Abstract
The comprehensive understanding of the dormant pruning patterns in pear trees, along with the accurate identification of shoots suitable for pruning, is essential for implementing automated pruning and fruit production. Due to the complexity of tree architecture, previous descriptions of pruning strategies were qualitative summaries based on experience. In this study, we proposed a high-precision shoot extraction pipeline through point cloud alignment at different times, enabling a quantitative analysis of the pruning patterns. The structural parameters of 126 full bearing period pear trees, encompassing two cultivars and three architectures, were characterized, including the shoot number, single shoot angle and length, as well as shoot length density. The validation results demonstrated that the method attained an R (2) of 0.82, 0.92, and 0.85 for shoot number, single shoot angle and length, respectively, with mean absolute error of 18.72, 6.08°, and 0.13 m. The findings indicate that tree architecture exerts a greater influence on pruning compared to cultivar, particularly in Cuiguan, where significant differences were observed across diverse tree architectures. The characters of the corresponding annual (one-year-old) shoots (AS) and pruned shoots (PS) exhibit similar distribution. The AS, constituted 78.62% of the PS number, and 94.90% of length of AS were pruned, indicating that dormant pruning in full bearing period pear tree primarily targets at the annual shoots, and the pruning of annual shoots is mainly by thinning. This study could help the automatic pruning system make pruning decisions and promotes the development of fine orchard management.