Occipital condyle width (OCW) is a highly accurate predictor of body mass in therian mammals

Table 2 Accuracy statistics for the regression model between natural log OCW and natural log body mass using the all taxon, species average dataset under several different ordinary least squares (OLS) and phylogenetic least squares (PGLS) regression models

Method	Regression	AIC	BIC	logLik	df	r²_adj	%PE	CF	%PE_cf	%SEE
OLS	Linear	441	453	− 217	402	0.9784	34.88	0.905	38.99	51.50
OLS	1/2 power	409	421	− 201	402	0.9810	32.70	1.145	31.28	49.09
OLS	1/3 power	455	467	− 224	402	0.9777	35.22	1.260	33.80	52.63
OLS	2/3 power	389	401	− 192	402	0.9811	32.03	1.047	31.09	47.69
OLS	3/4 power	391	403	− 193	402	0.9809	32.21	1.004	32.10	47.82
OLS	Quadratic	391	407	− 192	401	0.9810	32.05	1.068	30.98	47.73
OLS	Cubic	393	413	− 191	400	0.9809	32.04	1.055	31.05	47.79
PGLS	Linear	420	432	− 207	402	–	70.12	1.725	34.30	309.11
PGLS	1/3 power	367	379	− 180	402	–	75.61	1.953	32.95	276.45
PGLS	2/3 power	368	380	− 181	402	–	68.88	1.754	31.56	276.00
PGLS (OU)	2/3 power	394	406	− 194	402	–	32.00	1.046	31.08	47.45

Abbreviations: PGLS (OU), phylogenetic least squares under an Ornstein-Uhlenbeck model (rather than Brownian); AIC, Akaike Information Criterion; BIC, Bayesian Information Criterion; logLik, log likelihood; df, degrees of freedom; r²_adj, adjusted r² value; %PE, percent prediction error; CF, averaged correction factor (see “Methods”); %PE_cf, percent prediction error after applying correction factor; %SEE, standard error of estimate

ISSN: 1741-7007