Abstract: To clarify the changes and driving mechanism of the net primary productivity (NPP) in Yuanyang, Yunnan Province, we simulated the spatio-temporal distribution of vegetation NPP in Yuanyang from 2005 to 2015 based on the improved light energy utilization model (CASA) and MODIS-NDVI data. The correlations between NPP and climate factors or land use types were analyzed. From 2005 to 2015, the spatial distribution of NPP was significantly different in Yuanyang, increasing from the northeast to southwest, with an annual average of 380.57 g C·m^-2·a^-1. The vegetation NPP differed among land use types. The monthly and interannual NPP showed substantial variation. The uneven distribution of solar radiation had greater influence on NPP than that of precipitation during the year. The annual NPP presented a fluctuant increasing trend during the past decade, with an increment amplitude of 0-37 g C·m^-2·a^-1. The overall annual NPP was increased slightly, with an annual growth rate of 29600 kg C·a^-1. With the implementation of returning farmland to forest and grassland, the area of cultivated region with NPP of 200-300 g C·m^-2·a^-1 was decreased, and that of the woodland and grassland with NPP of 300-400 g C·m^-2·a^-1 showed an increasing trend. The change trends of NPP in different areas of Yuanyang could be classified as significant increase (P<0.01), slight increase (P<0.05), and no significant change, of which 29.5% of the area increased significantly, and 16.3% of the area increased slightly. The trend of NPP increasing varied across different vegetation types. The proportions of areas with significantly increasing NPP were 58.1% for woodland,19.2% for grassland, 17.4% for arable land, 11.4% for construction land, and 8% for garden. The changes of NPP were influenced by natural factors and human factors. The correlation coefficient between NPP and temperature or solar radiation was higher than that of precipitation. The large-scale increase of NPP was a result of heat and energy coupling. The comprehensive influence of vegetation characteristics in forested area and human activities including returning farmland to forest and grassland resulted in the greatest seasonal variability of vegetation NPP in forest, which was the main land-use type with significant changes of NPP. Our results provide valuable reference for land use, resources management, and ecosystem reconstruction.

Key words: spatial interaction, geographical detector, spatial distribution, Fujian Province, invasive plant