (EFA). T-tests and analyses of variance (ANOVAs) were used to determine how groups of visitors differ in terms of their place attachment. Lastly, multivariate analysis of variance was used to explore the effect of the key identified variables (age, camp visited, nationality, gender, wild card membership, level of education, travelling party and number of visits) as well as their possible interaction effects in explaining the variance in the two factors of place attachment. Multivariate analysis of variance (MANOVA) is a technique that is used to test for the difference in two or more vectors of means. The multivariate tests (Pillai’ trace, Wilks’ lambda, Hotelling’s trace and Roy’s largest root) all test the MANOVA null hypothesis, namely, that the mean on the composite variable is the same across groups. The test thus determines the equality of a composite of the means (optimised to yield the maximum possible F-ratio) across groups. The focus of the identification of meaningful effects will be in using Wilks’ lambda in conjunction with the partial eta squared value. Although the Pillai–Bartlett criterion is considered the most robust and powerful test statistic, the Wilks’ lambda is used as it provides an indication of the variance not accounted for by the combined dependent variables with (1 – λ) the variance that is accounted for by the best linear combination of dependent variables, which enables the explorative understanding of key effects.

Table 1 provides the demographic profile of the respondents. From the table, it is clear that almost an equal number of men (48%) and women (52%) completed the questionnaire. Almost two-thirds of respondents indicated that they are wild card members (a loyalty card that entitles members to reduced entrance fees at all national parks in South Africa). The biggest age group responding to the questionnaire were the 31–50 year-olds. Forty-five percent of respondents indicated that they hold a postgraduate degree. Nearly two-thirds of respondents visited the park with family and 60% of respondents have visited the park more than three times. Official SANParks data on the gender of visitors were not available; however, in a large sample survey (n = 4369), conducted by the agency in 2018 on overnight visitors to the KNP, the gender split of respondents was 59% men and 41% women. According to SANParks, however, this is not necessarily an indication of actual visitation, because the online survey that they use to collect it is sent to the email address of the person making the booking, which is dominated by men. Our sample is similar to the sample of Kruger, Viljoen and Saayman (2017). The biggest age group (35%) in their sample was between the ages of 45 and 59 years. Again, according to Kruger et al. (2017), visitors to Kruger generally hold a degree or diploma, and in our sample this was the same. In terms of frequency of visits, SANParks survey data show that 61% of their visitors have been to the Kruger for one to three times in the preceding 3 years, and 23% between four and nine times. This is similar to our sample.

Ethical approval for undertaking the research was obtained from the University of Pretoria, Faculty of Economic and Management Sciences Ethics Committee (protocol no. EMS014/17). The research was also conducted under the consent and approval of Tourism Development and Marketing Division of South African National Parks. All the participants in the study took part voluntarily after they were informed of the objectives of the study and the completion of an informed consent agreement. All participants were entitled to withdraw from the study at any point. The completed questionnaires were also completed anonymously and confidentially.

Visitors were asked to indicate the extent to which they agreed with a number of statements regarding the specific camps at which they were surveyed. Place attachment was measured on a five-point scale (1 = strongly agree and 5 = strongly disagree). Results are shown in Table 2. It is interesting to note that visitors to the two camps did not have a particularly strong place attachment to the specific camps, even though some preferred the particular camps for their specific settings and facilities. Of the four dimensions measuring place attachment, respondents seemed to agree the strongest with the place dependence dimension, followed by place identity and place affect. Place social bonding scored the lowest mean. When compared, day visitors to Satara felt a consistent stronger attachment to the camp than overnight visitors to Satara and Tamboti. A plausible explanation could be that day visitors could include repeat visitors who come to the camp because they are attached to it, for example, people living near the KNP. Only in terms of place dependence did Tamboti visitors score higher agreement means than day and overnight visitors to Satara. The low agreement mean given to place social bonding is also interesting, especially given the fact that 95% of respondents indicated that they visit the park with friends and family. The most probable explanation could perhaps be that social bonding does not adequately explain a visitor’s attachment to Tamboti and Satara.

Because of the modifications necessary to customise Ramkissoon et al.’s (2013b) scale to our context (Ramkissoon et al. 2013b) measured place attachment in the context of national parks, whereas our study measured place attachment in the context of camps in the parks), the place attachment scale was subjected to an exploratory factor analysis (EFA) using maximum likelihood extraction and direct oblimin rotation (a confirmatory factor analysis was conducted first; however, model fit indicated an inadequate fit, and therefore an EFA was conducted to determine the underlying dimensional structure). The purpose of the EFA was to ensure unidimensionality and internal consistency of this construct in the present context. The Kaiser–Meyer–Olkin measure of sampling adequacy (0.908) and the Bartlett’s test of sphericity which was significant (p = 0.000) both indicate that a factor analysis is appropriate. The EFA was conducted using a sample of 201 respondents who were intercepted at different locations in the two camps. No items were eliminated from the scale, with the number of items remaining at 12. These 12 items used to measure the place attachment construct were loaded onto two factors (see Table 2). Based on the items, which showed that respondents would continue their attachment to specific camps, factor 1 was labelled ‘continued attachment’, while factor 2 was labelled ‘interrupted attachment’, because respondents indicated that if their visitation to the camp would stop, they would disappoint or lose contact with their friends and family, and hence interrupt their attachment to the specific camp. These two factors thus become composites of specific variables and include specific items that are a facet of the broader place attachment dimension (Hair et al. 2014). Internal consistency was evaluated using Cronbach’s alpha and both factors showed a measurement greater than 0.9, indicating strong levels of internal consistency (Nunnally 1978). Together, these two factors explain 75.42% of the variance.

Differences in levels of attachment between groups were measured in terms of age, the camp visited, nationality, gender, wild card membership, level of education, travelling party and number of visits. Only those where significant differences were shown in at least one of the factors are given in Table 3. From the table it is clear that South Africans have a higher level of continued attachment than international visitors have, confirming the results of Hausmann et al. (2017), who found that more experienced national tourists have a higher SoP perception. As expected, wild card members showed a stronger attachment than non-wild card members, for both factors of place attachment. Those visitors with only a matric or high school qualification had the highest levels of place attachment, followed by those with a postgraduate degree. Visitors with a matric or high school qualification felt a stronger continued attachment than visitors with a degree, while visitors with a postgraduate qualification showed a stronger continued attachment than visitors with a degree. This contradicts the results of Kyle et al. (2004) and Ednie et al. (2010) who reported no significant differences between levels of education. In terms of number of visits, those who have visited most frequently showed a stronger continued attachment than those who have only visited 1–3 times. At the same time, those who have visited most frequently also had a stronger attachment than those who have visited between 4 and 20 times. Interestingly, day visitors to Satara were more attached to the camp than overnight visitors to Satara. At the same time, overnight visitors to Tamboti showed a stronger continued attachment than overnight visitors to Satara. Furthermore, day visitors to Satara felt a stronger interrupted attachment to the camp than overnight visitors to Tamboti. The results thus support hypothesis 1 in that groups of visitors differ in the levels of attachment that they experience towards specific camps in the Kruger.

Multivariate analysis of variance was used to explore the effect of the key identified variables, namely, age, the camp visited, nationality, gender, wild card membership, level of education, travelling party and number of visits as well as their possible interaction effects in explaining the variance in the two factors of place attachment, namely, continued attachment and interrupted attachment. Multivariate analysis of variance has greater power to identify an effect because it can identify whether groups are different along a combination of variables, whereas ANOVA can identify only if groups are different along a single variable field. The MANOVA measures whether or not the independent grouping variable simultaneously explains a statistically significant amount of variance in the dependent variable (continued attachment and interrupted attachment). Unfortunately, the MANOVA cannot predict which groups are significantly different from each other, it only tells us that at least two groups are different and that the independent variables influence some patterning of response on the dependent variable.

The set of two dependent variables considered were not highly correlated (> 0.7); thus, multi-collinearity was not present and thus were adequate for the purposes of the MANOVA analysis. The analysis revealed the following key statistical significant multivariate main and interaction effects:

From the above it is evident that a number of key variables influence the set of attachment factors, with the wild card membership, camp visited and the interaction between wild card membership, age and gender being the most influential. The fact that wild card membership explains the variance in the two factors of place attachment is not surprising, as one would expect members to have different levels of attachment than non-members. The factor camp visited also influences some patterning of the response on the dependent variable, again providing support for hypothesis 1, that visitors differ in the levels of attachment they feel towards specific camps. Even though alone age and gender do not influence the variance in the two factors, it is interesting that the interaction between wild card membership, age and gender simultaneously explains a statistically significant amount of variance in the two attachment factors. The results from the MANOVA thus support the second hypothesis that certain variables influence the level of place attachment that visitors experience towards specific camps in the Kruger. The fact that wild card membership showed significant differences in place attachment in both the ANOVA and MANOVA results should encourage SANParks to find ways of boosting their membership sales. Yet, visitors will only become loyal to a park (and camp) if they are satisfied with the overall experience, which should provide more motivation for camp and park management to scrutinise the overall satisfaction levels of visitors, and create strategies to improve the scores of items with which visitors are not particularly happy.