Table 3. Tests of neutral evolution using paml

Family	Feature	Number of elements	Gene	d_N/d_S ratio	P value*
HERV-T=	Segmentally duplicated	9	pol	0.85	NS
HERV-T=	Not segmentally duplicated	17	pol	0.13	< 0.001
HERV-H	With all five of the shared deletions	43	env (part)^‡	1.08	NS
	With all five of the shared deletions		pol (part)^§	0.91	NS
	With none of the shared deletions	19	env (part)^‡	0.42	< 0.025
	With none of the shared deletions		pol (part)^§	0.21	<0.001
HERV-K (HML2)	Segmentally duplicated	6	All genes	1.21	NS
HERV-K (HML2)	Not segmentally duplicated	38	All genes	0.21	<0.001

We investigated d_N/d_S ratios in several examples of elements and regions that we assume are not under purifying selection (due to either being copied by segmental duplication of the host genome or sharing multiple inactivating deletions). In each case, the results obtained for the comparable regions of elements that do not have these assumptions of neutral evolution are shown in the row below. The d_N/d_S ratios given are all of internal branches using the two-ratio model, with neighbor joining trees built from the regions and elements under consideration. Also shown is the significance of the difference in likelihood if this d_N/d_S value is fixed at 1. NS, not significant.

*Taken from critical values on the c ² distribution.

^{=
This family is also termed HERV-S71 (the pol gene is used as it is more intact).
^{‡A short (207-bp) region immediately downstream of the env deletion site "E" (Fig. 4)
^{§A 663-bp region between pol deletion sites "C" and "D" (Fig. 4).}}}