A comparison of GAM and GWR in modelling spatial distribution of Japanese mantis shrimp (Oratosquilla oratoria) in coastal waters

Abstract

Japanese mantis shrimp (Oratosquilla oratoria) is an economically important species in the coastal fisheries of East Asian countries, but the yield shows a declining trend in recent years under intense fishing pressure. The spatial distribution and biomass of this species vary geographically, sexually and periodically due to its adaptability of diverse environmental factors, which leads to substantial challenges in modelling and predicting the spatio-temporal distribution and adaptive management. In this study, based on distribution data of Japanese mantis shrimp in coastal Shandong trawling surveys from 2016 to 2017, period- and sex-specific generalized additive models (GAM) and geographically weighted regression models (GWR) were built to evaluate influential factors on the occurrence and biomass of Japanese mantis shrimp and to test their performance and predictability. Our results showed that the spatial distribution of Japanese mantis shrimp was significantly associated with depth, sea bottom temperature, and silt and clay content in sediment. Specifically, Japanese mantis shrimp tended to aggregate in higher content of silt and clay, whereas the preference of depth and sea temperature. Following relatively higher bottom temperature, the distribution of Japanese mantis shrimp was concentrated in inshore water in the breeding period and in offshore water in the wintering period. There was remarkable difference in the distribution between male and female, and the influential factors differed between gender groups and changed in different life periods. Regarding the performances of models, GWR showed advantages over GAM in identifying influential factors and predicting distribution in the cross validation, and GWR was recommended in similar applications in the future. This study might contribute to better understanding of the distributional patterns of Japanese mantis shrimp and inform stock monitoring and sustainable management.