Abstract: Species distribution model has been widely used to estimate the habitat distribution of fishery resources. For rare fish species, the excessive zero- values in dataset is an important issue to be considered in the prediction of their spatiotemporal distribution. Zero-inflated model would be an effective way to solve this problem. With the survey data of Coilia nasusresources collected in the Yangtze estuary from 2009 to 2016, zero-inflated Poisson (ZIP) and zero-inflated negative binomial (ZINB) models were used to establish the relationship between the density of Coilia nasus and the environmental factors. The optimal model was selected to predict the density distribution of Coilia nasus in the Yangtze estuary in all seasons of 2017. In the first part of the results of zero-inflated models, water temperature had a significant positive effect on the occurrence of non-zero resource density of Coilia nasus.In the second part of results of zero-inflated models, longitude had a significant negative effect on Coilia nasusdensity and water temperature had a significant positive effect on Coilia nasusdensity. The optimal ZINB model has better fitting and prediction ability than the optimal ZIP model. The predicted value of Coilia nasus density was gradually decreasing from west to east, with a consistency spatial distribution between the predicted and observed value. The predicted value of Coilia nasus density showed a temporal distribution: summer> spring> autumn> winter, consistent with the seasonal variation of water temperature. Our results provide reference for the studies of the relationship between rare fishes and environmental factors and related studies on spatial and temporal distribution of rare fishes.

Key words: Coilia nasus, zero-inflated model, count data, the Yangtze estuary.