Résumé
A tree’s architectural development can be described using statistically-based models. Here, we demonstrate the integration of an architecture described with hidden semi-Markov chains (HSMC) with L-PEACH, a carbon partitioning model for simulating developing peach trees. In previous studies, it was shown that in peach trees, several axillary buds can be found at each node, organised as a central bud, which can be blind, floral, vegetative (bud formed) with or without one or two lateral flower buds or latent (bud not fully formed). Axillary buds were also shown to be organised along a shoot into successive zones. Within each zone, the bud fates were fairly homogeneous, but between zones, they differed strongly. To represent these architectural patterns, bivariate HSMCs that were previously parameterised for unpruned trees in France were adapted to model pruned trees in California. From different models that have been previously estimated depending on shoot categories (as defined by the number of metamers), we considered limitations on the shoot sizes to be related to carbon availability and physiological factors. We also modelled responses to pruning and included algorithms for automatically handling some pruning and fruit thinning cases. This work demonstrates that HSMC concepts for describing tree architecture can be used to model pruning responses to simulate peach trees visually similar to those found in orchards