. 2018 Apr 3;18:45. doi: 10.1186/s12862-018-1166-5

Table 1.

Statistics describing the relationships shown in Fig. 1. Models were fitted using both OLS and PGLS regressions

Model	N	Slope (±SE)	r ²	λ (95%CI)	AIC	ln likelihood	F _{1, 194}	P-value
tSMR vs body mass
OLS	196	0.85 (0.02)^**	0.91		316.2	−153.5	1407.97	< 0.001
	196	0.76 (0.02)^**	0.88	0	314.6	−151.3	1408.00	< 0.001
PGLS	196	0.85 (0.03)^**	0.86	0.61 (0.51/0.91)	308.8	−148.4^a	1113.50	< 0.001
	196	0.87 (0.03)^**	0.79	1	312.9	−150.5	716.21	< 0.001
mSMR vs temperature
OLS	196	−0.42 (0.08)^**	0.41		311.8	−150.9	21.97	< 0.001
	196	−0.47 (0.08)^**	0.24	0	309.8	−148.9	61.57	< 0.001
PGLS	196	−0.45 (0.09)^**	0.16	0.51 (0.39/0.87)	293.8	−140.9^a	23.09	< 0.001
	196	−0.38 (0.10)^**	0.17	1	312.7	−150.4	14.97	< 0.001

CI confidence interval

^a the PGLS model is significantly better than the OLS model (likelihood ratio test)

^** P < 0.0001