Understanding patterns of distribution shifts and range expansion/contraction for small yellow croaker (Larimichthys polyactis) in the Yellow Sea

Detecting and analyzing patterns of distribution shifts and range expansion/contraction is important to understand population dynamics and changes in stock status. Here, we develop a spatio-temporal model for yellow croaker (Larimichthys polyactis), which was fitted to bottom trawl survey biomass data collected in the Yellow Sea in the winter of 2001-2017. The model accounts for both spatial and spatio-temporal structure and can potentially include the effects of surface temperature and of an annual index, the Pacific Decadal Oscillation, represented using a recently developed spatially varying coefficient model. We selected a spatio-temporal model with no covariates based on Akaike's information criterion. The center of gravity for yellow croaker was found to move northwest between 2001 and 2010, and then southwest over the period 2010-2017. These results reflected the predicted progressive disappearance of yellow croaker density hotspots (i.e., highest density areas) in the north and southeast areas of the Yellow Sea, which resulted in the central area of the Yellow Sea becoming the only yellow croaker density hotspot in 2017. This finding has important implications for fisheries management in the context of the China-South Korea fisheries agreement, as it indicates a measurable displacement of yellow croaker biomass toward China. The exclusion of covariates from the spatio-temporal model was not expected a priori and may be due to the facts that environmental variations are not pronounced in winter in the Yellow Sea and that the representation of spatial and spatio-temporal structure in spatio-temporal models accounts for a large proportion of the variability in the data.

Automated summary

A spatio-temporal model was developed to analyze the distribution shifts of yellow croaker in the Yellow Sea, revealing a progressive disappearance of density hotspots in the north and southeast areas, with the biomass moving towards China. The exclusion of covariates in the model was unexpected, possibly due to limited environmental variations in winter and the comprehensive representation of spatial and spatio-temporal structure.