Abstract:

Nowadays, plant functional groups (PFGs) have been recognized as a promising tool in management and restoration of degraded species-rich ecosystems. However, their potential distribution, especially in a tropical forest landscape, is rarely studied. Based on an investigation of 135 plots in a tropical natural forest landscape of Bawangling, Hainan Island, South China, the PFGs were classified by employing two criteria of successional status and potential maximum height. In this study the genetic algorithm for rule-set prediction (GARP) was used to estimate the potential distribution of the PFGs and then the receiver operating characteristics (ROC) was used to evaluate the predictive performance. Multiple linear regression analysis was applied in order to find out the major factors determining potential distributions of the PFGs. Results showed that the predictive accuracies for the eight PFGs were all over 70%. Compared with the climax functional groups, the pioneer functional groups had higher potential maximum distributional range, smaller maximum suitable range and presented more dispersed distribution pattern. This indicated there were different in autecological characteristics and responses to environment between pioneer and climax species. The analysis of correlation revealed that minimum temperature, annual mean temperature, maximum temperature, annual mean relative humidity, annual mean precipitation, aspect and elevation were the key factors determining potential distributions of PFGs. The results further also demonstrate that the GARP ecological niche model using presence-only data and large-scale environmental variables is a powerful tool for predicting species' geographic range.

Key words: woody plant, functional groups, potential distribution, ecological niche model, tropical rain forest