关键词: 丹霞梧桐, 极小种群植物, Maxent模型, 生境评价, 无人机, 丹霞山国家级自然保护区

Abstract: [Objective] The Maxent model was used to evaluate the potential habitat of Firmiana danxiaensis in order to provide a scientific basis for in situ and ex situ conservation.[Method] A total of 186 distribution points data of F. danxiaensis in the yellow leaf stage were obtained by unmanned aerial vehicle (UAV). Combined with six environmental variables, the Maxent model was used to predict the potential suitable area of F. danxiaensis, and the impact of main environmental variables on its distribution was analyzed.[Result] The AUC values of the training set and the testing set were 0.947 and 0.906, respectively. The result of Jackknife test indicated that DEM, slope and aspect were the main environmental variables affecting the potential distribution of F. danxiaensis, contributing 51.2%, 30.3% and 7.8%, respectively. Under the joint influence of the three main environmental variables, F. danxiaensis was mainly distributed on the sunny slopes with an elevation of 150-530 m and a slope of 20°-60° in Danxiasha national nature reserve. The potential suitable habitat was divided into four grades:highly suitable, moderately suitable, less suitable and unsuitable by natural break method. The total area of high and moderate suitable area of F. danxiaensis was 30.27 km², accounting for only about 10.41% of the total area under study. The less suitable area was 41.01 km², and the unsuitable area was 219.5 km², accounting for 89.59% of the total area. The field survey data verified that 92.1% of field distribution points of F. danxiaensis fell into high and moderately suitable area, and the kappa coefficient was 0.958, indicating that the prediction accuracy of Maxent model was high.[Conclusion] The Maxent model can accurately predict the potential suitable habitat of F. danxiaensis in Danxiashan national nature reserve, and revealed the three main environmental variables that affected its distribution, indicating that the model has a good effect on small-scale regional prediction. Two suggestions were put forward for the protection of F. danxiaensis: 1) in terms of in situ conservation, it is necessary to strictly restrict the construction of tourist facilities in the highly and moderately suitable areas, and establish a long-term UAV remote sensing monitoring system to track the dynamic changes of the population; 2) in terms of ex situ conservation, reintroduction experiments should be carried out as soon as possible to expand their population distribution according to their suitable area. This study also verified the advantage and feasibility of low-altitude UAV remote sensing in species identification and population surveys, and revealed its potential application in the monitoring and evaluation of Danxia landform.

Key words: Firmiana danxiaensis, plant species with extremely small populations(PSESP), Maxent model, habitat assessment, unmanned aerial vehicle(UAV), Danxiashan national nature reserve