Accumulation of slightly deleterious mutations in mitochondrial protein-coding genes of large versus small mammals -- Popadin et al. 104 (33): 13390 Data Supplement - HTML Page - index.htslp -- Proceedings of the National Academy of Sciences

Popadin et al. 10.1073/pnas.0701256104.

Supporting Information

Files in this Data Supplement:

Table 3. Comparison of average molecular traits for small versus large mammals, and the parameters of ordinary linear regressions for transition/transversion ratio k fixed at 3 across all groups

Trait	Average molecular traits			Ordinary linear models
Trait	Small	Large	P *	Intercept	Slope	P
K_a/K_s	0.040	0.052	0.004	0.030	0.002	<0.001
polarity-based K_r/K_c	0.686	0.988	<0.001	0.525	0.034	<0.001
volume-based K_r/K_c	0.728	1.103	<0.001	0.473	0.048	<0.001
polarity-volume-based K_r/K_c	0.481	0.554	0.003	0.421	0.010	<0.001
charge-based K_r/K_c	0.401	0.402	0.971	0.358	0.005	0.151
Grantham distance	55.288	58.592	<0.001	52.614	0.4692	<0.001

All the averages for large mammals are significantly greater than those for small ones, except for charge-based K_r/K_c(t-test). The regressions are all significant and positive, again except for charge-based K_r/K_c. Small and large mammals are defined in the same way as in the main text.

*Comparison of average molecular traits for small versus large mammals.

Table 4. Comparison of average molecular traits for small versus large mammals, and the parameters of ordinary linear regressions for a sub-sample of species with branches of less than one nucleotide substitution per codon (n = 89)

Trait	Average molecular traits			Ordinary linear models
Trait	Small	Large	P *	Intercept	Slope	P
K_a/K_s	0.041	0.055	0.002	0.029	0.002	<0.001
Polarity-based K_r/K_c	0.781	1.009	<0.001	0.603	0.030	<0.001
Volume-based K_r/K_c	0.830	1.025	0.001	0.582	0.034	<0.001
Polarity-volume-based K_r/K_c	0.479	0.536	0.048	0.432	0.008	0.0202
Charge-based K_r/K_c	0.396	0.456	0.117	0.302	0.012	0.0074
Grantham distance	55.411	58.835	0.002	51.885	0.528	<0.001

All the averages for large mammals are greater than for small ones at a 5% significance level except for charge-based K_r/K_c (t-test). The regressions are all significant and positive. Small mammals are those whose body mass ln W < 10.43 and large mammals are those whose body mass ln W > 10.43 where 10.43 is the median for n = 89 species (W in grams).

*Comparison of average molecular traits for small versus large mammals.

SI Text

A Check for Alternative Parameter Estimates.

For each subtree, under either transition / transversion ratio (free or fixed at 3) the PAML's model 1 with a free, branch-specific K_a/K_s yields a significantly higher maximum likelihood value L as compared with the PAML's model 0 with constant K_a/K_s across all branches of that subtree (i.e., 2DlnL is larger than a 5% critical value for a c² distribution with the number of the degrees of freedom dependent on the subtree in question: P(c²) < 0.0001 throughout except Lagomorpha where P(c²) = 0.0348). Thus, models with free K_a/K_s provide a better fit for all of the analyzed subtrees.

The transition/transversion parameter k found from model 1 with free, branch-specific K_a/K_s is 2.1 for Afrotheria, 10.6 for Artiodactyla, 7.0 for Carnivora, 10.1 for Cetacea, 3.5 for Chiroptera, 2.0 for core insectivores, 1.6 for Rodentia, 4.3 for Lagomorpha, 1.9 for Metatheria, 6.8 for Perissodactyla, and 5.6 for Primates and Dermoptera. The branch-specific values of K_a/K_s and the above estimates of k were used to compute K_a/K_s and K_r/K_c throughout the paper.

To check whether our results are robust with regard to the estimates of k, the basic analyses were repeated with k fixed at 3 across all groups (SI Table 3). The average molecular traits for small versus large mammals, and the parameters of ordinary linear models follow the same pattern as obtained for group-specific k values (see Comparison of Large Versus Small Mammals and Ordinary Linear Models sections in the main text).

A Check for Saturation Effect.

To be sure that the obtained regressions are immune to the presence of long branches, we repeated the basic analyses on a subsample of species with sufficiently short branches whose length was less than one nucleotide substitution per codon (89 species; SI Table 4). The results are found to be qualitatively similar to those for the full sample of 110 species (see Comparison of Large Versus Small Mammals and Ordinary Linear Models sections in the main text).