Skip to content
Frontiers
Search

Original Research

Frontiers in Sustainable Food Systems ·

Volume 10 - 2026 | https://doi.org/10.3389/fsufs.2026.1976172

Digital capability and supply chain resilience in rural food tourism ventures: the role of food supply chain integration

  • Wei Li1,2
  • Peng Yin3*corresponding author
  • Ruirui Zhou4
  1. 1Business School, Beijing Technology and Business University, Beijing, China
  2. 2School of Business Administration, Shandong Women's University, Jinan, China
  3. 3Business School, LuDong University, Yantai, China
  4. 4Shandong Provincial Archives of Natural Resources (Shandong Geological Museum), Jinan, China
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Rural food tourism ventures coordinate perishable local supplies, food-service operations and fluctuating tourist demand. How digital capability and food supply chain integration relate to resource-saving practices and enterprise-level resilience remains unclear. A four-wave matched survey of 467 women operators was analyzed using covariance-based structural equation modeling, bootstrap tests of statistical indirect associations and an ordinal-estimator sensitivity analysis. Public user-generated content was analyzed separately to characterize platform discourse. Digital capability was positively associated with food supply chain integration and enterprise-level supply chain resilience. Integration was positively associated with both the low-carbon-oriented and resource-saving food practices index and resilience. The direct association between the practices index and resilience was not statistically significant. Bootstrap estimates supported the indirect association between digital capability and the practices index through integration, but not the serial indirect association extending to resilience. Only the resilience text classifier met the prespecified reporting threshold; the remaining labels were retained as exploratory semantic cues. The findings distinguish resource-saving practices from disruption resilience and identify supply chain integration as an organizational correlate of both. Future research should test these associations in probability-based and gender-comparative samples and validate reported practices against measured energy use, food loss and emissions.

Keywords

1 Introduction

Local food tourism research increasingly examines how food-based tourism affects community wellbeing and the social sustainability of destinations, alongside its established concern with visitor experience and cultural heritage (Agyeiwaah and Zhao, 2024; Orea-Giner et al., 2026; Wan et al., 2025). In rural destinations, food connects agricultural production with hospitality and tourist consumption. Agritainment, homestay catering, local food retail and traditional food experiences organize these activities through short, place-dependent supply chains. Such arrangements can retain value locally and strengthen producer–consumer relations, but their sustainability and resilience depend on sourcing, transport, resource use and coordination rather than geographical proximity alone (Jia et al., 2024; León-Bravo et al., 2025; Stein and Santini, 2022). Limited inventory, capital reserves and bargaining power can constrain rural food tourism ventures. Ingredient seasonality, food-safety requirements, variable tourist flows and weather-related access can therefore disrupt menus and service. Women operators are prominent in this setting and often draw on food knowledge and community relations while facing constraints in finance, care responsibilities and resource access (Khazami et al., 2024; Martini et al., 2020; Nordbø, 2022; Tao et al., 2025). The all-women operator sample defines the empirical boundary; the design does not test gender differences.

Digital booking, social media, e-commerce, mobile payment, electronic procurement and source tracing are reshaping information connections from supply to consumption in local food systems. Digital tools increase the visibility of orders, inventory, ingredient quality and customer preferences. Their organizational value, however, depends on the capability to embed data in procurement, processing, sales and customer interaction (Belhadi et al., 2024; Khin and Ho, 2019; Taghikhah et al., 2025). Research on agri-food digitalization similarly locates transformation in data processing, interoperability, user adoption and cross-actor connection rather than in isolated technology deployment (Vahdanjoo et al., 2025; Wang et al., 2024). In rural food tourism, dispersed farmers often lack timely demand information, while operators must coordinate seasonal ingredients, meal preparation and tourist bookings with limited internal information. Food supply chain integration shortens information transmission and helps enterprises align procurement volume, menus and processing arrangements with demand changes (Flynn et al., 2010). Low-carbon-oriented and resource-saving practices, including on-demand procurement, resource-efficient processing, packaging reduction and food-loss prevention, also require suppliers, employees and customers to share information and adjust behavior. Existing studies examine digital technology and green integration, supply chain integration and sustainable performance, and digital capability and agri-food supply chain resilience (Belhadi et al., 2024; Han and Huo, 2020; Yadav et al., 2024). The route through which digital capability enters daily resource-saving operations and enterprise-level disruption response in rural food tourism ventures remains under-specified.

Research on rural tourism entrepreneurship has concentrated on entrepreneurial entry, empowerment, resource constraints and local embeddedness, leaving the organization of procurement, processing, service and consumption across local food supply chains underexamined (Ali et al., 2025; Martini et al., 2020; Nordbø, 2022). Agri-food digitalization studies more often examine larger production, processing or retail organizations and seldom distinguish digital capability from interorganizational integration and operating practices (Taghikhah et al., 2025; Vahdanjoo et al., 2025). Evidence also remains limited on whether resource-saving routines have an independent association with disruption response in rural food tourism ventures (Paredes-Rodríguez et al., 2024; Yuan et al., 2024). The analysis addresses two research questions. RQ1 asks how digital capability and food supply chain integration are associated with resource-saving practices and enterprise-level supply chain resilience in rural food tourism ventures. RQ2 asks whether the low-carbon-oriented and resource-saving food practices index retains an independent association with resilience and whether the serial statistical indirect association from digital capability through integration and the practices index is supported. Organizational information processing theory provides the theoretical basis for these associations. Evidence is drawn from 467 four-wave matched questionnaires, while public user-generated content is examined separately to characterize the surrounding platform discourse.

2 Literature review

2.1 Rural food tourism entrepreneurship and local food systems

Rural food tourism enterprises sit at the intersection of tourism services and the agri-food supply chain. Food constitutes an important medium through which tourists perceive local authenticity and connects farmers, processors, food service operators, and consumers. Short food supply chain research shows that shortening the organizational distance between producers and consumers can improve source transparency, relational trust, and local value retention, and create conditions for small-scale producers to enter the market (Jia et al., 2024; Sciortino et al., 2025). Local food tourism research links food consumption with tourist experience, social sustainability, community benefits and responsible tourism (Agyeiwaah and Zhao, 2024; Orea-Giner et al., 2026; Wan et al., 2025). However, the environmental and resilience effects of local sourcing are not uniform: shorter transport distances can be offset by low load factors, high-frequency delivery, or single-origin dependence, and local systems can also be exposed to common shocks due to supply concentration (León-Bravo et al., 2025; Stein and Santini, 2022).

A ‘women-led rural food tourism enterprise’ is a rural venture in which women participate in core operating decisions and local food procurement, processing, sales or experiential services constitute the principal business activity. These enterprises commonly combine household labor, place-based food knowledge and community ties with limited financial, managerial and logistical resources. Care responsibilities and locally embedded business relations further shape access to information, suppliers and external support (Martini et al., 2020; Nordbø, 2022; Ali et al., 2025; Khazami et al., 2024; Tao et al., 2025). Digital capability increases the visibility of demand and operating information, whereas supply chain integration coordinates information and resources across farmers, platforms, customers and food-service activities.

2.2 Digital capability and food supply chain integration

Digitalization research increasingly distinguishes technology adoption from digital capability. Technology adoption focuses on whether a firm uses a certain tool, whereas digital capability emphasizes the firm's ability to identify digital opportunities, configure digital resources, and use them to renew products, services, and processes (Khin and Ho, 2019). The popularity of social media or mobile payment can lower the access barrier, but if fragmented tools are not connected to procurement, inventory, menus, and customer management, the resulting data cannot readily support consistent decision-making. Agri-food system research also shows that the value of digital transformation comes from data processing, interoperability, user adoption, and cross-actor connection rather than the individual technology itself (Vahdanjoo et al., 2025; Wang et al., 2024). The value of digital capability therefore lies not in the number of tools but in whether information can be acquired, interpreted, shared, and embedded in organizational processes.

Supply chain integration describes the extent of information sharing, joint planning and process coordination within a firm's functions and between the firm and its suppliers and customers (Flynn et al., 2010). Internal integration connects procurement, processing, sales and service; supplier integration improves the visibility of ingredient quality, quantity and delivery status; customer integration brings bookings, reviews and preference feedback into supply arrangements. In this setting, environmental uncertainty refers to unpredictable changes in ingredient availability and quality, supplier delivery, tourist demand, weather-related access and online-platform conditions. Organizational information processing theory links digital capability with supply chain integration because these changes increase information requirements, while digital systems and cross-boundary coordination expand the capacity to collect, transmit and use relevant information (Galbraith, 1974; Tushman and Nadler, 1978). Recent agri-food research further shows that associations between digital capability and supply chain resilience depend on complementary capabilities in logistics, data security, customer value and decision support (Belhadi et al., 2024; Taghikhah et al., 2025).

Existing digital supply chain research often examines visibility, integration, innovation and performance as parallel outcomes, leaving the organizational use of digital information in cross-actor collaboration underexamined in rural food tourism ventures. Yadav et al. (2024) used organizational information processing theory to explain the link among digital technology, green integration, and agri-food supply chain sustainable performance, showing that digitalization and integration are complementary; Jantapoon and Amornwattahcharoenchai (2025) also examine the relationships among digital capability, knowledge integration, resilience, and environmental sustainability in agri-food supply chains. However, these studies do not capture the procurement-processing-experience chain of rural food tourism ventures. In settings characterized by informal relations and seasonal supply, formal digital records and local business relations jointly shape the coordination of resource-saving practices and disruption responses.

2.3 Low-carbon-oriented and resource-saving food practices

Low-carbon-oriented and resource-saving food practices comprise operating routines intended to reduce avoidable energy and material use and limit food loss across procurement, processing, service and residual handling. The construct concerns reported operating practices rather than measured emissions or verified carbon reductions. Green supply chain research places environmentally informed purchasing and supplier cooperation within upstream environmental management (Zhu et al., 2008). Restaurant studies document related practices in local and seasonal purchasing, energy and water management, portion and menu planning, packaging reduction and food-residue handling (Maynard et al., 2020). Evidence from chefs further identifies food-waste tracking, ingredient reuse, menu adjustment, donation and composting as responses to surplus food (Reardon et al., 2024). In rural food tourism ventures, these activities span sustainable procurement, resource-efficient processing, and food-loss and circular management.

2.4 Enterprise-level supply chain resilience

Enterprise-level supply chain resilience is a venture’s capacity to anticipate supply disruptions, maintain or adjust procurement and food-service operations while disruptions unfold, and restore supply and service afterward (Ambulkar et al., 2015). In rural food tourism, disruptions include ingredient shortages, delivery delays, food-quality problems and sudden changes in tourist demand. Resilient responses include replacing ingredients, adjusting menus, switching supply or sales channels and maintaining essential services. Short food supply chains can support adaptation through relational proximity and rapid feedback, but source concentration, limited storage and transport capacity, and inadequate infrastructure can also amplify common risks (Jia et al., 2024; Sciortino et al., 2025). Resilience therefore rests on information visibility, alternative resources and adjustment capacity rather than chain length alone.

Evidence remains fragmented across digital capability, interorganizational coordination, resource-saving operations and disruption adaptation. Research on women’s rural tourism entrepreneurship seldom follows procurement, processing and consumption across organizational boundaries. Agri-food digitalization studies often combine technological resources with organizational integration, whereas sustainability and resilience research generally examines resource efficiency and disruption adaptation separately (Zhong et al., 2026). Distinguishing digital capability, food supply chain integration, low-carbon-oriented and resource-saving practices, and enterprise-level supply chain resilience permits their direct and indirect statistical associations to be assessed separately.

3 Theoretical background and hypotheses

3.1 Organizational information processing theory

Organizational information processing theory explains how organizations acquire, process and transmit information under uncertainty. Galbraith (1974) argued that task uncertainty increases information-processing requirements and that organizations respond by developing rules, plans, lateral relations and information systems. Tushman and Nadler (1978) further proposed that organizational performance depends on the fit between information-processing requirements and information-processing capacity. When information flow is insufficient or coordination mechanisms are weak, organizations face delayed decisions, misallocated resources and inconsistent actions.

Rural food tourism enterprises operate in local supply chains shaped by seasonal agricultural production, variable tourist demand and place-specific business relationships. Agricultural supply, food-service processing, local food sales, tourist consumption and platform communication are interconnected; operators must process information on ingredient seasonality, tourist-flow fluctuations, food safety, logistics delays and online reviews simultaneously. Digital booking, social media, e-commerce, mobile payment and source records increase the visibility of orders, inventory, customer preferences and ingredient supply. These information sources support stable operational adjustment only when incorporated into coordination among suppliers, employees, customers and platforms.

Digital capability (DC) represents the capacity to acquire and process operational information (Khin and Ho, 2019); food supply chain integration (SCI) represents the internal and cross-boundary coordination through which information enters joint planning and operations (Flynn et al., 2010). The low-carbon-oriented and resource-saving food practices index (LCP) records complementary resource-saving routines, while enterprise-level supply chain resilience (SCR) represents disruption preparedness, adaptation and recovery (Ambulkar et al., 2015; Zhu et al., 2008). This distinction follows organizational information processing theory, which separates information-processing requirements, processing capacity and organizational coordination arrangements (Galbraith, 1974; Tushman and Nadler, 1978).

3.2 Digital capability and food supply chain integration

Digital capability refers to a firm's ability to identify digital opportunities, configure digital resources and use data to improve operational processes (Khin and Ho, 2019). In rural food tourism supply chains, demand information comes from bookings, platform reviews, instant communication and on-site consumption, whereas supply information comes from farmers, wholesalers, logistics providers and inventory records. Dispersed and frequently updated information increases the risk of order omissions, stocking errors and delivery delays when firms rely on oral transmission or personal memory. Digital capability improves the timeliness, traceability and transferability of data, creating a basis for joint planning. Recent agri-food system research identifies interoperability, user adoption, data processing and cross-actor connection as key conditions for digital transformation, which aligns closely with the integration needs of rural food tourism enterprises (Vahdanjoo et al., 2025; Wang et al., 2024).

Digital capability increases the availability and transferability of operating information, whereas supply chain integration embeds that information in joint planning across organizational boundaries. Operators can arrange procurement and scheduling based on digital bookings, feed sales and surplus data back to suppliers, and coordinate quality and delivery through traceable records; customer feedback can also be incorporated into menu design and portion adjustment. When digital capability is weak, even if the firm has stable local relations, information remains confined to ad hoc communication rather than becoming embedded across procurement, processing, and sales. Existing research finds that digital technology can be associated with green integration and supply chain collaboration (Yadav et al., 2024), and agri-food enterprises also use different levels of digital tools to build sensing and response capabilities according to scale and chain position (Belhadi et al., 2024; Taghikhah et al., 2025).

H1: Digital capability is positively associated with food supply chain integration.

3.3 Food supply chain integration and the low-carbon-oriented and resource-saving food practices index

Food supply chain integration contains three interrelated dimensions: internal, supplier and customer integration. Internal integration aligns plans across procurement, processing, sales and service; supplier integration supports joint decisions on product categories, quality, quantities and delivery schedules; customer integration incorporates booking, preference and consumption feedback into operational decisions (Flynn et al., 2010). The low-carbon-oriented and resource-saving food practices index spans the same operating links. Aligned demand and supply information can reduce excess purchasing and food loss while supporting seasonal menu planning and more efficient production batches.

Supply chain integration is associated with low-carbon-oriented practices because it brings demand and supply information into routine operational coordination. Internal integration links order information with procurement and meal-preparation plans, reducing waste from fragmented information. Supplier integration improves negotiation over seasonal ingredients, delivery batches and packaging methods, supporting local procurement and consolidated delivery. Customer integration supports portion forecasting, preference identification and booking guidance, reducing surplus from the demand side. Green supply chain research shows that environmental improvement depends on the joint action of internal, supplier and customer integration and that different integration dimensions correspond to different sustainable outcomes (Han and Huo, 2020). Fresh-food supply chain research also indicates that material reduction, delivery optimization and biological-waste management often require cross-organizational cooperation (Toniolo et al., 2024). Coordination among farmer supply, internal operations and tourist demand is therefore expected to be associated with the adoption of low-carbon-oriented and resource-saving practices.

H2: Food supply chain integration is positively associated with the low-carbon-oriented and resource-saving food practices index.

3.4 Food supply chain integration and enterprise-level supply chain resilience

Enterprise-level supply chain resilience requires enterprises to identify risks before disruptions, adjust resources during disruptions, and restore basic supply after disruptions (Ambulkar et al., 2015). The disruptions faced by rural food tourism enterprises include raw material shortages, supplier default, traffic interruption, food-quality anomalies, and sudden demand changes. Such disruptions usually cross enterprise boundaries and are difficult to handle promptly by relying solely on internal inventory or operators' experience.

Supply chain integration is expected to expand adaptive capacity by improving visibility and coordination speed. Information shared across procurement, food preparation, customer-facing operations and online channels allows menus, portions and fulfillment commitments to be adjusted together. Earlier notice of shortages or delivery changes also permits operators and suppliers to identify alternative ingredients, batches or transport arrangements, while communication with customers supports booking and service adjustments. Organizational information processing theory predicts that shorter information paths and lateral coordination help contain disruptions before they spread across operations. Evidence from digital supply chains similarly associates visibility, risk-information processing and cross-actor collaboration with enterprise-level resilience (Belhadi et al., 2024). Partner and customer information is especially consequential for ventures with limited internal buffers.

H3: Food supply chain integration is positively associated with enterprise-level supply chain resilience.

3.5 Low-carbon-oriented and resource-saving food practices index and enterprise-level supply chain resilience

The low-carbon-oriented and resource-saving food practices index and enterprise-level supply chain resilience concern distinct organizational outcomes: resource-saving operations and adaptation to disruptions. On-demand procurement, portion management, food-loss recording and flexible seasonal menus can preserve resources for disruption response. Conversely, inventory compression, concentrated sourcing and implementation costs can constrain operational buffers. These resource-efficiency and buffer-constraint arguments imply competing directions for the LCP–SCR association.

The net association is therefore examined with a non-directional hypothesis:

H4: The low-carbon-oriented and resource-saving food practices index is associated with enterprise-level supply chain resilience.

3.6 Digital capability and enterprise-level supply chain resilience

Digital capability is expected to be positively associated with enterprise-level supply chain resilience. Digital booking and sales data support demand sensing; electronic procurement and platform connections expand potential supply alternatives; and online channels help enterprises adjust transactions and fulfillment modes when on-site tourist flows are disrupted. Stronger digital capability also helps enterprises retain order, quality and delivery records, providing continuous information for disruption assessment. Agri-food supply chain research shows that enterprises use digital sensing, seizing and reconfiguring capabilities to formulate differentiated risk-response strategies (Belhadi et al., 2024), and digital supply chain studies link digitalization with risk management and resilience.

H5: Digital capability is positively associated with enterprise-level supply chain resilience.

3.7 The statistical mediating role of food supply chain integration

Digital capability supplies operational information, but this information must enter joint planning and routine operations before it can be reflected in low-carbon-oriented practices. Food supply chain integration links digital records on orders, inventory and ingredient quality with supplier coordination, internal workflow alignment and customer participation. Such coordination supports on-demand procurement, consolidated delivery, portion management and food-loss reduction. Evidence linking digital technology, green integration and sustainable performance indicates that digital resources are associated with environmental practices through upstream and downstream coordination (Han and Huo, 2020; Yadav et al., 2024). Food supply chain integration is therefore expected to transmit the association between digital capability and low-carbon-oriented practices.

H6: Digital capability has a positive statistical indirect association with the low-carbon-oriented and resource-saving food practices index through food supply chain integration.

3.8 Serial statistical indirect association

Digital information can enter coordinated procurement and resource-saving routines, but the association of these routines with disruption response remains theoretically indeterminate. The proposed DC → SCI → LCP → SCR sequence therefore tests whether the statistical indirect association extends to enterprise-level resilience without prespecifying the sign of the final link.

H7: Digital capability has a serial statistical indirect association with enterprise-level supply chain resilience through food supply chain integration and the low-carbon-oriented and resource-saving food practices index.

Figure 1 presents the theoretical framework.

Theoretical framework. H1–H5 denote hypothesized direct associations; the DC → LCP arrow is a prespecified direct path. H6 (DC → SCI → LCP) and H7 (DC → SCI → LCP → SCR) are bootstrap-tested products of the displayed direct-path coefficients rather than additional structural paths.
Figure 1. Theoretical framework. H1–H5 denote hypothesized direct associations; the DC → LCP arrow is a prespecified direct path. H6 (DC → SCI → LCP) and H7 (DC → SCI → LCP → SCR) are bootstrap-tested products of the displayed direct-path coefficients rather than additional structural paths.

4 Materials and methods

4.1 Research design and analytical logic

A four-wave matched questionnaire provides the enterprise-level data, and public user-generated content (UGC) supplies a separate account of platform discourse. The rural food tourism venture is the questionnaire unit of analysis, and the questionnaire sample comprises women-led ventures. Digital capability (DC), food supply chain integration (SCI), the low-carbon-oriented and resource-saving food practices index (LCP), and enterprise-level supply chain resilience (SCR) were measured at different waves. Structural equation modeling (SEM) was implemented in covariance-based form (CB-SEM) to estimate enterprise-level statistical associations among these constructs. Public UGC describes platform-visible discourse concerning digitalization, resource-saving practices, resilience and operating constraints and is not included in the SEM hypothesis tests.

Measurement and structural models were estimated sequentially following Anderson and Gerbing (1988). Confirmatory factor analysis (CFA) assessed the reflective measurement model before the three integration dimensions were specified as a second-order SCI construct. The structural model estimates six direct paths. Case-level bootstrap resampling estimates H6 and H7 as products of coefficients derived from those paths; neither hypothesis adds a structural arrow. A weighted least squares mean- and variance-adjusted (WLSMV) model re-estimates the same six direct paths as an ordinal-item sensitivity analysis (Flora and Curran, 2004; Li, 2016). LCP was specified from complementary operating activities (Diamantopoulos and Winklhofer, 2001; Jarvis et al., 2003), excluded from reflective CFA, and entered into the structural model as a precomputed observed composite with an estimated residual variance.

4.2 Data sources, survey procedure and analytical sample

The survey targeted women entrepreneurs who participated directly in operating decisions in rural food tourism ventures. Eligible ventures were located in rural areas or relied primarily on rural resources and directly engaged in food procurement, processing, sales or experiential activities. Each venture had operated continuously for at least 12 months, maintained at least one established relationship with a farmer or food supplier, and used at least one digital operating tool during the preceding year. The four-wave survey was conducted from June to October 2025 in Jinan, Yantai, Linyi, Tai'an, Zaozhuang and Rizhao, Shandong Province. The final sample comprised 89 ventures in Jinan, 87 in Yantai, 79 in Linyi, 85 in Tai'an, 65 in Zaozhuang and 62 in Rizhao. Recruitment drew on contacts established through the authors' participation in rural outreach training in hospitality service skills organized by the Shandong Provincial Tourism Bureau since 2015. Training for farm-based and fishing-based rural tourism operators provided channels for contacting eligible women operators and arranging follow-up. Recruitment used these established contacts rather than a population-based probability sampling frame. Before accessing the questionnaire, participants read the study information on the opening page and provided documented electronic informed consent by selecting “I agree to participate.” The survey platform recorded this selection before permitting access to the questionnaire.

Wave 1 (T1) measured digital capability and enterprise characteristics; Wave 2 (T2) measured internal, supplier and customer integration; Wave 3 (T3) measured the LCP index and self-reported operating-check indicators; and Wave 4 (T4) measured supply-disruption experience and enterprise-level supply chain resilience. T1 was administered in mid-June, T2 in mid-July, T3 in mid-August and T4 in mid-October 2025. The intervals between successive waves were approximately one month, one month and two months, respectively. Anonymous identification numbers linked responses across waves. Records were obtained from 501 women operators; questionnaire-quality review retained 467 matched records for analysis (93.2%). Full-time employment and annual revenue described enterprise scale; neither was used as an eligibility cutoff. Thirty-four records were excluded under the applied screening rules: completion in less than 90 s at T1, less than 120 s at any of T2–T4, or selection of the same response option for all scale items within a wave. Mandatory-response validation prevented submission of incomplete questionnaires. LCP responses coded as not applicable were retained as valid responses rather than missing data. Mean respondent age was 42.82 years (SD = 9.05), mean operating tenure was 5.07 years (SD = 2.62), mean full-time employment was 8.63 workers (SD = 9.26; median = 6; IQR = 3–10; range = 1–60), and enterprises had a mean of 6.00 major food supply sources (SD = 2.89). Education levels of junior secondary school or below, vocational/high school diploma, associate degree, bachelor's degree and postgraduate degree accounted for 17.1%, 30.4%, 29.6%, 18.6% and 4.3%, respectively. Annual revenue was below RMB 300,000 for 254 ventures (54.4%) and at least RMB 1 million for 46 ventures (9.9%); all five revenue categories are reported in Table 1. Supply disruptions during the preceding 12 months were reported by 291 enterprises (62.3%). Table 1 reports respondent and enterprise characteristics.

Table 1. Characteristics of respondents and enterprises (N = 467).

CharacteristicCategory/statisticn%CharacteristicCategory/statisticn%
Age (years)M = 42.82, SD = 9.05, range = 25–65467—Food-related revenue share0–20%469.9
EducationJunior secondary school or below8017.121–40%7516.1
Vocational/high school diploma14230.441–60%15433.0
Associate degree13829.661–80%12526.8
Bachelor’s degree8718.681–100%6714.3
Postgraduate degree204.3Digital-skills trainingYes24051.4
Main business typeRural restaurant7115.2No22748.6
Homestay catering7415.8Green/loss-reduction trainingYes19341.3
Local food retail8217.6No27458.7
Agricultural experience8317.8Financing availabilityVery difficult132.8
Integrated lodging and catering7215.4Difficult10522.5
Other8518.2
Annual revenue< RMB 100,00010622.7Moderate21546.0
RMB 100,000–299,99914831.7Easy12226.1
RMB 300,000–499,9998618.4Very easy122.6
RMB 500,000–999,9998117.3Supply disruption in past 12 monthsYes29162.3
≥ RMB 1,000,000469.9No17637.7
Weekly unpaid care (hours)M = 8.41, SD = 4.81; median = 8, IQR = 5–12; range = 0–21Full-time employeesM = 8.63, SD = 9.26; median = 6, IQR = 3–10; range = 1–60
Operating history (years)M = 5.07, SD = 2.62, range = 1–15Main food suppliersM = 6.00, SD = 2.89, range = 1–17

M, mean; SD, standard deviation; IQR, interquartile range.

4.3 Measures and scale formation

Except for the low-carbon-oriented and resource-saving food practices index, all core constructs used five-point Likert scales ranging from 1 (strongly disagree) to 5 (strongly agree). Cognitive interviews involved 12 women operators, followed by a pretest with 45 operators. Item wording was adapted to the respondents’ operating roles, business activities and disruption contexts while retaining the original construct definitions.

Digital capability was measured with five items adapted from Khin and Ho (2019), covering the acquisition of digital technology, identification of digital opportunities, response to technological change and incorporation of digital tools into procurement, sales and food-related operations. Food supply chain integration was measured with items adapted from Flynn et al. (2010). Four items represented each of its three dimensions: internal integration, supplier integration and customer integration. Internal integration concerns information connections among procurement, processing, sales and service; supplier integration concerns information sharing and joint arrangements with farmers and food suppliers; customer integration concerns the connection of bookings and feedback with operating plans. The three first-order factors represent the second-order SCI construct.

The LCP index was constructed for this study using practice domains from Zhu et al. (2008) manufacturing-sector green supply chain measure and operating activities documented in Maynard et al. (2020) restaurant sustainability checklist. The 12 items and their 0–5 response format were adapted to rural food tourism rather than reproduced verbatim from either source instrument. Responses range from 0 to 5: 0 denotes that an activity is not applicable to the enterprise, 1 denotes not implemented, and 5 denotes systematically implemented. A 0 response was excluded from the denominator of the relevant dimension. Each dimension equals the mean of its applicable items, and LCP equals the arithmetic mean of the three dimension scores; each dimension therefore has a weight of one-third. The formative specification follows from the content structure: the three dimensions represent complementary components, and omitting any dimension would change the domain represented by LCP. The dimension weights were fixed rather than estimated from the SEM. LCP measures operator-reported low-carbon-oriented and resource-saving practices, not actual emissions or carbon-reduction performance.

Enterprise-level supply chain resilience was measured primarily with the enterprise-level scale developed by Ambulkar et al. (2015). Four items retained the original domains of continuity, adjustment, resource mobilization and recovery. A fifth item captured learning and improvement after disruptions in rural food tourism enterprises. The items referred to raw material shortages, supplier default, logistics disruption and sudden demand changes. The fifth item was developed for the rural food tourism context and evaluated through expert review, cognitive interviews and pretesting.

4.4 SEM analytical procedure

Primary SEM analyses used Python 3.9.7 and semopy 2.3.11; the ordinal sensitivity analysis used R 4.6.1 and lavaan 0.7–2. The main CFA and structural model treated the five-point reflective items as continuous and used maximum-likelihood estimation with the Wishart likelihood (MLW), optimized by sequential least squares programming (SLSQP). Model-based standard errors came from the inverse expected information matrix. Two-sided Wald z tests evaluated structural coefficients at α = 0.05, and two-sided 95% percentile bootstrap confidence intervals evaluated statistical indirect associations (Preacher and Hayes, 2008; Hayes and Scharkow, 2013). Model fit was assessed using the chi-square statistic (χ2), degrees of freedom (df), the comparative fit index (CFI), Tucker–Lewis index (TLI), root mean square error of approximation (RMSEA) and standardized root mean square residual (SRMR) (Hu and Bentler, 1999; Kline, 2016). Latent variables were scaled by fixing the first loading to 1; the second-order SCI factor was identified by its three first-order integration factors. No correlated residuals were added. LCP was not modeled as a reflective latent construct in the CFA and entered the structural model as a precomputed observed composite with an estimated residual variance.

Internal consistency was assessed with Cronbach's alpha (α) and composite reliability (CR); convergent validity used standardized loadings and average variance extracted (AVE) (Fornell and Larcker, 1981; Raykov, 1997), and discriminant validity used the heterotrait–monotrait ratio (HTMT) (Henseler et al., 2015). Variance inflation factors (VIFs) assessed collinearity among the three formative LCP dimensions; their content coverage informed the composite specification (Diamantopoulos and Winklhofer, 2001). Temporal separation, anonymous matching and neutral wording were procedural safeguards against common-method bias. The theoretical measurement model was also compared with a single-factor model (Podsakoff et al., 2003; Richardson et al., 2009). All constructs remained operator-reported. Eligibility requirements for prior digital-tool use and an established supplier relationship excluded digitally inactive and non-networked enterprises and may restrict the variance of DC and SCI.

4.5 Public UGC text data and analytical procedure

Public-text analysis characterizes the retrieved platform discourse surrounding rural food tourism and related entrepreneurship, agriculture and consumption issues. The corpus includes comments, personal accounts, promotional posts and media material; it is not a sample of verified enterprise observations. Record identifiers link audit decisions, source metadata, adjudicated labels, classifier probabilities and topic assignments. Text-derived associations are analyzed separately from the questionnaire-based SEM.

The initial retrieval contained 36,380 records from Zhihu, Bilibili, Weibo and Xiaohongshu. The initial audit classified 34,533 records as eligible and flagged 1,847 records because of source, text-length or duplication concerns. Removing empty texts and duplicate URL–text pairs yielded 18,565 records. Missing comment permalinks were linked to retained comment and parent-content identifiers. Records classified as eligible satisfied the audit criteria but were not independently verified for topical relevance.

Two coders assessed a stratified sample of 656 records, and disagreements were adjudicated. Raw agreement ranged from 0.730 to 0.892 and Cohen's kappa (κ) from 0.151 to 0.413. Records sharing the same parent content or identical text were assigned to the same cross-validation group, producing 459 groups.

The multi-label classifiers used logistic regression with inverse-frequency class weights and word- and character-level term frequency–inverse document frequency (TF–IDF) features. Vocabulary and inverse document frequency weights were fitted within each training fold. Five-fold stratified group cross-validation used seed 20260821. A label qualified for full-corpus model-predicted reporting only when its out-of-fold F1 score and average precision (AP) both reached 0.70 at the retained decision threshold of 0.50. Threshold sensitivity was examined from 0.30 to 0.70. Supplementary Table S5 reports agreement and threshold results.

Topic modeling used 1,914 non-comment texts containing at least 50 characters; 270 shorter non-comment records were excluded. Non-negative matrix factorization (NMF) was applied to Chinese-tokenized TF–IDF features with a maximum vocabulary of 5,000 terms, a minimum document frequency of 4 and a maximum document frequency of 0.75. The five-topic model used non-negative double singular value decomposition with zeros filled by the mean (NNDSVDa) initialization, seed 20260821 and 600 maximum iterations. Solutions containing four to seven topics and five random initializations were examined to assess sensitivity to topic number and initialization. Supplementary Table S6 reports the alternative solutions, random-initialization stability and topic diagnostics.

5 Results

5.1 Data quality, descriptive statistics and correlations

The 467 matched records contained no missing core reflective items, out-of-range values or duplicated identification numbers. The 599 zero responses among the 12 LCP items represented questionnaire-defined non-applicability and were excluded from applicable-item means. Item distributions showed no severe univariate non-normality (maximum absolute skewness = 0.593; maximum absolute excess kurtosis = 0.769).

Table 2 reports descriptive statistics and correlations. Mean scores were 3.630 for DC, 3.619 for SCI, 3.258 for LCP and 3.948 for SCR. The strongest bivariate association was between SCI and SCR (r = 0.513); correlations among the four constructs ranged from 0.250 to 0.513.

Table 2. Descriptive statistics and correlations.

ConstructMSD1234
1. Digital capability (DC)3.6300.7381
2. Food supply chain integration (SCI)3.6190.6380.4751
3. Low-carbon-oriented and resource-saving food practices index (LCP)3.2580.5710.2500.3151
4. Enterprise-level supply chain resilience (SCR)3.9480.7110.4450.5130.2511

DC, digital capability; SCI, food supply chain integration; LCP, low-carbon-oriented and resource-saving food practices index; SCR, enterprise-level supply chain resilience; M, mean; SD, standard deviation.

5.2 Competitive CFA and common method bias diagnosis

The hypothesized five-factor reflective measurement model fitted the data well (χ2 = 226.823, df = 199, CFI = 0.994, TLI = 0.992, RMSEA = 0.017, SRMR = 0.029). Both the three-factor alternative and the single-factor model fitted substantially worse (Table 3), supporting the distinction among DC, the three first-order integration dimensions and SCR.

Table 3. Competitive CFA results.

Modelχ2dfχ2/dfCFITLIRMSEASRMR
Five-factor measurement model226.8231991.1400.9940.9920.0170.029
Three-factor alternative model829.5602064.0270.8540.8370.0810.063
Single-factor model1719.7002098.2280.6470.6100.1250.103

df, degrees of freedom; CFI, comparative fit index; TLI, Tucker–Lewis index; RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.

The single-factor model reproduced the covariance structure substantially less well than the theoretical measurement model (ΔCFI = 0.347, ΔRMSEA = 0.108, ΔSRMR = 0.074). Because all constructs were operator-reported, the single-factor comparison cannot exclude common-method bias. Structural coefficients are interpreted as statistical associations.

5.3 Reliability, convergent validity and discriminant validity

All standardized loadings were positive and statistically significant, ranging from 0.610 to 0.783. Cronbach's α and CR ranged from 0.822 to 0.845, and AVE ranged from 0.510 to 0.557 (Table 4). No prespecified reflective item was removed after model estimation.

Table 4. Reliability and convergent validity.

ConstructItemsLoading rangeCronbach’s αCRAVE
DC (digital capability)50.610–0.7830.8360.8380.510
II (internal integration)40.667–0.7710.8290.8310.551
SI (supplier integration)40.697–0.7740.8330.8340.557
CI (customer integration)40.674–0.7780.8220.8230.539
SCR (enterprise-level supply chain resilience)50.679–0.7630.8450.8450.523

CR, composite reliability; AVE, average variance extracted; II, internal integration; SI, supplier integration; CI, customer integration.

HTMT values ranged from 0.419 to 0.611 and remained below 0.85 (Table 5). Together with the competitive CFA, these results support empirical separation of DC, internal integration, supplier integration, customer integration and SCR.

Table 5. HTMT discriminant-validity matrix.

ConstructDCIISICISCR
DC (digital capability)—
II (internal integration)0.419—
SI (supplier integration)0.4890.565—
CI (customer integration)0.4750.6110.565—
SCR (enterprise-level supply chain resilience)0.5300.4910.4550.544—

HTMT, heterotrait–monotrait ratio; DC, digital capability; II, internal integration; SI, supplier integration; CI, customer integration; SCR, enterprise-level supply chain resilience.

The standardized second-order SCI loadings were 0.755 for internal integration, 0.739 for supplier integration and 0.784 for customer integration; the two freely estimated loadings were statistically significant (both p < 0.001).

5.4 Low-carbon-oriented and resource-saving food practices index

The three LCP dimensions—sustainable procurement, resource-efficient processing, and food-loss and circular management—received equal weights. Their VIFs were 1.676, 1.672 and 1.645, respectively.

Item-level applicability ranged from 86.3% to 91.4%. Mean applicable-item counts were 3.580 for sustainable procurement, 3.544 for resource-efficient processing and 3.593 for food-loss and circular management. The corresponding dimension means were 3.228, 3.283 and 3.263. Item-level response distributions, dimension-level applicability and the complete distribution of applicable-item counts are reported in Supplementary Tables S1 and S2.

The six-path model was re-estimated using the equal-weight dimension score, the applicable-item mean and a score that recoded non-applicable responses as 1. The two alternative scores correlated 0.996 and 0.923 with the primary score. Across all three specifications, the LCP → SCR path remained statistically non-significant, while the direction and significance of SCI → LCP, SCI → SCR and DC → SCR were unchanged. Supplementary Table S2 reports the complete sensitivity estimates.

The external operating-check composite averaged six standardized self-report indicators: local and seasonal sourcing shares, online booking/sales revenue share, reverse-coded reported food-loss rate, resource-use record keeping and portion-choice availability. In an ordinary least squares (OLS) regression, the three LCP dimensions explained 0.4% of its variance (R2 = 0.004; F(3, 463) = 0.687, p = 0.561). None of the dimension coefficients was statistically significant; their estimates are reported in Supplementary Table S3. The operating-check coefficients were not used as LCP weights. Because the checks do not constitute a content-equivalent global criterion, the weak associations provide limited external convergent evidence.

5.5 Structural model and direct effects

The six-path structural model fitted the data well (χ2 = 256.569, df = 222, CFI = 0.992, TLI = 0.991, RMSEA = 0.018, SRMR = 0.032) and explained 37.4% of the variance in SCI, 13.9% in LCP and 45.8% in SCR. DC was positively associated with SCI and SCR; SCI was positively associated with LCP and SCR. The DC → LCP and LCP → SCR paths were not statistically significant (Table 6).

Table 6. CB-SEM structural path estimates and hypothesis tests.

HypothesisPathBSEStandardized βzpDecision
H1DC → SCI0.4940.0570.6128.640<0.001Supported
H2SCI → LCP0.3240.0800.3064.073<0.001Supported
Direct pathDC → LCP0.0820.0590.0961.4050.160Not significant
H3SCI → SCR0.5700.0940.5156.069<0.001Supported
H4LCP → SCR0.0230.0480.0220.4690.639Not supported
H5DC → SCR0.1920.0600.2153.1820.001Supported

B, unstandardized coefficient; SE, standard error; β, standardized coefficient; DC, digital capability; SCI, food supply chain integration; LCP, low-carbon-oriented and resource-saving food practices index; SCR, enterprise-level supply chain resilience.

5.6 Specific indirect effects

The six-path latent-variable SEM was re-estimated in 5,000 successful case-level bootstrap resamples. Percentile confidence intervals supported the indirect association from DC to LCP through SCI (estimate = 0.160, 95% CI [0.087, 0.253]) and the indirect association from DC to SCR through SCI (estimate = 0.282, 95% CI [0.195, 0.392]). The H7 serial indirect association through SCI and LCP was not supported (estimate = 0.004, 95% CI [−0.014, 0.020]). Table 7 reports direct, indirect and total effects.

Table 7. Bootstrap direct, indirect and total effects.

EffectPoint estimateBoot SE95% CISuccessful/failed runsDecision
H6: DC → SCI → LCP0.1600.042[0.087, 0.253]5,000/0Supported
DC → LCP (direct)0.0820.062[−0.040, 0.202]5,000/0Not significant
DC → LCP (total)0.2420.045[0.155, 0.335]5,000/0Significant
DC → SCI → SCR0.2820.051[0.195, 0.392]5,000/0Significant
DC → LCP → SCR0.0020.005[−0.007, 0.015]5,000/0Not significant
H7: DC → SCI → LCP → SCR0.0040.008[−0.014, 0.020]5,000/0Not supported
DC → SCR (direct)0.1920.056[0.079, 0.300]5,000/0Significant
DC → SCR (total)0.4800.048[0.387, 0.575]5,000/0Significant

Boot SE, bootstrap standard error; CI, confidence interval; DC, digital capability; SCI, food supply chain integration; LCP, low-carbon-oriented and resource-saving food practices index; SCR, enterprise-level supply chain resilience.

5.7 Ordinal-estimator sensitivity analysis

The ordinal-item sensitivity analysis used WLSMV in lavaan 0.7–2. The 22 reflective items were treated as ordered; LCP remained a continuous observed composite. The model used the delta parameterization, marker-variable identification, robust standard errors and a mean- and variance-adjusted test statistic. The same six direct paths were estimated, with no correlated residuals. Supplementary Table S4 provides the complete settings and estimates.

The scaled WLSMV fit statistics were χ2 = 230.210, df = 222, CFI = 0.9989, TLI = 0.9988, RMSEA = 0.0089 and SRMR = 0.0324. These favorable values are reported as estimator-specific fit information rather than independent evidence of robustness.

The WLSMV analysis retained the MLW significance pattern: DC → SCI, SCI → LCP, SCI → SCR and DC → SCR were statistically significant, whereas DC → LCP and LCP → SCR were not. Complete unstandardized and standardized estimates, robust standard errors, z statistics and p-values are reported in Supplementary Table S4.

5.8 Public UGC text analysis results

The audited retrieval corpus contains 18,565 records: Zhihu 9,117, Bilibili 6,022, Weibo 2,043 and Xiaohongshu 1,383. It includes 16,381 comments and 2,184 body or video-description texts (Supplementary Table S7). Timestamp checks identified records published before August 2023; the corpus is therefore not treated as a complete 2023–2026 publication window. Platform composition reflects purposive retrieval rather than the population distribution of rural food tourism discourse.

The adjudicated sample contained 656 records. Only the resilience classifier met the prespecified F1/AP rule (F1 = 0.719; AP = 0.868). F1 values for digital capability, integration, low-carbon-related practices and constraints were 0.517, 0.538, 0.298 and 0.513, respectively; these labels were restricted to exploratory semantic retrieval (Table 8).

Table 8. Multi-label classifier validation.

LabelPositivePrecisionRecallF1ROC-AUCAPUse
Digital capability290.5170.5170.5170.9580.602Exploratory semantic retrieval
Food supply chain integration280.5830.5000.5380.9260.562Exploratory semantic retrieval
Low-carbon-oriented practices260.3330.2690.2980.8750.273Exploratory semantic retrieval
Supply-chain resilience (text label)780.8200.6410.7190.9780.868Model-predicted distribution
Operational constraints700.4560.5860.5130.9040.541Exploratory semantic retrieval

F1, harmonic mean of precision and recall; ROC–AUC, area under the receiver operating characteristic curve; AP, average precision.

After refitting on all adjudicated records, the resilience classifier identified 389 of 18,565 retrieved texts at the 0.50threshold (2.10%). The predicted count varied from 1,287 at 0.30 to 197 at 0.70. The model-predicted share is threshold-dependent and uncalibrated and is not an estimate of population prevalence.

Low inter-coder agreement limits interpretation even for resilience (κ = 0.151). Classifier accuracy against adjudicated labels and agreement between independent coders measure different sources of uncertainty. The predictions describe the retained coding scheme; they do not establish the validity of enterprise-level constructs.

The five-topic NMF solution assigned 1,914 eligible non-comment texts to rural development and agri-tourism (n = 424), entrepreneurship and household roles (n = 683), homestay operations (n = 207), agrifood e-commerce and packaging (n = 422), and agritainment pressures and consumption (n = 178). Table 9 reports topic terms and interpretations.

Table 9. Topic interpretation based on public UGC texts.

TopicInterpretationTop words (English)n
1Rural development and agri-tourismrural, development, ecology, industry, revitalization, agriculture, construction, cultural tourism, green, sustainability424
2Entrepreneurship and household rolesentrepreneurship, women, rural, entrepreneurs, life, work, children, family, men683
3Homestay operationshomestay, rural, hotel, guests, tourism, industry, visitors, operations, occupancy, design207
4Agrifood e-commerce and packagingagricultural products, e-commerce, packaging, design, logistics, rural, brands, unsold stock, platforms422
5Agritainment pressures and consumptionagritainment, environmental protection, owners, experience, consumption, catering, business, consumers, closures178

The topic solution was stable across five random initializations (adjusted Rand index = 0.851–0.868), whereas agreement between the primary and alternative-preprocessing assignments was lower (adjusted Rand index = 0.659). Topic diversity was 0.920 and top-term document co-occurrence normalized pointwise mutual information was 0.239. Residual general entrepreneurship and lifestyle content limits food-tourism-specific interpretation.

Figure 2 presents topic–platform–retrieval–stratum co-occurrences in the 1,914 topic-model texts.

Public-discourse linkages among topics, platforms and retrieval strata. Ribbon widths are proportional to text counts and use the same scale across all three columns (*n* = 1,914). Colors identify the five NMF topics reported in Table 9; numbers beside nodes give marginal counts. Retrieval strata identify the keyword groups used to collect texts, not classifier-predicted practices. The diagram depicts discourse co-occurrence, not temporal flows or enterprise-level statistical associations.
Figure 2. Public-discourse linkages among topics, platforms and retrieval strata. Ribbon widths are proportional to text counts and use the same scale across all three columns (n = 1,914). Colors identify the five NMF topics reported in Table 9; numbers beside nodes give marginal counts. Retrieval strata identify the keyword groups used to collect texts, not classifier-predicted practices. The diagram depicts discourse co-occurrence, not temporal flows or enterprise-level statistical associations.

6 Discussion

6.1 Main findings

The SEM estimates supported H1, H2, H3, H5 and H6, but not H4 or H7. Digital capability was associated with food supply chain integration and enterprise-level resilience, while integration was associated with both the LCP index and resilience. The LCP–SCR coefficient and the DC → SCI → LCP → SCR serial indirect association were not statistically significant. The public UGC analysis characterized platform discourse separately.

The positive DC–SCI association (β = 0.612) is consistent with the information-processing role assigned to digital capability. In the sampled enterprises, digital capability connected booking, customer-feedback, procurement, ingredient-quality and service information across operating activities. Such information becomes organizationally relevant when it enters procurement planning, supplier communication and internal workflow coordination. This coefficient pattern accords with organizational information processing theory, which distinguishes information-processing capacity from the coordination arrangements through which information is used.

Food supply chain integration was associated with both the LCP index (β = 0.306) and enterprise-level supply chain resilience (β = 0.515). The first association corresponds to coordination of purchasing quantities, seasonal menus, preparation schedules and food-loss management. The second corresponds to information sharing and joint adjustment when ingredient shortages, delivery changes or demand fluctuations occur. The model did not separately estimate the contributions of coordination, inventory redundancy and supplier diversification; the SCI–SCR coefficient therefore represents an overall statistical association rather than the effect of a single resilience mechanism.

The LCP index did not retain a statistically significant independent association with enterprise-level resilience after digital capability and food supply chain integration were included. The estimate does not support H4 within the specified model. Resource-saving routines concern procurement, processing efficiency and food-loss management, whereas disruption response also depends on alternative supply, operational reconfiguration and recovery capacity. The unsupported H7 estimate likewise provides no evidence for a statistical sequence in which digital capability is associated with resilience through both supply chain integration and the LCP index.

Only the resilience classifier met the prespecified reporting threshold. Low inter-coder agreement and threshold sensitivity nevertheless limit the interpretation of its model-predicted distribution. The remaining labels and the NMF topics provide exploratory descriptions of platform discourse rather than corroboration of the enterprise-level SEM estimates.

6.2 Theoretical contributions

Research on agri-food digitalization associates digital technologies with green integration, sustainable performance and resilience but often places technological capability, interorganizational coordination and operating outcomes within a broad transformation process (Belhadi et al., 2024; Taghikhah et al., 2025; Yadav et al., 2024). In the estimated model, DC was associated with SCI, whereas SCI retained separate associations with both LCP and SCR. The significant DC → SCI → LCP indirect association, combined with the non-significant DC → LCP coefficient, distinguishes information-processing capacity from the cross-boundary arrangements through which operating information enters coordinated action.

Sustainability and resilience research increasingly treats resource efficiency, adaptation and recovery as complementary attributes of food systems (Jia et al., 2024; Paredes-Rodríguez et al., 2024; Yuan et al., 2024). SCI is associated with both LCP and SCR in the present model, but neither the LCP → SCR coefficient nor the DC → SCI → LCP → SCR serial indirect association is statistically significant. A shared organizational correlate therefore does not establish a sequential association between resource-saving practices and resilience. The evidence qualifies a linear interpretation in which procurement, processing efficiency and food-loss routines necessarily correspond to alternative supply, rapid reconfiguration and recovery capacity.

Rural tourism entrepreneurship research has primarily examined entrepreneurial entry, empowerment, role transition and local embeddedness (Martini et al., 2020; Nordbø, 2022; Tao et al., 2025). Rural food tourism ventures also coordinate agricultural supply, food processing, hospitality services and tourist consumption. Positioning these ventures as supply-chain actors connects tourism entrepreneurship with short food supply chains and agri-food digitalization. The all-women sample sets the empirical boundary; gender-specific differences require direct comparison with male-led and mixed-gender ventures.

6.3 Practical implications

Digitalization programs for rural food tourism operators could prioritize the integration of digital tools with procurement, inventory, menu planning and customer feedback. Training should cover platform content production, online booking management, customer feedback analysis, electronic procurement records and inventory-menu linkage, instead of focusing only on account opening or live-streaming sales. Low-carbon-oriented policy support should provide operational toolkits tailored to food-service settings for seasonal menu design, local procurement coordination, portion management, packaging reduction, kitchen-residue recording and recycling. Resource-saving routines should be implemented for their operational and environmental objectives rather than treated as substitutes for supplier diversification, demand visibility or contingency coordination. Local governments, platforms and industry organizations can include supplier alternatives, demand-information sharing and contingency coordination in support programs for enterprises meeting the sample conditions. Platform governance can also improve the visibility of high-quality rural food tourism products through local food labels, sustainability labels for food tourism enterprises, dedicated rural food marketplace sections and credible review systems.

6.4 Limitations and future research

The sample comprises women operators recruited through established contacts in six Shandong cities, with prior digital-tool use and supplier relationships required for inclusion. These conditions limit generalization to digitally inactive enterprises, firms without established supplier relationships and enterprises outside the recruitment channels or region. Comparisons with male-led and mixed-gender enterprises are needed to assess gender differences. Four-wave measurement separates reporting occasions, but operator self-report and the observational design leave common-method bias and causal ambiguity unresolved; independently recorded transactions and longitudinal interventions would strengthen inference. LCP represents self-reported practices, and its weak associations with heterogeneous operating-check indicators provide limited external convergent evidence; future validation should incorporate measured energy use, food loss and emissions. Public UGC was purposively retrieved rather than sampled from a complete annual archive. Low coder agreement, imbalanced labels, threshold sensitivity and residual content beyond rural food tourism restrict its interpretation; independent domain-relevance coding and calibrated external validation are needed before estimating discourse prevalence.

7 Conclusion

The four-wave survey showed that digital capability was associated with food supply chain integration and enterprise-level resilience, while integration was associated with both the LCP index and resilience. Neither the independent LCP–SCR association nor the DC → SCI → LCP → SCR serial indirect association was statistically supported. Resource-saving operations and disruption resilience therefore shared an association with supply chain integration but did not constitute an empirically supported linear sequence. Public-text results characterized platform discourse independently of the enterprise-level SEM estimates. Digital support for rural food tourism ventures can connect booking, procurement, inventory and supplier information, while resilience planning requires supply alternatives, demand visibility and contingency coordination. These conclusions apply to women operators from six Shandong cities whose ventures already used digital tools and maintained supplier relationships. Replication with probability-based, gender-comparative and geographically diverse samples, together with objective measures of energy use, food loss and emissions, is required to assess wider applicability.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. OpenAI’s ChatGPT was used for English-language editing and stylistic revision of the manuscript and for reviewing the reproducibility documentation associated with the analysis and plotting code. The statistical analyses and figures were produced using the software and procedures reported in the manuscript, not by generative AI. The authors independently reran and inspected all analysis and plotting code and verified all numerical results, figures, references, and final manuscript text.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fsufs.2026.1976172/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Bing He, Jiangsu Ocean University, China

Reviewed by: Tomo Kawane, Hiroshima University, Japan

Han Zhang, Qingdao University, China

Statements

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The author(s) declared that financial support was received for this work and/or its publication. This work was supported by Motivation Management and Value Pursuit among Women's Rural Tourism Entrepreneurs (2022GSPSJ04).

Data availability

The de-identified, processed questionnaire dataset and analytical materials supporting the findings of this study are available from the corresponding author upon reasonable request. Full raw survey export cannot be shared externally due to platform service agreements and participant-privacy constraints. Data sharing is subject to participant confidentiality, platform terms of service, and applicable data-protection requirements. Requests to access the datasets should be directed to LW, liwei@sdwu.edu.cn.

Ethics approval

Ethical approval and formal ethics‑committee waiver were not obtained prior to data collection. This study conducted an anonymous, non‑interventional online questionnaire targeting adult rural food‑tourism operators. No sensitive personal or identifiable information was collected; the research posed minimal risk to participants and complied with the principles of the Declaration of Helsinki. Before starting the survey, respondents read full study briefings on the opening page and gave documented electronic informed consent by actively selecting “I agree to participate”. Participation was fully voluntary. No separate paper‑based informed‑consent forms were applied. Our institutional ethics committee does not review low‑risk social‑science questionnaire studies, hence no official waiver document can be issued. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

WL: Data curation, Investigation, Writing – original draft, Writing – review & editing. PY: Supervision, Writing – review & editing. RZ: Investigation, Software, Writing – original draft.

Acknowledgements

The authors thank the rural tourism operators who participated in the questionnaire survey and the training organizers who assisted with field access.

References

  1. Agyeiwaah E., Zhao Y. (2024). Residents' perceived social sustainability of food tourism events. Tour. Manag. Perspect. 53: 101276. doi: doi:10.1016/j.tmp.2024.101276
  2. Ali A., Iqbal Z., Ali I. (2025). Women in mountain tourism: exploring the links between women tourism entrepreneurship and women empowerment in Hunza Valley. Tour. Recreat. Res. 50, 1391 – 1407. doi: doi:10.1080/02508281.2024.2386869
  3. Ambulkar S., Blackhurst J., Grawe S. (2015). Firm's resilience to supply chain disruptions: scale development and empirical examination. J. Oper. Manag. 33–34, 111 – 122. doi: doi:10.1016/j.jom.2014.11.002
  4. Anderson J. C., Gerbing D. W. (1988). Structural equation modeling in practice: a review and recommended two-step approach. Psychol. Bull. 103, 411 – 423. doi: doi:10.1037/0033-2909.103.3.411
  5. Belhadi A., Kamble S. S., Subramanian N., Singh R. K., Venkatesh M. (2024). Digital capabilities to manage agri-food supply chain uncertainties and build supply chain resilience during compounding geopolitical disruptions. Int. J. Oper. Prod. Manag. 44, 1914 – 1950. doi: doi:10.1108/IJOPM-11-2022-0737
  6. Diamantopoulos A., Winklhofer H. M. (2001). Index construction with formative indicators: an alternative to scale development. J. Mark. Res. 38, 269 – 277. doi: doi:10.1509/jmkr.38.2.269.18845
  7. Flora D. B., Curran P. J. (2004). An empirical evaluation of alternative methods of estimation for confirmatory factor analysis with ordinal data. Psychol. Methods 9, 466 – 491. doi:,. doi:10.1037/1082-989X.9.4.466
  8. Flynn B. B., Huo B., Zhao X. (2010). The impact of supply chain integration on performance: a contingency and configuration approach. J. Oper. Manag. 28, 58 – 71. doi: doi:10.1016/j.jom.2009.06.001
  9. Fornell C., Larcker D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res. 18, 39 – 50. doi: doi:10.1177/002224378101800104
  10. Galbraith J. R. (1974). Organization design: an information processing view. Interfaces 4, 28 – 36. doi: doi:10.1287/inte.4.3.28
  11. Han Z., Huo B. (2020). The impact of green supply chain integration on sustainable performance. Ind. Manag. Data Syst. 120, 657 – 674. doi: doi:10.1108/IMDS-07-2019-0373
  12. Hayes A. F., Scharkow M. (2013). The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis: does method really matter? Psychol. Sci. 24, 1918 – 1927. doi:,. doi:10.1177/0956797613480187
  13. Henseler J., Ringle C. M., Sarstedt M. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Mark. Sci. 43, 115 – 135. doi: doi:10.1007/s11747-014-0403-8
  14. Hu L.-T., Bentler P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: conventional criteria versus new alternatives. Struct. Equ. Model. A. Multidiscip. J. 6, 1 – 55. doi: doi:10.1080/10705519909540118
  15. Jantapoon K., Amornwattahcharoenchai P. (2025). Integrating sustainability and resilience in agri-food supply chains: a knowledge-based view of Thailand's pineapple industry in the digital era. Sustain. Futur. 10: 101264. doi: doi:10.1016/j.sftr.2025.101264
  16. Jarvis C. B., MacKenzie S. B., Podsakoff P. M. (2003). A critical review of construct indicators and measurement model misspecification in marketing and consumer research. J. Consum. Res. 30, 199 – 218. doi: doi:10.1086/376806
  17. Jia F., Shahzadi G., Bourlakis M., John A. (2024). Promoting resilient and sustainable food systems: a systematic literature review on short food supply chains. J. Clean. Prod. 435: 140364. doi: doi:10.1016/j.jclepro.2023.140364
  18. Khazami N., Nefzi A., Yahyaoui A. (2024). Role of rural women on the agritourism entrepreneurial behavior in Tunisia. Cogent Bus. Manag. 11: 2292313. doi: doi:10.1080/23311975.2023.2292313
  19. Khin S., Ho T. C. F. (2019). Digital technology, digital capability and organizational performance: a mediating role of digital innovation. Int. J. Innov. Sci. 11, 177 – 195. doi: doi:10.1108/IJIS-08-2018-0083
  20. Kline R. B. (2016). Principles and Practice of Structural Equation Modeling. 4th ed. New York, NY: Guilford Press.
  21. León-Bravo V., Borrello B., Ciccullo F., Caniato F. (2025). Unpacking proximity for sustainability in short food supply chains. Bus. Strateg. Environ. 34, 1792 – 1809. doi: doi:10.1002/bse.4079
  22. Li C.-H. (2016). Confirmatory factor analysis with ordinal data: comparing robust maximum likelihood and diagonally weighted least squares. Behav. Res. Methods 48, 936 – 949. doi: doi:10.3758/s13428-015-0619-7
  23. Martini U., Malacarne K., Pederzolli Giovanazzi S., Buffa F. (2020). Sustainable tourism development in rural and marginal areas and opportunities for female entrepreneurship: lessons from an exploratory study. Worldw. Hosp. Tour. Themes 12, 421 – 430. doi: doi:10.1108/WHATT-05-2020-0023
  24. Maynard D. C., Zandonadi R. P., Nakano E. Y., Botelho R. B. A. (2020). Sustainability indicators in restaurants: the development of a checklist. Sustainability 12: 4076. doi: doi:10.3390/su12104076
  25. Nordbø I. (2022). Female entrepreneurs and path-dependency in rural tourism. J. Rural. Stud. 96, 198 – 206. doi: doi:10.1016/j.jrurstud.2022.09.032
  26. Orea-Giner A., Fusté-Forné F., Todd L., Park E. (2026). The nexus between local food tourism events and social sustainability: future prospects and priorities. Tour. Manag. Perspect. 60: 101430. doi: doi:10.1016/j.tmp.2025.101430
  27. Paredes-Rodríguez A. M., Orejuela-Cabrera J. P., Osorio-Gómez J. C. (2024). Integrating sustainability and resilience in agri-food supply chains. Int. J. Sustain. Eng. 17, 1122 – 1138. doi: doi:10.1080/19397038.2024.2430514
  28. Podsakoff P. M., MacKenzie S. B., Lee J.-Y., Podsakoff N. P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. J. Appl. Psychol. 88, 879 – 903. doi:,. doi:10.1037/0021-9010.88.5.879
  29. Preacher K. J., Hayes A. F. (2008). Asymptotic and resampling strategies for assessing and comparing indirect effects in multiple mediator models. Behav. Res. Methods 40, 879 – 891. doi: doi:10.3758/BRM.40.3.879
  30. Raykov T. (1997). Estimation of composite reliability for congeneric measures. Appl. Psychol. Meas. 21, 173 – 184. doi: doi:10.1177/01466216970212006
  31. Reardon J., Way K. A., Garrison M. E. B. (2024). Food waste in restaurants: a qualitative investigation of chefs' practices. Food Humanit. 3: 100405. doi: doi:10.1016/j.foohum.2024.100405
  32. Richardson H. A., Simmering M. J., Sturman M. C. (2009). A tale of three perspectives: examining post hoc statistical techniques for detection and correction of common method variance. Organ. Res. Methods 12, 762 – 800. doi: doi:10.1177/1094428109332834
  33. Sciortino C., Giamporcaro G., Sgroi F., Modica F. (2025). Exploring the role of short food supply chains in agricultural sustainability and resilience: a literature review. Agric. Food Econ. 13: 75. doi: doi:10.1186/s40100-025-00420-3
  34. Stein A. J., Santini F. (2022). The sustainability of "local" food: a review for policy-makers. Rev. Agric., Food Environ. Stud. 103, 77 – 89. doi:,. doi:10.1007/s41130-021-00148-w
  35. Taghikhah F. R., Prior D. D., Hafezi R., Baker D., Matous P. (2025). Understanding digital capabilities and their impacts on Australian agri-food supply chain resilience: engineering vs. socio-ecological thinking. Technol. Forecast. Soc. Chang. 218: 124191. doi: doi:10.1016/j.techfore.2025.124191
  36. Tao H., Zhang M., Wang J. (2025). "She" power: role transition and self-empowerment of women in rural tourism. J. Sustain. Tour. 33, 1991 – 2009. doi: doi:10.1080/09669582.2024.2418358
  37. Toniolo S., Russo I., Bravo I. (2024). Integrating product-focused life cycle perspectives in the fresh food supply chain: revealing intra- and inter-organizational views. Sustain. Prod. Consum. 48, 46 – 61. doi: doi:10.1016/j.spc.2024.05.009
  38. Tushman M. L., Nadler D. A. (1978). Information processing as an integrating concept in organizational design. Acad. Manag. Rev. 3, 613 – 624. doi: doi:10.5465/amr.1978.4305791
  39. Vahdanjoo M., Grøn Sørensen C., Nørremark M. (2025). Digital transformation of the agri-food system. Curr. Opin. Food Sci. 63: 101287. doi: doi:10.1016/j.cofs.2025.101287
  40. Wan C., Lee D., Wut T. M., Banerjee R. (2025). Tourists' local food consumption: travel experience, responsible tourism, or both? Sustain. Futur. 9: 100663. doi: doi:10.1016/j.sftr.2025.100663
  41. Wang G., Li S., Yi Y., Wang Y., Shin C. (2024). Digital technology increases the sustainability of cross-border agro-food supply chains: a review. Agriculture 14: 900. doi: doi:10.3390/agriculture14060900
  42. Yadav S., Samadhiya A., Kumar A., Luthra S., Kumar V., Garza-Reyes J. A.. (2024). The interplay effects of digital technologies, green integration, and green innovation on food supply chain sustainable performance: an organizational information processing theory perspective. Technol. Soc. 77: 102585. doi: doi:10.1016/j.techsoc.2024.102585
  43. Yuan M., Hu H., Xue M., Li J. (2024). Framework for resilience strategies in agricultural supply chain: assessment in the era of climate change. Front. Sustain. Food Syst. 8: 1444910. doi: doi:10.3389/fsufs.2024.1444910
  44. Zhong S., Zheng C., Fang P. (2026). Transformative capacity for resilience in short food supply chains: the dynamic matching mechanism of sustainable restaurants in China. Int. J. Hosp. Manag. 138: 104714. doi: doi:10.1016/j.ijhm.2026.104714
  45. Zhu Q., Sarkis J., Lai K.-H. (2008). Confirmation of a measurement model for green supply chain management practices implementation. Int. J. Prod. Econ. 111, 261 – 273. doi: doi:10.1016/j.ijpe.2006.11.029

Citation

Li, W., et al. (2026). Digital capability and supply chain resilience in rural food tourism ventures: the role of food supply chain integration. Frontiers in Sustainable Food Systems, Volume 10. https://doi.org/
Received
Accepted
Published
Volume
10 - 2026

Copyright

Copyright © 2026 Li, Yin and Zhou.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: Peng Yin

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

Original ResearchMaking inequities visible: a call for participatory food system dashboards for SDG monitoringRositsa T. Ilieva, Yvette Ng, Nevin CohenReviewAvocado (Persea americana Mill) seed starch: composition, extraction, and applications—a comprehensive reviewNehal Chavda, Rajdip Khaniya, Margi Gokalani, Ajay SinghBrief Research ReportComparison of food carbon footprints using 20- and 100-year timeframesKatie Osborn, Silas Crocker, Kira Presley, Alexis Shenkiryk, Maggie Spire, Jennifer Molidor, Mansha Kapur, Jennifer A. Jay