Table 3. Regression statistics of maximum prolonged-swimming speed on predictor variables.

MPF subset
AICc	βf	P(f)	βf 2	P(f 2)	βAR	P(AR)	β logM	P(logM)	AdjR2
70.07	0.14	0.051	-	-	0.76	∼0	0.21	0.004	0.81
71.88	-	-	-	-	0.82	∼0	0.15	0.025	0.80
72.21	0.17	0.067	−0.043	0.61	0.76	∼0	0.21	0.004	0.80

Estimates from phylogenetic generalized least squares regression with λ estimated by maximum likelihood (λ is a parameter that controls the influence of the phylogenetic variance-covariance matrix on the estimates). In all models in both datasets, λ was less than 0.0001, indicating that the resulting GLS regression was reduced to an OLS regression. The β are standard partial regression coefficients, P is the probability of the effect. AICc is the small sample size corrected Akaike Information Criterion. AdjR² is the R² adjusted for the number of parameters in the model. The predictor variables are f (body fineness ratio), f ², AR (propulsive fin aspect-ratio), and logM (body mass). The model-averaged regression coefficients are given in Fig. 5.