Primary data were collected through structured surveys conducted across 11 mountain villages: Bogë, Çerem, Dedaj, Dragobi, Lekbibaj, Lepushë, Reç, Selcë, Theth, Valbonë, and Vermosh. The sample includes 82 households, 107 businesses, and 310 working-age individuals engaged in agricultural and tourism activities. Several households operated multiple economic activities, including agricultural and tourism-related businesses. To avoid duplication of observations and ensure consistency across survey waves, business-level information was aggregated to the household level. Consequently, all econometric models were estimated at the household level, with households serving as the panel units observed during the 2023–2025 study period. Data collection covered agricultural production, biodiversity-related activities, tourism participation, economic performance, migration intentions, and socio-demographic characteristics. A stratified sampling approach was applied. Villages were selected for their importance to agrobiodiversity and rural tourism, while households were randomly selected using fixed quotas for agricultural and tourism businesses. The sample represents approximately 8% of households in the CABRA project area and captures major mountain communities characterized by strong integration of agriculture and tourism. Although geographically concentrated, the dataset provides detailed longitudinal evidence from some of the most important agrobiodiversity regions in Northern Albania.

To operationalize multidimensional rural development concepts, five composite indices were developed using standardized z-scores to ensure comparability across variables measured on different scales. Standardization was performed according to the formula:

where Xij represents the value of indicator j for observation i, Xj is the sample mean, and SD(Xj) is the standard deviation of indicator j. Composite indices were calculated as the arithmetic mean of the standardized indicators comprising each construct. Equal weighting was applied because no strong theoretical or empirical justification existed for assigning differential weights. Higher index values indicate stronger performance in the corresponding dimension. Reliability tests confirmed acceptable internal consistency (Cronbach’s α > 0.70).

All models are estimated at the household level with year fixed effects and household-clustered standard errors. Model diagnostics indicated no evidence of problematic multicollinearity (VIF = 1.00–1.69). Breusch–Pagan tests revealed significant heteroskedasticity (BP = 17.57–52.04, p < 0.01), justifying the use of household-clustered robust standard errors.

Direct Effects Model: The baseline model estimates the relationship between agrobiodiversity and rural resilience:

where Xit includes gender, education, business type, access to training, and additional income sources.

Mediation Analysis: The mediation framework evaluates whether farm culture and biodiversity communication mediate the relationship between agrobiodiversity and rural resilience. The mediation analysis followed the causal mediation framework proposed by Preacher and Hayes (2008). Indirect effects were estimated using bias-corrected bootstrapped confidence intervals based on 5,000 replications.

Indirect effects were estimated using bootstrapped confidence intervals with 5,000 replications.

Moderation Analysis: The moderating role of mountain rural heritage was estimated through an interaction model:

A positive interaction coefficient indicates that heritage assets strengthen the relationship between agrobiodiversity and rural resilience.

Migration Model: Migration intentions were estimated using Firth’s penalized logistic regression due to the low incidence of emigration intentions:

where MIit indicates whether household members considered permanent migration abroad.

To contextualize the findings regionally, the analysis incorporates comparative benchmarks from Albania, Serbia, Bosnia and Herzegovina, Montenegro, and North Macedonia.

These countries share similar characteristics, including fragmented agriculture, rural depopulation, and increasing reliance on rural tourism.

Households with higher Agrobiodiversity Index (ABI) values report significantly stronger economic performance. High-ABI households achieve higher sales, income, and investment levels, suggesting that diversified agricultural systems improve economic resilience through broader production structures and value-added activities.

The direct-effects model confirms a significant positive relationship between agrobiodiversity and rural resilience:

The ABI coefficient (β=0.38, p<0.001) indicates that greater agricultural diversification is associated with stronger resilience outcomes.

The mediation analysis demonstrates that Farm Culture (FCI) and Biodiversity Communication (BCI) partially mediate the relationship between agrobiodiversity and rural resilience. Agrobiodiversity positively influences both the transmission of cultural knowledge and biodiversity communication practices, thereby strengthening resilience outcomes.

The inclusion of FCI and BCI reduces the direct ABI effect, indicating partial mediation. The findings suggest that agrobiodiversity generates value not only through diversification itself but also through cultural practices and visitor-oriented communication about biodiversity.

Mountain Rural Heritage (MRH) significantly moderates the relationship between agrobiodiversity and resilience. The interaction term between ABI and MRH remains positive and statistically significant.

Households with stronger heritage assets, including traditional accommodations and tourism infrastructure, obtain substantially greater resilience benefits from agrobiodiversity activities.

The logistic regression results indicate that rural resilience and tourism engagement are associated with lower migration intentions.

Higher resilience significantly reduces intentions to migrate, while tourism participation is associated with lower out-migration pressures. Although agrobiodiversity does not have a direct, significant effect on migration intentions, its indirect effect through rural resilience remains important. Overall, the findings support the argument that integrated agrobiodiversity–tourism systems contribute to stronger territorial resilience in mountain regions.

Comparative analysis indicates that Northern Albania shares several structural characteristics with other rural economies in the Western Balkans, including fragmented agriculture, rural depopulation, and increasing dependence on tourism as a driver of diversification. However, the findings also reveal stronger agriculture–tourism integration and higher biodiversity communication compared with regional benchmarks.

The results suggest that Northern Albania exhibits relatively stronger biodiversity communication and tourism integration than many other comparable regions in the Western Balkans. Approximately 37% of tourism businesses source agricultural products locally, indicating stronger territorial linkages between agriculture and tourism activities. Migration intentions also remain below regional averages.

At the same time, persistent financing constraints, weak infrastructure, and limited access to credit continue to restrict broader rural investment and diversification. The comparative evidence suggests that Northern Albania shares several structural characteristics with other Western Balkan rural regions, including fragmented agriculture, demographic pressures, and increasing dependence on tourism diversification.

The stronger integration of agriculture, tourism, and biodiversity-related activities observed in the study area indicates a comparatively advanced stage of rural diversification. The findings position Northern Albania as a transitional mountain economy in which agrobiodiversity, tourism, and heritage increasingly serve as interconnected mechanisms for rural development.

The findings provide strong empirical evidence that agrobiodiversity is positively associated with rural resilience in mountain regions. Households with higher levels of agricultural diversification report stronger economic performance, higher investment activity, and greater adaptive capacity. These results are consistent with previous studies emphasizing diversification as a key mechanism that supports resilience under environmental and economic pressures (Lin 2011; Darnhofer et al. 2016; Wezel et al. 2014).

The mediation analysis further demonstrates that farm culture and biodiversity communication function as important socio-cultural mechanisms linking agrobiodiversity with resilience outcomes.

Traditional knowledge systems, local production practices, and biodiversity storytelling appear particularly relevant for value creation and territorial differentiation, supporting earlier findings on the importance of cultural capital and experiential rural tourism (Berkes et al. 2000; Tew and Barbieri 2012).

The moderating role of mountain rural heritage highlights the importance of broader territorial systems in shaping rural development outcomes. The resilience effects of agrobiodiversity become significantly stronger in areas with developed heritage assets and tourism infrastructure. This supports previous research arguing that interactions between natural and cultural capital are central to sustainable rural development (Lane and Kastenholz 2015; Živković et al. 2025).

Migration-related findings also reinforce the economic significance of rural diversification. Higher resilience levels and tourism participation are associated with lower migration intentions, consistent with studies identifying economic vulnerability as a key driver of rural out-migration (de Haas 2010). These findings suggest that integrated agriculture–tourism systems may contribute to demographic stability in mountain economies.

Overall, the results indicate that agrobiodiversity contributes to rural resilience through interconnected ecological, cultural, and tourism-related mechanisms rather than through agricultural production alone. This perspective is particularly relevant for the Western Balkans, where fragmented agriculture, demographic decline, and uneven market integration continue to constrain rural development.

The findings therefore support integrated rural development approaches combining biodiversity conservation, tourism diversification, and cultural heritage preservation within broader territorial resilience strategies.

Although the panel design and robustness checks strengthen the analysis, the findings should be interpreted as longitudinal associations rather than definitive causal relationships. Nevertheless, the study provides important empirical evidence from Northern Albania and contributes to a broader understanding of resilience dynamics in transition economies in mountainous regions.