The Application of the Generalized Additive Model to Represent Macrobenthos near Xiaoqing Estuary, Laizhou Bay

SIMPLE SUMMARY: Macrobenthos is widely used as an indicator of ecological health in marine monitoring and assessment. Xiaoqing Estuary has been subjected to different degrees of freshwater injection, resulting in significant changes in salinity, which has affected macrobenthos in the area. In this paper, we determined the main environmental variables driving the macrobenthic community and established a generalized additive model (GAM) model of macrobenthos based on the Margalef diversity index (d(M)), following which the distribution of macrobenthos richness can be predicted. The present findings will provide useful information for future studies on the correlation between macrobenthic communities and environmental factors in salinity-stressed areas of estuaries. ABSTRACT: Macrobenthos is widely used as an indicator of ecological health in marine monitoring and assessment. The present study aimed to characterize the interrelationships between the distribution of the macrobenthos community and environmental factors near Xiaoqing Estuary, Laizhou Bay. Responses of species richness to environmental factors were studied using the generalized additive model (GAM) and the Margalef diversity index (d(M)) as indicators of species diversity instead of individual indicator species. Six factors were selected in the optimal model by stepwise regression: sediment factors (organic matter, phosphate, nitrate nitrogen, and ammonium nitrogen) and water factors (salinity, and ammonium nitrogen). The response curves generated by the GAM showed a unimodal relationship among taxa diversity, salinity in water, and sediment organic matter. d(M) was positively correlated with ammonium nitrogen in water and was negatively correlated with phosphate in the sediment. The model optimized by forward stepwise optimization explained 92.6% of the Margalef diversity index with a small residual (2.67). The model showed good performance, with the measured d(M) strongly correlated with the predicted d(M) (Pearson R(2) = 0.845, p < 0.05). The current study examined the combined influence of multiple eco-factors on macrobenthos, and the Margalef diversity index of macrobenthos was predicted by the GAM model in a salinity-stressed estuary.