Construction of growth model and growth simulation of Yunnan Arabica coffee.

Yunnan is the primary coffee-producing region in China. With the changing consumption habits of younger consumers, the coffee industry has experienced rapid growth, and enhancing both quality and yield has become a major focus of research. However, crop growth models are predominantly applied to the simulation of annual grain crops. To date, there has been no reported research on the development of growth models specifically for Arabica coffee in Yunnan.This study is based on the coffee growth model initially assessed and calibrated using long-term experimental sites in Costa Rica and Nicaragua in Central America (Ovalle-Rivera O. et al. Agrofor. Syst. 94:2033-2051, 2020), and has been locally calibrated to construct a growth model for Arabica coffee in Yunnan. The main tasks are as follows: Focusing on the key coffee-growing regions in Yunnan, data on local climate, soil, phenology, management practices, and yield were collected. The study explored the calibration of model parameters using Bayesian optimization, gradient descent, and particle swarm optimization algorithms. By comparing two indicators-root mean square error (RMSE) and mean absolute error (MAE)-it was determined that particle swarm optimization produced the best results under Yunnan's ecological and climatic conditions, and thus, this algorithm was used to construct the Yunnan Arabica coffee growth model. Furthermore, simulations were conducted to examine the effects of varying fertilization amounts, shading practices, shading intensities, altitudes, and weather conditions on coffee growth. The study discusses the impact of these factors on the growth and yield of Yunnan Arabica coffee, providing scientific evidence for optimizing cultivation management strategies. The results indicate that moderate fertilization and shading can significantly improve coffee yield, although both factors have a "saturation point." Within a specific range of total fertilizer application, the impact of single application rates on final yield is not significant. This suggests that when formulating a scientifically sound fertilization strategy, the total nitrogen fertilizer application should be optimized first. Among the major production areas, the climate conditions in the Pu'er region are the most ideal, while other regions require tailored management measures. In summary, this research successfully constructed the Yunnan Arabica coffee growth model and simulated its performance under various conditions, revealing potential response trends of coffee to environmental changes. It provides theoretical support for the cultivation management and sustainable development of the Arabica coffee industry in Yunnan.