Table 4.

Results for models with descending numbers of limiting factors

Model	Covariates included	df	AICc	Δi	W i	R 2
1	Stock, Nov–Dec, Mar, Apr–Jun, Jul–Aug, Sep–Oct	23	593.0	42.0	0.0000	0.62
2	Stock, Mar, Apr–Jun, Jul–Aug, Sep–Oct	26	575.1	24.1	0.0000	0.62
3	Stock, Nov–Dec, Apr–Jun, Jul–Aug, Sep–Oct	26	575.1	24.1	0.0000	0.62
4	Stock, Nov–Dec, Mar, Jul–Aug, Sep-Oct	26	586.7	35.7	0.0000	0.49
5	Stock, Nov–Dec, Mar, Apr–Jun, Sep–Oct	26	582.8	31.8	0.0000	0.54
6	Stock, Nov–Dec, Mar, Apr–Jun, Jul–Aug	26	575.1	24.1	0.0000	0.62
7	Stock, Apr–Jun, Jul–Aug	32	551.0	0.0	0.5794	0.62
8	Stock, Apr–Jun	35	564.7	13.7	0.0006	0.31
9	Stock, Jul–Aug	35	551.6	0.6	0.4199	0.51
10	Stock	38	571.3	20.3	0.0000	0.01

Note: df–degrees of freedom, AICc–bias adjusted Akaike’s information criterion, Δ_i is–simple difference in AICc from the model with the minimum AICc, W_i–the Akaike weight, R²–the coefficient of determination. See Table 1 for a key to abbreviations of covariate names. Model 1 includes all limiting factors. Models 2 through 6 have one factor eliminated. Food availability in November–December, March and September–October provided the least value and these factors were eliminated in model 7. Models 7 and 9 could not be distinguished on the basis of our selection criteria: Δ_i < 2. Models without stock (prior abundance in the fall) were not ecologically plausible