Species occurrence data form the basis of the establishing Species Distribution Models (SDMs). As a new technology, environmental DNA (eDNA) has been widely used for species monitoring and species diversity assessment, but it is still unclear whether it can replace or supply traditional trawl surveys to provide occurrence data for SDMs. This study took a bifurcated estuary with seawater flow backward phenomenon - the Yangtze River estuary as the research area, and the typical marine benthos - Collichthys lucidus as the research object, we carried out a comparative survey during two trips in August and November 2021, aiming to use trawl-survey data as the control group to assess the results of habitat modeling using eDNA data alone or in combination. The results showed that the trawl surveys indicated that C. lucidus inhabited almost exclusively the nearshore waters of estuaries with high salinity, which is consistent with the traditional perception that it does not inhabit freshwater (i.e. marine species). However, eDNA sampling suggested that this species was also widely distributed in the freshwater rivers of the south branch of the Yangtze River, which is likely a result of seawater intrusion from the north branch. Consequently, the Maximum Entropy Model (MaxEnt) predictions based on eDNA expanded the suitable habitat range of C. lucidus in the estuarine area. Moreover, imprecise spatial information affects the identification of key environmental requirements, underestimating the importance of salinity in habitat selection for C. lucidus. Our study highlights that in highly dynamic and open water environments like estuaries, cautious evaluation is necessary when using eDNA as species occurrence data for prediction with spatially explicit models requiring precise spatial information.