Objectives The subtropical island forest ecosystem is one of the important components of the Earth's terrestrial forest ecosystem. The construction of a biomass model of typical tree species in subtropical island forests is of great significance for the enrichment of the biomass model library of typical tree species in terrestrial forests and for the advancement of understanding of the biomass characteristics of typical tree species in subtropical island forests.

Methods The study focused on Dachen Island, Taizhou City, Zhejiang Province, China, an island in the subtropical climate zone. The measured above- and below-ground biomass, diameter at breast height, tree height, and crown spread data of two representative tree species ( Sapium sebiferum, Mallotus tenuifolius) in Dachen Island were collected. These data were then used to develop a biomass model for various tree species. Firstly, the correlations between total biomass and diameter at breast height, tree height, and crown width were subjected to analyzed. Secondly, unitary, binary, and ternary biomass power function models of total biomass with diameter at breast height, tree height, and crown width were constructed and evaluated separately. Thirdly, the data pertaining to the above- and below-ground biomass data were employed to analyze the characteristics of above- and below-ground biomass share and its relation with diameter at breast height, tree height, and crown width.

Results The total biomass and diameter at breast height of the two typical tree species on Dachen Island are most closely related, followed by tree height and crown width. The determination coefficients of the univariate power function model for the total biomass and diameter at breast height of the two typical tree species are both 0.97 and the average residuals are both below 0.33 kg. Similarly, the determination coefficients of the bivariate power function model for total biomass, diameter at breast height, and tree height all reached 0.97, and the average residual and root mean square error were reduced to a certain extent compared to the bivariate model. When the biomass model increased from binary to ternary, the accuracy of the model decreased slightly, especially the accuracy of the ternary total biomass model of M. tenuifolius was significantly lower than that of the binary model. The ratio of aboveground biomass to underground biomass of S. sebiferum was significantly positively correlated with tree height and crown width. The ratio of aboveground biomass to underground biomass of M. tenuifolius was significantly positively correlated with tree height. The average proportion of aboveground biomass of S. sebiferum and M. tenuifolius was70.1% and 66.9%, respectively, indicating that the aboveground biomass was significantly higher than that of underground.

Conclusions The experimental comparison revealed that the binary power function model yielded the most optimal estimation of total biomass for the typical tree species on Dachen Island.