Table 2. Comparison of alternative models describing variation in population growth rate (PGR), assessing the effect of controlling for migration strategy and decomposing risk scores into sub-components by forest type, season, nesting or foraging and individual forest changes.

Model	No scaling		Quantitative scaling		Qualitative scaling
	AICc	Δ AICc	AICc	Δ AICc	AICc	Δ AICc
PGR∼migration+total risk	−285.5		−278.9		−284.7
PGR∼migration+coniferous risk+broadleaf risk+Mediterranean risk	−298.4	−12.9	−289.4	−10.5	−293.4	−8.7
PGR∼migration+foraging risk +nesting risk	−285.5	0	−280.3	−1.4	−284.3	0.4
PGR∼migration+summer foraging+winter foraging+nesting risk	−284.3	1.2	−278.1	0.8	−288.2	−3.5
PGR∼coniferous risk+broadleaf risk+Mediterranean risk	−280.7	4.8	−279.0	−0.1	−278.1	6.6
PGR∼summer foraging+winter foraging+nesting risk	−280.7	4.8	−276.9	2	−278.1	6.6
PGR∼foraging+nesting risk	−277.5	8	−279.3	−0.4	−273.2	11.5
PGR∼total risk	−263.6	21.9	−269.6	9.3	−263.5	21.2
PGR∼migration+change 1 risk+change 2 risk+…+change 22 risk	−236.7	48.8	−243.2	35.7	−252.5	32.2
PGR∼change 1 risk+change 2 risk+…+change 22 risk	−230.2	55.3	−245.8	33.1	−249.2	35.5

Note that model fit was compared between models within the same scaling mechanism and that Δ AICc is calculated as the difference in AICc value from the baseline model of migration plus total risk; this is the most parsimonious formulation of risk score and all other models represent more complex formulations of this rather than containing independent data.