Memorable gastronomic experience in sustainable tourism destinations: evidence from Croatia

The aim of this study is to examine the nature of the relationship (direct and indirect) between perceived quality aspects (PQ), memorable gastronomic experiences (MGE) and tourist behavioral intentions (BI) in two sustainable destinations in Croatia - Istra and Međimurje. A sample of 414 respondents was collected through a questionnaire. To examine the relationships between PQ, MGE and BI, the primary data were then subjected to PLS-SEM analysis. The findings of this study indicate that a stronger direct relationship between MGE and BI was found in Istria compared to Međimurje. Moreover, among tourists who visited Istria, MGE proved to be a stronger moderator between local food (LF) and BI of tourists. By following the research results in both sustainable destinations, DMOs can create sustainable gastronomic routes with special attention to key elements of quality that promote the MGE.


Introduction
MGE represents a long-term retrospective subjective gastronomic memory of the consumer, which is simultaneously permeated by cognitive and affective processes (Rašan & Laškarin Ažić, 2023).The ability of tourism providers to offer a memorable experience of local gastronomy goes a long way in helping tourists form strong attachments to the destination they visit.More recently, sustainability has become a central issue in gastronomy tourism research (Leer, 2020), but no research on MGE in sustainable destinations has been conducted so far.However, there are studies that examined the perception of tourists about MGE in general tourist destinations (Antón et al., 2019;Di-Clemente et al., 2019;Badu-Baiden et al., 2022a).In these studies, scholars mainly focused on examining the relationship between concepts PQ, MGE and BI.
The literature is dominated by scholars' attitudes toward the MGE as a mediator between antecedents and consequences (Jeannot et al., 2022).Souki et al. (2022) state that the antecedent of MGE (environmental stimuli) is a PQ experience.The antecedent role of PQ aspects has also been confirmed in many other MGE and BI studies (Muskat et al., 2019;Prayag et al., 2015;Cao et al., 2019;Kovalenko et al., 2023).It is undisputed that scholars perceive behavioural destination loyalty as a consequence of MGE (Antón et al., 2019;Badu-Baiden et al., 2022b;Kovalenko et al., 2023).In this context, BI as a type of destination loyalty represents the common denominator for the WOM and repeat intentions of tourists (Rašan & Laškarin Ažić, 2021).
However, Rasoolimanesh et al. (2021) note that there is a lack of theoretical unanimity in the literature about the nature of the relationship between PQ, MGE, and BI tourists.Some scholars have found that aspects of the PQ experience have a direct relationship (Prayag et al., 2015;Hernández-Rojas & Huete Alcocer, 2021), while others have an indirect relationship to destination and facility loyalty of tourists (Tuncer et al., 2021;Bufquin et al., 2017).Discrepancies were also found when examining the relationship between aspects of PQ and MGE.Most studies found a direct relationship between aspects of PQ and MGE (Lin, 2022).In some studies, MGE was found to be a moderator between aspects of PQ and BI tourists (Horng & Hsu, 2021).
The main objective of the study is to determine the contrast in the nature of the relationship between the PQ, MGE and BI constructs between tourists who have visited two different sustainable destinations.Other objectives are as follows: ⮲ expand existing knowledge of MGE and its implications for future tourist behaviour in sustainable destinations; ⮲ to test whether there are significant differences between two sustainable destinations in hypothetical relationships (PQ, MGE, and BI); and ⮲ to provide directions for future MGE research in sustainable destinations.
Identifying differences between sustainable destinations in creating memorable experiences and their impact on future behavioural intentions adds to the current body of knowledge, as there are no previous studies that specifically address this topic.The results of this study provide both sustainable destination managers and restaurant operators with greater insight into the gastronomic factors that affect tourists' loyalty to the destination where they consumed local food.The results obtained also provide restaurants with a solid basis for improving their gastronomic offer in order to meet tourists' desires and needs in the long term.
principles of sustainable development is more important today than ever before, there are not many destinations worldwide that have a sustainability certificate.
According to the Euromonitor International report (2020), Croatia ranks 13th in the world in terms of sustainable tourism and green transformation standards and indicators.In Croatia, a country where tourism accounts for a quarter of the total GDP, there are several destinations that have received a sustainability certificate.Two sustainable destinations stand out the most in Croatia: • Istria: Providers of tourist products and services, and therefore Istria as a destination, have a large number of certificates for sustainable development, such as Eco Domus, Domus bonus, Travelife Gold Award, SGS certificate (Institute of Agriculture and Tourism, 2020).Istria was awarded and certified for its sustainability and included in the list of 100 Green Destination Stories.
• Međimurje: This region is a three-time winner of the EDEN award (2007, 2015 and 2019) -National European Destination of Excellence.In addition, Međimurje is in the process of repositioning itself as a sustainable destination and won the prestigious "Green Destination" award in 2023 (Tourist Board of Međimurje, ed.).
In Istria County, in 2021 there were recorded 21.734.119overnight stays while the number of overnight stays in Međimurje County in the same year was even 150 times lower (144.061) Croatian Bureau of Statistic ( 2022).The difference in the number of arrivals was obvious, which is confirmed by the fact that there were 50 times more arrivals in Istria (3.372.081)than in Međimurje (64.051).Based on the above statements, there are discrepancies in the structure of tourist arrivals and overnight stays between these two sustainable destinations.The identified differences served to formulate the following hypothesis: H1: There are statistically significant differences in the perception of gastro-tourists between those who visited Istria and those who visited Međimurje.

Conceptualization of PQ and its relationship to MGE and BI
"Perceived quality refers to tourists' perceived performance of products or services in terms of attributes that can meet their needs and expectations compared to competitors" (Souki et al., 2022).Numerous scholars have used a multidimensional approach in examining the quality of gastronomic products and services (Souki et al., 2020;Wu et al., 2021).As a result, PQ of gastronomy is a complex, multidimensional concept.Various PQ aspects are mentioned in the gastronomic literature.However, the most common PQ aspects used in the gastronomic studies are: (a) physical environment (Tsaur & Lo, 2020;Kala, 2020), (b) service quality (Kala, 2020;Kim et al., 2021;Tsaur & Lo, 2020), and (c) food quality (Stone et al., 2018;Kim et al., 2021;Zhang et al., 2019a).
In the destination, facilities that offer gastronomic delights can create PQ conditions that result in a memorable experience for tourists.Fernandes et al. (2015) state that sensory stimuli are key to creating a memorable experience, as tourists experience support from the environment through their senses.When examining tourists' perceptions of the memorability of an experience, Oh et al. (2007) suggest including as many senses as possible.Empirical studies in tourism acknowledge that there is a direct relationship between the PQ aspects and the MGE (Lin, 2022), as well as the PQ aspects and the BI of tourists (Prayag et al., 2015;Davras & Özperçin;2021;Konuk, 2019;Wang et al., 2020).Some scholars pointed out that a MGE is mediator between PQ aspects and BI of gastro tourist (Horng & Hsu, 2021;Badu-Baiden et al., 2022a).Therefore, the following hypotheses were tested in this study: H2: PQ aspects have a direct positive effect on MGE in both sustainable destinations.H3: PQ aspects have a direct positive effect on BI in both sustainable destinations.H4: MGE is a mediator between PQ aspects and BI in both sustainable destinations.

Conceptualization of MGE and relationship with BI
There is no consensus on theorizing and studying memorability of experiences (Bigné et al., 2022;Packer & Ballantyne, 2016;Hosseini et al., 2023;Hosany et al., 2022).According to Rašan et al. (2022), there are large differences in the settings of the memorable experience depending on the context in which the research was conducted.For example, the memorable tourist experience at the destination level has been studied mainly by scholars using the MTE model, while there is no consistency in the research design in the gastronomic literature.Skavronskaya et al. (2020) noted that the memorable experience includes a cognitive and neuropsychological aspect.While some other scholars state that the experience is a subjective assessment of tourists and their affective, cognitive, and behavioral reliving of events from the past (Tung & Ritchie, 2011).For Becker and Jaakkola (2020), the experience represents tourists' responses to individual stimuli.When recalling events, tourists rely on episodic memory, which allows them to record and store information about temporal-spatial relationships as well as the people they were surrounding themselves with at the time (Tulving, 2002).
Two people can be in the same facility, sit at the same table, eat the same food, and receive the same service, but their gastronomic experience will never be the same.This is mainly because the MGE represents consumers' subjective consumer assessments of gastronomic experiences (Cao et al., 2019), which are remembered even after the experience has already taken place (Tsaur & Lo, 2020;Rašan & Laškarin Ažić, 2023).Memory enables tourists to relive the events they experienced (Lee et al., 2021) in order to make BI -the decision to return and spread positive WOM (Sthapit & Coudounaris, 2018).There is no doubt that these memories are a stimulator and the most important predictor of tourists' return to a destination (Wirtz et al., 2003).Therefore, it is not surprising that previous research has shown that the MGE is a construct directly related to tourists' BI (Horng & Hsu, 2021;Di-Clemente et al., 2019;Soonsan & Somkai, 2023).Based on the above, the following hypothesis is defined: H5: The MGE has a direct effect on BI of tourists in both sustainable destinations.
Following previous discussion and the proposed hypotheses, a theoretical model is presented in Figure 1.

Questionnaire design
The measurement instrument is designed based on the relevant scientific literature in the field of gastronomy (Table 1).The items were adapted to the requirements of this survey, in which the questionnaire served as an instrument to examine the attitudes of gastro-tourists.The questionnaire consisted of four parts: perceived quality (PQ), memorable gastronomic experience (MGE), intention of tourist behaviour towards the visited sustainable destination (BI) and profile of respondents.Gatrotourists in the two sustainable destinations were first asked to recall the object where they consumed gastronomic products and to answer questions based on these memories.
The first part of the survey was related to items about the PQ of gastronomic products and services in the destination visited.In the second part of the questionnaire respondents were asked to determine their level of agreement with MGE items.While the third part of the questionnaire determined their loyalty to a sustainable destination.The items in all three parts of the questionnaire were structured and examined using a 5-point Likert scale ranging from 1 = strongly agree to 5 = strongly disagree.The last part of the survey contained questions about the sociodemographic characteristics of the respondents.Although a total of 500 questionnaires were prepared for the field survey, the response rate was only 31.4% (157 questionnaires).When checking the completion rate of each of the collected questionnaires, 19 cases with missing values were found.Since these were cases in which the questionnaires contained more than 15% missing values, these questionnaires were excluded from further data processing in accordance with the recommendation of Hair et al. (2017).Thus, 138 questionnaires collected in the field were included in further analysis.Those respondents who indicated they did not have time to respond on-site were sent an email containing the questionnaire.In this way, the authors received an additional 276 properly completed questionnaires, which indicates that 414 respondents were included in the total sample.

Sampling
The target group of this study was gastro-tourists visiting either Istria or Međimurje.Since there is no list of names of gastro-tourists in Croatia, it was not possible to draw a random sample.For this reason, this study used a convenience sample based solely on the availability of units.As the appropriate sample size can only be determined for convenience samples, the choice of sample size depends on the subjective assessment of the researcher.In terms of ethical dilemmas, it is important to emphasize that at the beginning of each questionnaire there was a statement that completion of the questionnaire was completely voluntary.In other words, participation in the research was voluntary and anonymous, without any incentives.

Data analysis
Data analysis was performed in two phases as recommended by Abeyrathna et al. (2022): (1) descriptive and inferential analysis using the SPSS 23.0 programme and (2) construction of an initial Structural Equation Model (SEM) using SmartPLS software.In addition to determining the demographic characteristics of the subjects, descriptive statistical analysis was performed to determine the central tendency (sample mean: x) and standard deviation (sd: σ) of the construct PQ.Then, exploratory factor analysis (EFA) was performed to determine the dimensionality and reduction of the PQ items.The reliability of the proposed measurement instrument PQ was examined using Cronbach's alpha, and items with a value below 0.3 were excluded from further analysis (Churchill, 1979).According to the research objectives and the characteristics of the data collected (sample size, non-normal data) and the nature of the research (exploratory), partial least squares analysis SEM (PLS-SEM) was selected as the most appropriate analysis for this study.The authors used SmartPLS v.4.0.8.4 to analyse the model and its relationships.

Results of empirical research 4.1. Descriptive and inferential statistics in SPSS
The authors first analysed the sociodemographic profiles of the tourists, the results of which are summarised in Table 2.The sociodemographic profile of the respondents is shown in Table 2. From the summary of the data collected, it can be concluded that female respondents slightly predominate with 53.6% of the sample, while male respondents account for 46.4% of the sample.The majority of the sample are younger respondents between the ages of 21 and 30 (35.3%), who have a university degree (54.6%) and are employed full-time (55.3%).The distribution of destinations between Istria and Međimurje is almost 50/50.Respondents who remembered their MGE were mostly in the sustainable destination for the first time.
To identify the variables that contribute most to the variation in the data, the authors decided to apply Exploratory Principal Component Analysis (PCA) with varimax rotation as a technique to select the characteristics.PCA led to the distinction of four quality aspects for the creation of an MGE.Following Field (2013), the authors suppressed factor loadings of less than 0.3 and applied the concept of statistical power (Hair al., 2014), adhering to the cut-off value for sample size (n= 414).Four components (quality aspects) with eigenvalues higher than 1 explaining 60% of the variance were retained and named as follows: Aspect 1: The physical environment -PE (22 % of variance); Aspect 2: Service quality -SQ (17 % of variance); Aspect 3: Food and beverage quality -FBQ (12 % of variance); and Aspect 4: Local food -LF (9 % of variance).The reliability test (Cronbach's alpha) for the first three aspects showed good internal consistency, while Cronbach's alpha for the last aspect -LF -was close to 0.6.Although reliability for exploratory research should be 0.6 (Hair et al., 2019), the authors decided to reevaluate this factor by conducting a reliability test for the entire measurement scale with and without the factor in question.The results showed that Cronbach's alpha remained the same, so the measurement scale was retained with all the extracted items listed in Table 3.The degree of intercorrelation between variables was assessed using the Kaiser-Meyer-Olkin measure of sampling adequacy (KMO), which was 0.8777, and Bartlett's Test of Sphericity with Chi-square 1885.164,df =105, significance of p< 0.05.

Reflective measurement model
The outer loading matrix shows that 19 of 21 items satisfy item reliability.Since loadings above 0.708 are recommended (Hair et al., 2019), the deletion of three questionable items with loadings between 0.658 and 6.665 was considered.After reexamination, it was decided to proceed with further analyses and retain these items, as the removal of these items didn't result in any changes in content validity.
Internal consistency reliability was also confirmed using the pA (rho) measure proposed by Dijkstra and Henseler (2015), as the values were above the minimum acceptable value of 0.7.Moreover, AVE was higher than 0.5 for each construct, suggesting satisfactory convergence validity.In addition, HTMT values are below 0.85, indicating the presence of discriminant validity (Franke & Sarstedt, 2019).

Structural model
To evaluate the structural model, the authors followed the procedure developed by Hair et al. (2019) and proposed the standard evaluation criteria (coefficient of determination R2, blindfolding-based cross-validated redundancy measure Q2, and statistical significance and relevance of path coefficients).
The bootstrapping procedure with 5000 iterations was applied to asses the whole model (Table 4) for both groups of interest (Istra and Međimurje).The predictive power of the dependent variable (R2 bihevioral intentions = intentions is 0.370 and R2 MGE=0.206), which means that future behavioural intentions explain 37% of the variance, while MGE explains 21%).Before assessing the structural relationships, the authors checked whether there were problems with the collinearity of the model; however, all VIF values were less than 2, which is the recommended value of less than 5 (Hair et al., 2019).
According to the results, the strongest relationship between MGE and BI (ß = 0.749, p < 0.01) is in the case of Istria.Among the common relations (for Istra and Međimurje) there are insignificant relations (FBQ-> BI, p > 0.05) and (SQ -> MGE, p > 0.05), and significant relationships (LF -> BI; MGE -> BI; PE -> MGE; SQ -> BI, p < 0.05).The effects of LF on BI (ß=0.321,p< 0.05) and of PE on BI (ß=0.202,p< 0.05) were significant only in the case of the Međimurje target.Mediation effects were examined by bootstrapping with 5.000 permutations, following the mediator analysis procedure in PLS by Zhao et al. (2010).The analysis confirmed three indirect relationships only in Istria; FBQ -MGE -BI (ß = 0.097; p < 0.05), PE -> MGE -> BI (ß = 0.202; p < 0.05), and LF->MGE->BI (ß = 0.205; p < 0.05).Since in all three cases the direct effects were insignificant, this implies that the MGE fully mediates these relationships.To test whether the differences between two sustainable destinations are significant, a multigroup analysis was performed on indirect relationships (Table 6).According to the multi-group results presented in Table 6, there are significant differences between two sustainable goals in the indirect paths "PE-> BI" and "LF -> BI" mediated by an MGE.In particular, this result implies that in Istria MGE have a significantly stronger influence on the relationships between PE and BI than in Međimurje.Moreover, in Istria MGE have a significantly stronger influence on the relationships between LF and BI.These results could be explained by the fact that Istria pays special attention to the tradition of excellence in gastronomic offer, standardizing the production of local food and wine.Moreover, Istria is known among gastro-tourists for the variety of local food.

Theoretical implications
More and more frequently, tourists visit Croatian destinations with the aim of consuming local food typical of the region they are staying in during their holiday.Despite the obvious need to understand the perceptions of gastro-tourists, this study is, to our knowledge, the only one focused on examining the MGE of tourists in sustainable destinations.Examining gastro-tourist perceptions in sustainable destinations theoretically contributes to progress in the literature, as previous studies have focused on examining MGE at the level of general tourist destinations (Sthapit, 2017;Hsu & Scott, 2020;Antón et al., 2019;Kovalenko et al., 2023).
Inevitably, there is little agreement on the theories (Bigné et al., 2020) and research settings in the study of MGE (Rašan & Laškarin Ažić, 2023).Therefore, the results of this study contribute to the existing literature in several areas.First, the multidimensional nature of the construct PQ (H2) was demonstrated.The identified aspects of PQ experience -PE, SQ, and FBQ -are commonly known in the literature (Tsaur & Lo, 2020;Kala, 2020;Kim et al., 2021;Zhang et al., 2019b).In contrast, the LF aspect is mentioned in only a few MGE studies (Badu-Baiden et al., 2022a;Stone et al., 2018).LF refers to authentic products that not only symbolize tourist destinations but also vividly represent local traditional culture.According to Choe and Kim (2018), Demirkol andCifci (2020), andRodrigues et al. (2023), LF expresses national, regional, and personal identity and can enhance the destination's image.In order for a sustainable destination to be gastronomically recognizable, the range of LF should be as diverse as possible.
The results of this research suggest that three of the four aspects of PQ (FBQ, LF, and PE) have a direct relationship with MGE in both destinations (H3).Previous studies have shown that employee service delivery behaviors strongly contribute to tourists' perceptions of the quality of the service itself (Line & Hanks, 2019), which ultimately has a direct impact on tourists' overall experience (Keith & Simmers, 2011).However, the results of this research suggest otherwise, as the relationship between SQ and MGE was not found to be significant.
Third, H4 cannot be confirmed for these cases because there are differences among sustainable destinations in the effects of certain elements of perceived quality (FBQ, LP, and PE) on behavioural intentions.However, H4 is confirmed in the case of the effect of FBQ on BI, as there is a significant direct relationship for both sustainable destinations.
Fourth, the analysis process suggests that for the most part (with the exception of SQ), the MGE mediates the relationships between aspects of PQ and BI tourists in both sustainable destinations (H5).The congruence of these results is evident with the research of Horng and Hsu (2021), which found the mediating role of MGE between all aspects of PQ, Pleasantness, and BI.Li et al. (2023) also demonstrated the mediating effect of MGE, but between food authenticity, sensory appeal, and BI.On the other hand, Shahzadi et al. (2018), in a study on fine dining restaurants, found that the mediating role between PQ and BI has tourist satisfaction (an affective component of the experience).
Fifth, the findings on the significance of the relationship between MGE and BI tourists (the intention to recommend and the intention to revisit) are consistent with previous conclusions from the literature (Horng & Hsu, 2021;Di-Clemente et al., 2019;Soonsan & Somkai, 2023;Antón et al., 2019).This suggests that H6 has been confirmed.Moreover, it is not surprising that this study revealed statistically significant differences in the nature of the relationship between the observed constructs in both samples (PE-> MGE-> BI and LF-> MGE-> BI), as there are discrepancies between sustainable destinations -Istria and Međimurje (see subsection 2.1).The above comments imply the assumption of H1.

Practical implications
The results of the current research can serve as a basis for raising awareness of the importance of promoting food sustainability in local and regional communities.Significant contributions of the work can be highlighted in the context of valuable inputs for destination management organizations (DMOs), with the aim of improving certain elements of the destination gastronomic offer.The identified factors can help tourism service providers such as destination managers, restaurateurs and gastronomic event organizers to improve their current elements of the gastronomic offer in their own environments.Since the results show a significant difference between two sustainable destinations in terms of impact on BI (PE, LF), it is recommended that each sustainable destination integrates and evaluates the identified factors in its own environment.To attract future destinations BI, DMOs should also be interested in integrating local products into the local community.For example, DMOs should provide clear instructions and explain "why using locally grown and produced food in their restaurants will pay off in the long run".
The most important contribution for destination management organisations is the recommendation to focus on the promotion of glocal gastronomy as the ultimate guideline for the future development of this segment of the tourism offer.The glocalness of gastronomy is crucial, as it has a positive impact not only on the loyalty of tourists, but also on the preservation of the local and regional cultural identity and gastronomic richness of a community.Therefore, the distribution of local gastronomic products to international tourists can create a marketing brand as a glocal tourist destination.

Limitations and future research
Although this research provides insights into the current state of the MGE, it has certain limitations.First, because the research results were generalized through the quantitative methods, the use of the qualitative approach is recommended for future research.
Further, this research was conducted on the model of two sustainable destinations, which does not necessarily guarantee transferability to a larger geographical area and highlights the need to modify the model according to the specifics of the destinations where further research will be conducted.Clearly, a recommendation for future research is to extend the research model to a larger geographic area.
As the sample of respondents included only gastro-tourists visiting two sustainable destinations, Međimurje and Istria, regardless of their number (n = 414), it is not possible to generalize data about tourists visiting other destinations, this study examined the loyalty of gastro-tourists based on tourists' intention to return and their recommendation.Although this approach is widely used in the literature, it might be better for future research to use a multidimensional loyalty approach for a more detailed understanding (Suhartanto et al., 2013).

Table 1 .
Operationalization of research constructs Data collection was conducted in two Croatian sustainable destinations, Međimurje and Istria, from July 2021 to February 2022, considering both summer and winter seasons in Croatia.The selection of these two destinations was based on the criterion of promoting sustainable gastronomic development, which has a positive impact on tourism (see section 2.1.).The questionnaire was available to foreign and domestic tourists in English and Croatian.The employees of seven restaurants in Istria and Međimurje helped the professional surveyors with the exit data collection by kindly guiding them to the surveyors.

Table 3 .
PCA analysis -PQ aspects of experience

Table 4 .
Direct relationships