Table 2.
Outcomes of Generalized Linear Models (GLMs) and Generalized Additive Models (GAMs).
| Model | Dataset | Dependent variable | Independent variables | sign terms (estimate) | p-value | Dev. Expl. | AIC |
|---|---|---|---|---|---|---|---|
| GLM | 1997–2013 | biomass | SSTspawn + CHLspawn + SSTpost + CHLpost | non significant terms | |||
| GLM | 1997–2013 | biomass + catch | SSTspawn + CHLspawn + SSTpost + CHLpost | non significant terms | |||
| GLM | 1998–2013 | biomass | SSTspawn + CHLspawn + SSTpost + CHLpost | non significant terms | |||
| GLM | 1998–2013 | biomass + catch | SSTspawn + CHLspawn + SSTpost + CHLpost | non significant terms | |||
| GAM | 1997–2013 | biomass | s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | non significant terms | |||
| GAM | 1997–2013 | biomass + catch | s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | SSTspawn (Fig. 6.a) | 0.043 | 63.62% | 321.84 |
| GAM | 1998–2013 | biomass | s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | non significant terms | |||
| GAM | 1998–2013 | biomass + catch | s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | SSTspawn (Fig. 6.b) | 0.043 | 63.62% | 321.84 |
| GLM | 1997–2013 | biomass | LRI + SSTspawn + CHLspawn + SSTpost + CHLpost | LRI (2.871) | 0.001 | 80.33% | 311.94 |
| CHLspawn (−6.512e + 05) | 0.038 | ||||||
| GLM | 1997–2013 | biomass + catch | LRI + SSTspawn + CHLspawn + SSTpost + CHLpost | LRI (2.867) | 0.001 | 82.05% | 309.51 |
| CHLspawn (−6.078e + 05) | 0.037 | ||||||
| GLM | 1998–2013 | biomass | LRI + SSTspawn + CHLspawn + SSTpost + CHLpost | LRI (2.871) | 0.001 | 80.33% | 311.94 |
| CHLspawn (−6.512e + 05) | 0.038 | ||||||
| GLM | 1998–2013 | biomass + catch | LRI + SSTspawn + CHLspawn + SSTpost + CHLpost | LRI (2.867) | 0.001 | 82.05% | 309.51 |
| CHLspawn (−6.078e + 05) | 0.037 | ||||||
| GAM | 1997–2013 | biomass | LRI + s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | LRI (2.8663) | 0.001 | 85.70% | 308.91 |
| GAM | 1997–2013 | biomass + catch | LRI + s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | LRI (2.9253) | 0.005 | 91.90% | 300.66 |
| CHLspawn (Fig. 6.c1) | 0.035 | ||||||
| CHLpost (Fig. 6.c2) | 0.038 | ||||||
| GAM | 1998–2013 | biomass | LRI + s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | LRI (2.8663) | 0.001 | 85.70% | 308.91 |
| GAM | 1998–2013 | biomass + catch | LRI + s(SSTspawn) + s(CHLspawn) + s(SSTpost) + s(CHLpost) | LRI (2.9253) | <0.001 | 91.90% | 300.66 |
| CHLspawn (Fig. 6.d1) | 0.035 | ||||||
| CHLpost (Fig. 6.d2) | 0.038 | ||||||
In GAM, the use of s() stands for the application of the cubic regression spline as a smooth term. When significant terms (sign. terms) were detected, the model regression coefficients (estimate) and the associated p-value are shown, as well as the deviance explained and the AIC index, as an indicator of the model fitting performance.