Abstract
Abstract Understanding the relationship between tree height and diameter at breast height (DBH) is crucial for forest growth modeling, while understory vegetation plays a key role in maintaining biodiversity and ecosystem stability. However, in subtropical Cunninghamia lanceolata plantations, especially in the mountainous areas of southern Jiangxi, comprehensive studies simultaneously evaluating tree height-DBH models and understory diversity across different stand ages are still relatively scarce.This study selected 910 Cunninghamia lanceolata in the artificial forest of southern Jiangxi, China. Their height-diameter at breast height data were fitted with nine growth functions, while the composition, floristic elements, and species diversity of the understory shrubs and herbaceous plants were investigated and analyzed. Finally, Mantel correlation tests were conducted. Among the nine candidate height-diameter models evaluated using an internal validation approach, the Quadratic, Mixed, and Exponential models performed the best, with the Quadratic model having the highest R 2 value of 0.6429, an RMSE ranging from 2.3904 to 2.4255 m, and an MAE ranging from 1.8964 to 1.9237. Vegetation diversity analysis revealed a total of 43 families, 64 genera, and 74 species of understory plants. The Simpson diversity index for the shrub layer ranged between 0.9364 and 0.9381, and the Shannon-Wiener diversity index ranged between 0.1648 and 0.1687. It indicates that the species diversity value is highest in the middle-aged forest plots, while the evenness value is highest in the mature forest plots. The 43 families of understory plants were divided into 7 distribution types, and the 64 genera into 13 distribution types, with understory plants exhibiting a clear phylogeographic association in their distribution types. The Mantel correlation test indicated that both tree height and diameter at breast height showed mostly insignificant negative correlations with understory vegetation diversity indices, while there were significant differentiated association patterns among the diversity indices themselves. Given the presence of confounding factors and the limited sample size, this analysis did not provide strong evidence sufficient to demonstrate a consistent correlation.