Exploring Nonlinear Time Series Models Forecasting Relative Abundance of Red Snapper (Lutjanus campechanus) in the Gulf of Mexico

Red snapper (Lutjanus campechanus) is an iconic fisheries species in the Gulf of Mexico.  Assessment of the species has been complicated due to the high mortality attributed to shrimp trawl bycatch and regulatory discards as well as uncertainties of incorporating environmental processes.   Models assessing red snapper stocks have been shifted from VPA, ASAP, and CATCHEM to Stock Synthesis.  Nonlinear time series (NLTS) models using nonparametric methods have shown predictability for biological populations and may have potentials forecasting abundance of fish stocks because external processes can be implicitly expressed via the behavior of time series data, rather directly formulated into the models.  Applying NLTS models, we analyze red snapper time series of fisheries independent and fisheries dependent indices and examine dynamics of the species and predictability of the NLTS to develop short term forecasts of stock abundance.   Results showed that NLTS models exhibit better predictability for fisheries dependent indices than fisheries independent indices across the western and eastern Gulf.  Prediction can be improved with including the environmental indicators using multivariate NLTS technique.  We further explore the utility of the NLTS models to examine potential associations between the stock trends of Gulf red snapper and the ecosystem indicators.