Biological and physical processes operate collaboratively through spatial or temporal scales to form ecological patterns, which are considered as a key element for understanding the natural liens within an ecosystem. Given the importance of scaling in ecological inference, this study elucidates how physical and biological variables under or within scales interact with each other. Density of Sicyopterus japonicus and environmental variables are examined and quantified at 70 stream sections distributed among 14 reaches in the Datuan stream catchment of northern Taiwan during the fall and winter of 2007, as well as the spring and summer of 2008. Hierarchical linear regression analysis indicates that S. japonicus density and habitat features are related on two levels, i.e., sections within reaches and reaches within streams throughout the year. Moreover, parameter uncertainty is represented by the confidence interval, which is calculated by the variance-covariance matrix of hierarchical linear model (HLM) parameters. According to HLM results, environmental variables at the section level (water depth and current velocity) and the reach level (stream width, water temperature, stream slope, soil erosion index) influence S. japonicus density. Although S. japonicus density varies significantly among reaches and sections within reaches, cross-level interaction may not always influence the distribution, processes and activities of S. japonicus throughout the year. The HLMs of S. japonicus density associated with stream features describe thoroughly multiple processes and the activities of S. japonicus across scales and within scales during different seasons. The annual HLM results represent the overall biological and physical patterns of the Datuan stream annually, explaining why they do not reflect seasonal associations or explain S. japonicus-related activities and environmental features of the stream.