“…The inclusion of SST in the A and A 1 models increases the absolute forecasting errors for all species. Covariates in the ARIMAX and Bayesian settings are incorporated linearly; however, the effect of winter SST on groundfish catch is likely to be nonlinear (Laurel, Hurst, Copeman, & Davis, 2008;Rooper & Martin, 2009;Sadorus, Mantua, Essington, Hickey, & Hare, 2014), which is apparent in Supporting Information Figure S3; therefore, the effect's nonlinear shape is not being taken into account in the A, A 1 , B, and B 1 forecasting methods.…”