| Article ID | Journal | Published Year | Pages | File Type |
|---|---|---|---|---|
| 4492718 | Agriculture and Agricultural Science Procedia | 2015 | 8 Pages |
Litchi (Litchi chinensis Sonn. cv. Hong Huay) is one of the most economically important fruit products in Thailand, especially in the highlands of Chiang Mai where litchi can have natural off-season flowering even during the rainy season (June to August). The off-season production of litchi is not a common practice in the region because the physiological responses of litchi trees to flower induction techniques are still unclear. Therefore, we aim to use Random Forests (RF) as a tool to understand the effects of horticultural techniques on plant hormone dynamic or generative shoot development. A field experiment was conducted considering four horticultural treatments, namely (1) control, (2) girdling, (3) foliar spraying with 1% of monopotassium phosphate (0-52-34) and ethephon 800 mg/l, and (4) combination of girdling and foliar spraying. The girdling was applied at the end of leaf development stage. Foliar spray was applied three times at 7-d intervals from 14 days after girdling. Random Forests (RF) was applied to compute the percentage of flowering from plant hormone concentrations observed on 0, 7, 21, 35, 49, 56 and 63 days after the treatment, and to evaluate variable importance to illustrate physiological responses to horticultural treatments. In the field experiment, the three horticultural techniques could induce off-season flowering, while untreated trees had no flower, but leaf flushing. The RF model showed fair performance for modelling the percentage of flowering. The variable importance illustrated the effects of plant hormones in different tissues (i.e. wood, bark, leaves and apical bud) and timing (i.e. sampling dates). The variable importance of the total cytokinins and zeatin/zeatin riboside were observed to be high in wood, leaves and apical bud, while the importance of auxin in bark, wood and apical bud was found to be low on day 35. The RF model could present additional information for a deeper understanding of plant hormone dynamics in plant tissues and critical timing required for flower induction, which can be used for better and improved farmers’ management and practices in the future.
