Lake Dianchi is the largest shallow lake in Yunnan-Guizhou plateau and the sixth largest one in China. The lake has been experiencing cyanobacterial blooms in the last two decades. Although a few studies have investigated the tempo-spatial dynamics of cyanobacterial blooms and their underlying mechanisms, knowledge regarding the food web structure and trophic interactions in bloom-dominated ecosystems is scarce. In the present study, an Ecopath model was developed to assess the entire lake ecosystem on the basis of historical and survey data obtained between 2009 and 2010 at Lake Dianchi. The results showed that the aggregation of flows sensu Lindeman refers to six trophic levels (TLs), and most biomasses and trophic flows were primarily concentrated at the first three levels. About 77.5% of the trophic flows from TLI to TLII originated from detritus, whereas high proportions of under-utilised zooplankton biomass returned to the detritus because of low transfer efficiencies (2.9%) in TLII. The microbial loop was considered to be involved in linking the transfer between detritus and TLII. In addition, low values of connectance index and average mutual information implied that the food web tended to be lost in information diversity and had a less complicated structure. High cycling flows concentrated in the microbial loop reflected that the ecosystem enhanced recycling to forms positive feedback by which ecosystem locked the nutrients and promoted the inflation of biomass in plankton communities. Thus, Dianchi Lake was clearly thought to be a bottom-up control ecosystem. These characteristics of the food web partly explained why cyanobacterial blooms were exceptionally heavy and durable in this lake. Finally, the implications of artificially stocking filter-feeding fish (bighead and silver fish) and exotic zooplantivorous icefish on the ecosystem structure and function are discussed herein. (C) 2014 Published by Elsevier B.V.