Differences in the initial configuration of ecological communities may lead to contrasting trajectories when facing environmental changes. Here, we propose to uncover the determinism of benthic communities by carrying out a detailed investigation of their response to small-scale modification of environmental conditions, including physical, chemical, and geological factors. At ten locations (confounding site and depth) in Northern Taiwan, communities were delineated using a morpho-functional classification of the organisms. A k-means clustering was used to identify k homogenous groups among transects. Their environmental determinism was examined by combining this result with 16 environmental variables of transect conditions into a regression tree framework. Biotic and abiotic data were further analyzed with a Multivariate Regression Tree (MRT) to ascertain the hierarchical environmental determinism. The classifications produced by both approaches were compared using the Adjusted Rand index (ARI) to assess the predictive power of unsupervised clustering on its missing explanatory components (abiotic variables). k-means and MRT produced five clusters, respectively, with a similarity of 0.82 in ARI. Wave motion, followed by substrate types resolved most of the variance, while chemical factors in this study were uniform throughout the region. Comparable structures for both methods (clustering groups) demonstrated that the delineated clusters matched with contrasting environmental conditions which could be explained by the existence of various benthic communities. Further consideration of these different communities and their environmental context will be important in determining their trajectories under global changes and may help in the interpretation of community modifications with changing environmental conditions.