In forestry, the stand structure is typically described by fitting probability density functions to distributions of the empirical tree size dimensions. Selecting an optimal model is essential for accurately characterising structural dynamics, though its performance may vary depending on ecological conditions. In this study, we evaluate the performance of five statistical models used to describe the distribution of diameter at breast height and basal area in pedunculate oak (Quercus robur L.) forests in the Republic of Moldova, taking into account the influence of soil moisture. Distribution parameters were estimated using the MLE, and model fit was assessed through an integrated approach combining primary criteria (the AIC and KS test), secondary indicators (the log likelihood and BIC), and graphical analysis to identify the most suitable distribution by forest type. In dry pedunculate oak-cherry forests and most dry pedunculate oak-blackthorn forests, the Birnbaum-Saunders distribution provided the best fit for diameter and basal area. In contrast, in pedunculate oak-hornbeam forests, the gamma distribution showed superior performance. The Birnbaum-Saunders distribution, rarely applied in forestry, showed a strong fit in stands on dry soils, suggesting its potential for modelling stand structure under extreme ecological conditions. Graphical analysis revealed an overlap between the Birnbaum-Saunders, inverse Gaussian, and log-normal distributions for basal area, indicating that these models reflect similar structural patterns in the pedunculate oak-cherry forests and pedunculate oak-blackthorn forests. The results confirm that soil moisture is a key factor in stand organisation, influencing both tree dimension distributions and structural convergence trends in forest ecosystems.

Keywords: Quercus robur; probability density functions; model selection; diameter; basal area; stand structure