Spatiotemporal distribution and the estimated abundance indices of Larimichthys polyactis in winter in the Yellow Sea based on geostatisticaldelta-generalized linear mixed models

Larimichthys polyactis is an important economic bottom fish in China. Understanding the spatial and temporal distribution and abundance index of L. polyactis contributes to the scientific management of L. polyactis resources. In this study, the Geostatistical delta-generalized linear mixed models (geostatistical delta-GLMM) was used to analyze the spatial distribution of the wintering population of the Yellow Sea and Bohai Sea L. polyactis from 2001 to 2011 and from 2015 to 2017 in the central and southern Yellow Sea. The biomass index of L. polyactis was estimated by geostatistical delta-GLMM, ordinary Kriging interpolation-based method and design-based swept area method, respectively. The results showed that there were mainly three high-density areas of biomass areas in the overwintering ground of the Yellow Sea in 2001 and 2002, namely, the north (36°00′-37°37.5′N, 123°15′-124°15′E), the central (33°75′-36°00′N, 123°15′-124°75′E) and the southeast (32°00′-33°75′N, 124°00′-125°15′E), and the central region had the highest density. Since 2003, the biomass density of L. polyactis has declined, and the density core area in the north and southeast has declined more than the central density core area. In winter, the total biomass index of L. polyactis is contrary to the change trend of the annual yield of L. polyactis and fishing boat power, showing a downward trend. In most years, the number of stations is more than 37, and the range of stations covers the study area. The factors of sampling stations can be excluded, which indicates that the resources of L. polyactis are still facing overfishing and the population is in a declining state. By 2016—2017, the density core area became less obvious. The geostatistical delta-GLMM estimated the total biomass index of the L. polyactis in the winter of 2001—2017 is more accurate than that estimated by the swept area method and ordinary Kriging interpolation-based method.