Table 2.

Symbiosis genes tend to be present in more genomes, have higher copy number, higher rates of horizontal transfer than other genes and less phylogenetic signal in gene presence/absence (lower δ) than non-symbiosis genes. Median values are shown. Duplication and transfer are estimates from GeneRax for genes with at least four genomes represented and at least six total copies. δ and Ka : Ks calculated only for genes with at least three sequences (numbers of gene in parentheses; results including all genes were similar—see electronic supplementary material, table S4). The Welch p-value is from a t-test comparing nodulation to fixation genes. The lower part of the table shows the probability that values are greater than a random sample of genes found in the same number of genomes and having approximately the same number of copies as the symbiosis genes.

	strains	copies/genome	duplication	transfer	median pairwise Ka : Ks	R2	δ
non-symbiosis	2	1	10−7	0.08	0.22 (n = 11 152)	0.76	3.3 (n = 9737)
nodulation	6	1	10−7	0.24	0.20 (n = 48)	0.23	1.6 (n = 35)
fixation	22	1	10−7	0.14	0.20 (n = 47)	0.54	2.2 (n = 23)
p-value Welch	<0.001	0.55	0.17	<0.001	0.9	0.007	0.56
benefit	19	1.08	0.002	0.20	0.15 (n = 7)	0.29	1.1 (n = 3)
fitness	17	1.13	10−7	0.18	0.23 (n = 19)	0.19	1.9 (n = 12)
probability of the estimated values being less than random samples of genes (%)
nodulation			0.2	0	99.7	99.7	99.9
fixation			0	0	99.2	88.6	99.2
benefit			6.7	0.6	99.6	96.5	95.9
fitness			0.8	4.1	73.4	85.5	94.5