Table 2.

Generalized Least Squares Regression Results, with and without Phylogenetic Control.

Model	Nonphylogenetic			Phylogenetic
	T	df	P	t	df	P	Trees with Negative Effect	Trees with Significant Effect
All mammals
Genome ∼ population	–0.09	197	0 .93	1.2	197	0.23	16/100	0/100
genome ∼ population * mass * metmass	–2.95	65	0.004	–0.51	73	0.61	100/100	0/100
genome ∼ population * mass	–4.01	74	0.000	–0.59	75	0.56	96/100	0/100
genome ∼ population + mass	–0.07	75	0 .95	2.92	75	0.005	0/100	100/100
Body mass < 316 g
genome ∼ population	–3.31	37	0.002	0.7	37	0.49	0/100	0/100
genome ∼ population * mass * metmass	–1.39	29	0 .17	0	37	1	24/100	0/100
genome ∼ population * mass	–1.34	35	0 .19	0.02	39	0.98	0/100	0/100
genome ∼ population + mass	–2.6	36	0.01	0.82	39	0.41	0/100	0/100
Rodentia
genome ∼ population	–2.18	57	0.03	1.21	59	0.23	6/100	1/100
genome ∼ population * mass * metmass	–0.63	23	0.53	0.84	31	0.41	0/100	0/100
genome ∼ population * mass	–2.59	29	0.02	–0.5	33	0.62	100/100	0/100
genome ∼ population + mass	–0.78	30	0.44	1.36	33	0.18	0/100	0/100

Note.—Regressions incorporating phylogenetic covariation were repeated across 100 randomly chosen, equally likely trees (see supplementary information, Supplementary Material online), and the test statistics reported are from one randomly chosen tree. The last two columns indicate for how many of these 100 trees the effect of population size was negative (consistent with the prediction) and for how many of these trees the effect of population size was significant (with either positive or negative effect or population size). Test statistics are for the effect of population size, after accounting for other variables in the model, and the sign of the t statistic indicates the sign of the slope for population size.