Ecological studies of freshwater biota in understudied regions are often obstructed by poor taxonomic knowledge. We posit that molecular tools can help alleviate this issue and present an example where we combined molecular tools, environmental data and ecological statistics to investigate the distribution and community ecology of an unknown fauna of hydropsychid caddisflies along altitudinal gradients in four Himalayan river systems of Central and Eastern Nepal. A total of 484 larval specimens from 34 tributaries were examined. Phylogenetic analysis of the mitochondrial cytochrome c oxidase I (COI) and the nuclear ribosomal RNA 28S were used to delineate molecular operational taxonomic units (MOTUs) applying three analytical methods: general mixed Yule-coalescent (GMYC) model, Automatic Barcode Gap Discovery (ABGD) and Bayesian Phylogenetics and Phylogeography (BPP). Spatial distributional patterns and potential differences in ecological niches among MOTUs were statistically tested using regression and correlation approaches. Further, we examined the data for signs of non-random structure in MOTU communities. MOTU diversity within the family of Hydropsychidae was generally high but varied across evaluated gene fragments and slightly among delineation methods. Yet, the subsequent evaluation of environmental and spatial drivers and resulting distributional patterns were highly consistent among the different MOTU estimates. Within each river system, we found community composition varied greatly along the altitudinal gradients, with many MOTUs associated with specific altitudinal ranges. Prevalent MOTU turnover at the river system scale indicated high $β$-diversity in the hydropsychid community leading to high degrees of regional endemism. In the Langtang river system, we found fewer MOTU co-occurrences than expected by chance. These results highlight the utility of DNA-based approaches using variable genetic markers (mitochondrial or ribosomal nuclear) for primary biodiversity assessment of poorly studied groups or regions. Our study further shows that DNA-based biodiversity measures are suitable for downstream applications, such as exploring fundamental questions in stream ecology.