Abstract: The changes of urban surface landscape will lead to damage, fragmentation, and losses of biodiversity, resulting in urban biodiversity decline. In order to explore the responses of urban plant diversity to the changes of landscape pattern, we chose the urban area surrounded by round-city highway as research object, set 36 plots in different directions with the Gupeng Square as the center, investigated the plant group in each of the plots through coverage rank method of Braun-Blanquet, and examined the features of plant species diversity in Xuzhou’s greenlands. Meanwhile, we divided the landscape of the research area into seven categories: buildings, grasslands, forest lands, waste lands, water areas, roads and farmlands with the help of Erdas, GIS software and Landsat TM remote sensing images from 2005 to 2015 as basic data source. The features of landscape pattern in the research area were analyzed by Fragstats software. We set nine different buffer districts within each of the sample plot by applying the space analyzing function of GIS, calculated 39 landscape pattern metrics using Fragstats, and explored the features of how plant diversity responsed to landscape pattern. After that, the landscape pattern metrics with significant influence on the plant diversity of urban greenlands under different extents were selected by redundant analysis (RDA). The results showed that the plant diversity had obvious temporal and spatial scale effects in responses to landscape pattern and landscape elements. In 2005, the patch area and shape metrics of unused land and farmland could better explain plant diversity in 800 m buffer zone. The patch fragmentation and connectivity metrics of construction land had significant impacts on plant diversity within the extent of 800-2000 m. In 2015, grassland and forest landscapes better explained plant diversity in the 800 m buffer zone while construction sites and unused land had a remarkable impact on plant diversity in the extent of 1000-2000 m. The current plant diversity was more strongly influenced by the 2005 landscape type metrics, indicating that the plant diversity lagged in its response to the changes of landscape pattern. The pattern metrics of landscape level were more consistent with the landscape-class level metrics, indicating that the interaction between species and spatial scale was more dependent on the landscape type. This study would provide effective decision-making information for Xuzhou City planning and construction as well as effective protection of urban green space biodiversity.