Table IV. Multiple regression analyses of ε_c in six major food crops.

The final model and ranking of variables included in the final model were determined using the lowest AIC_c. Variable coefficients and significance were determined using the final model.

Crop	Variable Rank	Modela	AICc	Coefficient (10−3)	P
Peanut (n = 51)		YOR, St, T	−540.1		<0.001
	1	T	−536.0	0.426	<0.001
	2	YOR	−531.8	0.123	<0.05
	3	St	−526.6	−0.00238	0.12
Soybean (n = 117)		YOR, [CO2], St, T	−1,270.2		<0.0001
	1	T	−1,231.8	−1.04	<0.0001
	2	St	−1,226.3	−0.00824	<0.0001
	3	[CO2]	−1,204.3	0.0524	<0.05
	4	YOR	−1,201.3	0.0293	0.21
Rice (n = 102)		St, T	−1,133.8		<0.0001
	1	St	−1,125.0	−0.00960	<0.0001
	2	T	−1,014.3	−0.729	<0.01
Wheat (n = 159)		YOR, St, T	−1,451.2		<0.0001
	1	T	−1,447.9	−0.746	<0.001
	2	YOR	−1,438.4	0.105	<0.05
	3	St	−1,437.8	−0.00284	0.13
Sorghum (n = 23)	1	St	−228.3	−0.00739	0.11
Maize (n = 149)		YOR, T	−1,347.8		<0.0001
	1	YOR	−1,346.0	0.346	<0.0001
	2	T	−1,344.0	0.521	<0.01

^aIndependent variables included YOR, mean annual [CO₂] during the measurement period, T, and available S_t. Water available as precipitation and irrigation was included when sample size changed by less than 10% after including it in the analyses. Density was included in maize analyses but was not in the selected model.