Results/Conclusions Species-area relationship over landscape basically presents consistent on local (< 1-ha) rather than large scales. Species-area relationship is not linear, rather it is curvilinear with an inflexion at ca 1 ha in log-log space. The variation in the z-value, the slope of SAR in double-log space, is approximately stochastic over 24 ha plot on the local scales of less than 0.24 ha, only less than 12% of the variation is accounted for by biological, environmental and spatial variables. 51% of the variation is explained together by environmental and spatial variables on the scale of 0.96 ha, whereas the biological variables are insignificant. However, the variation is almost completely explained together by the three groups of variables on the scale of 1.5 ha. Biological and environmental variables jointly account for almost all of the variation in z-value across scales, but their independent contributions are negligible. We suggest that neutral theory basically explains the SAR on or across local scales, while a lot of non-neutral processes co-determine the variation on or across larger scales. Interspecific competition probably generates the neutrality and non-neutrality over landscape, while other variables might shape the community through interspecific competition. Arguably, a primary mechanism for SAR across scales is potentially dispersal limitation, through which environmental factors affect the SAR. We highlight reconciliation of neutral and non-neutral theory in a scale framework, and the incorporation of interactions between biological, environmental and spatial variables into understanding community.