Estimating TMRCA and mutation rates for the phase 3 Y chromosome STR clusters via ASD estimates

John McEwan

11^th March 2005

Disclaimer

These analyses are being undertaken to investigate the history of the R1b clusters, no claims are made about these clusters nor the methods used. They are provided for information purposes only. Specifically, many of the R1b clusters are not well supported, in addition genealogical and anthro-genealogical STR mutation estimates are a hotly disputed area.

Background

The phase 3 clustering was done using a distance measure Da that optimized identification of branches, but is not linear with branch age (Takezaki & Nei 1996). A distance measure which is approximately linear with age is based on the ASD measure (Goldstein et al 1995) and is commonly used in many forms of “model free” estimates of haplotype age and TMRCA estimates (Stumpf & Goldstein 2001; Zhivotovsky et al 2004). However, there has been considerable controversy about whether individual loci mutate at different rates, the actual mutation rate estimates, and the impact of population size and expansion on these estimates. In addition it appears that “effective” mutation rates (estimated on the basis of known divergence dates) are lower than estimates derived from observed mutation rates. The exact reasons are still being investigated but it appears that this is a consequence of a deviation from the neutral model.

In this analysis I am concentrating on the R1b clusters, individuals in these clusters are known to have expanded dramatically since the LGM and populations have typically grown at 3.5% per annum over that period. In addition I am using “effective” mutation rate derived from Zhivotovsky et al. (2004), but corrected for the different markers used.

Regards ASD calculations themselves the equation is best described by Stumpf and Goldstein (2001)

In the current context, The ASD was calculated for the various clusters defined in the phase 3 analysis, average ASD’s were calculated for all clusters (thereby providing an estimate of the clusters ) relative TMRCA and actual ages in years were estimated based on corrected mutation rates using the value (0.0007/generation) and generation length (25 years) from Zhivotovsky (2004). In addition the average ASD for all clusters was calculated for each marker thereby providing an estimate of individual marker mutation rates. These are also tabulated.

Results

Estimates of the TMRCA for the various STR derived clusters

The results are presented in table 1 below and this has a link to the complete table as well.

In the current context it must be remembered that these are estimates of the TMRCA for the cluster, they do not provide estimates of the time since divergence of various clusters, especially where these correspond to haplogroups. This is because narrow bottlenecks especially during the last ice age may mean the last common ancestor of a group as distinct from when the SNP derived haplogroup mutation occurred may differ. Similarly, the estimates are subject to considerable error. Perhaps what is surprising is the age estimates for the TMRCA of 7390 (1178) for R1b when it is known that the R1b1c SNP must have occurred before the start of the LGM approximately 20 Kyr bp. In part this has to do with sampling, because most individuals will be from the Iberian LGM refugia which rapidly expanded from a small group after the end of the last ice age about 9000 yrs bp. However, a further factor to be born in mind is the “effective” mutation rate used. It may still be too high.

R1bSTR19 aka North West Irish, or IMH

Perhaps the more surprising is the estimates fro the “well established” R1b STR clusters, R1bSTR19, R1bSTR22 and R1bSTR47. The Irish cluster has a particularly recent TMRCA of 3362 (SEM=609) yrs bp, and its best estimate is rather more recent than both the Scots and Irish clusters. The cluster itself is quite distinct from R1b, so this could infer that this group must have rapidly expanded from a small group around that time. Based on its current geographical distribution it is thought that this group represents the original hunter/gatherer population that inhabited Ireland after the LGM (~9000 yrs bp). The expansion may have been due to the rapid growth in population after the introduction of agriculture (~6500 yrs bp), or after metal working began ~4500 yrs bp when Beaker-ware also made its appearance. This cultural marker was most probably associated with the introduction of, horses and the Celtic language as well. More recent cultural events may also have contributed.

R1bSTR22 aka Frisian or Germanic

This cluster is exclusively associated with the S21+ SNP, although the SNP mutation itself is much older. It is thought to have originated in North Eastern Germany, although its centre of origin is disputed. Its estimated age of 5175 (SEM=813) yrs bp coincides best with the cultural changes that occurred around the time of the Kurgan expansion reached northern Europe.

R1bSTR47 aka Scots

At 4686 (SEM=903) yrs bp this group is intermediate between the Irish and Scots, but again may be related to a cultural shift with the introduction of the horse and altered agriculture. Again this group is thought to represent the indigenous hunter/gatherer inhabitants of Britain who settled after the LGM.

Perhaps another facet that is noticeable is most of the other R1b clusters are typically more diverse and have older TMRCA.

In summary, these estimates are presented as an independent chronological assessment of their ages. They do not provide estimates of their split from other groups (typically much older) rather the date of their initial rapid expansion. Observation suggests that the most important factor affecting the estimates is the “effective” mutation rate used.

Relative mutation rates of the various markers in the 37 FTDNA panel.

This is a fraught exercise but Table 2 is provided for consideration. The equations above suggest that the relative rates can be estimated via ASD, and for the sake of completeness these have been converted to mutation rates via Zhivotovsky et al. (2004) estimate. They are provided simply for comparison and no claims are made for their accuracy. For instance they are not weighted for the number of observations in each cluster (each is independent so the square root of numbers is an appropriate weighting factor). What is apparent is the rather dramatic difference in relative rate for individual markers.

Table 1. Phase 3 analysis cluster ASD estimates for each marker and average ASD and TMRCA estimates and relative ASD estimates for each marker. (Full table available at this link)

37 STR	n	aveASD	SEM(ASD)	TMRCA years^1	SEM(years)
AB	9	1.727	0.322	29512	5499
E	143	1.075	0.182	18361	3111
F	12	4.322	1.524	73856	26040
G	85	0.666	0.135	11385	2311
HO3	3	0.925	0.265	15805	4522
I	800	0.769	0.109	13147	1868
I1a	550	0.375	0.074	6401	1259
I1b	65	1.231	0.344	21042	5882
I1c	168	0.656	0.157	11214	2681
I?	19	0.936	0.154	15992	2635
Ix	17	0.451	0.091	7703	1557
K2	17	1.076	0.196	18394	3353
N	8	0.410	0.086	7007	1472
Q	24	1.021	0.168	17440	2866
R1a	196	0.517	0.078	8830	1341
R1b	2553	0.432	0.069	7390	1178
E	143	1.075	0.182	18361	3111
E3b	105	1.037	0.209	17721	3575
E3bSTR1	51	1.270	0.297	21708	5070
E3bSTR2	46	0.434	0.091	7419	1563
E3bSTR3	8	0.525	0.115	8977	1959
E3a	38	0.594	0.100	10156	1703
G	85	0.666	0.135	11385	2311
GG2	56	0.465	0.111	7947	1900
Gx	7	0.231	0.056	3940	951
GG2STR2	22	0.795	0.160	13587	2727
GG2(Fx)	8	0.248	0.139	4236	2381
I	800	0.769	0.109	13147	1868
I1c	168	0.656	0.157	11214	2681
I1cSTR1	103	0.467	0.118	7979	2023
IslesI1c	33	0.277	0.061	4736	1043
I1cSTR2	7	0.371	0.108	6334	1845
RootsI1c	18	0.430	0.081	7344	1376
I1cSTR3	7	0.428	0.138	7314	2360
I1b	65	1.231	0.344	21042	5882
I1bSTR1	39	0.552	0.095	9433	1631
WesternI1b	16	1.324	0.919	22618	15699
I1b2	10	0.995	0.602	16996	10289
I1a	550	0.375	0.074	6401	1259
I1aSTR1	42	0.378	0.087	6453	1492
I1aSTR2	56	0.344	0.071	5879	1210
I1aSTR3	37	0.278	0.076	4755	1304
I1aSTR4	69	0.269	0.062	4591	1052
I1aSTR5	71	0.285	0.081	4874	1376
I1aSTR6	59	0.317	0.050	5409	857
I1aSTR7	94	0.293	0.068	5008	1167
I1aSTR8	47	0.352	0.098	6007	1668
I1aSTR9	20	0.382	0.095	6529	1618
I1aSTR10	55	0.193	0.045	3292	761
Ix	17	0.451	0.091	7703	1557
I?	19	0.936	0.154	15992	2635
J	119	1.114	0.192	19030	3275
J2x	18	0.360	0.078	6151	1329
J1	21	1.161	0.339	19848	5796
J2e	19	0.362	0.069	6192	1173
J2STR4	8	0.812	0.165	13877	2816
J2	43	0.766	0.135	13081	2312
R1a	196	0.517	0.078	8830	1341
R1b	2553	0.432	0.069	7390	1178
R1bSTR1	57	0.340	0.071	5805	1210
R1bSTR2	56	0.380	0.090	6487	1544
R1bSTR3	53	0.236	0.038	4035	647
R1bSTR4	15	0.474	0.151	8108	2585
R1bSTR5	17	0.318	0.088	5437	1506
R1bSTR6	36	0.465	0.105	7939	1787
R1bSTR7	69	0.306	0.057	5222	978
R1bSTR8	29	0.481	0.133	8223	2273
R1bSTR9	57	0.380	0.118	6500	2023
R1bSTR10	60	0.309	0.095	5281	1622
R1bSTR11	53	0.288	0.079	4917	1343
R1bSTR12	41	0.378	0.098	6453	1667
R1bSTR13	14	0.302	0.093	5160	1593
R1bSTR14	43	0.333	0.071	5698	1220
R1bSTR15	60	0.308	0.058	5264	987
R1bSTR16	39	0.356	0.086	6089	1466
R1bSTR17	57	0.406	0.101	6943	1734
R1bSTR18	37	0.328	0.044	5613	750
R1bSTR19Irish	184	0.197	0.036	3362	609
R1bSTR20	37	0.331	0.061	5654	1049
R1bSTR21	26	0.319	0.060	5459	1019
R1bSTR22Frisian	117	0.303	0.048	5175	813
R1bSTR23	27	0.516	0.202	8811	3451
R1bSTR24	64	0.485	0.085	8296	1448
R1bSTR25	82	0.375	0.078	6410	1341
R1bSTR25a	35	0.355	0.079	6060	1354
R1bSTR26	26	0.494	0.160	8439	2741
R1bSTR27	55	0.345	0.066	5899	1124
R1bSTR28	56	0.471	0.068	8044	1154
R1bSTR29	19	0.308	0.083	5256	1416
R1bSTR30	33	0.341	0.068	5828	1163
R1bSTR31	9	0.394	0.096	6739	1646
R1bSTR32	43	0.379	0.063	6482	1084
R1bSTR33	21	0.286	0.061	4887	1049
R1bSTR34	45	0.385	0.064	6572	1092
R1bSTR35	27	0.385	0.092	6575	1578
R1bSTR36	39	0.369	0.072	6304	1228
R1bSTR37	64	0.389	0.070	6649	1201
R1bSTR38	33	0.300	0.075	5132	1277
R1bSTR39	69	0.429	0.078	7331	1329
R1bSTR40	65	0.309	0.078	5273	1338
R1bSTR41	31	0.374	0.076	6399	1302
R1bSTR42	79	0.422	0.079	7213	1348
R1bSTR43	97	0.391	0.062	6675	1059
R1bSTR44	91	0.289	0.046	4941	779
R1bSTR45	38	0.344	0.064	5870	1093
R1bSTR46	25	0.326	0.064	5579	1098
R1bSTR47Scots	133	0.274	0.053	4686	903
R1bSTR48	38	0.505	0.116	8623	1974
R1bSTR49	52	0.378	0.071	6454	1207
^1years were estimated by deviding ASD by the average generation length (25 yrs and the "corrected" estimated mutation rate)
The corrected mutation rate was derived as the average of Zhivotosky (2004) markers and his estimated mutation rate of 0.0007/25 year generation
scaled for the additional markers used in this analysis (DYS461 was assumed equivalent to the average) =2.09*0.0007

Table 2. Relative ASD values and estimated mutation rates for the 37 FTDNA panel. The mutation estimates are derived from Zhivotovsky et al. (2004) value derived from the shaded markers.

Marker	mean ASD	SEM ASD	mutation est	SEM mutation
393	0.158	0.019	0.000452	0.000055
390	0.354	0.045	0.001013	0.000129
19	0.203	0.024	0.000582	0.000069
391	0.190	0.018	0.000544	0.000052
385a	0.503	0.078	0.001442	0.000222
385b	0.917	0.164	0.002627	0.000469
426	0.044	0.014	0.000126	0.000040
388	0.186	0.048	0.000534	0.000137
439	0.404	0.036	0.001158	0.000103
389i	0.242	0.029	0.000693	0.000082
392	0.135	0.033	0.000388	0.000094
389ii*	0.326	0.042	0.000935	0.000121
458	0.819	0.060	0.002347	0.000171
459a	0.111	0.020	0.000317	0.000057
459b	0.127	0.018	0.000364	0.000050
455	0.052	0.018	0.000148	0.000052
454	0.057	0.017	0.000163	0.000047
447	0.733	0.136	0.002099	0.000391
437	0.193	0.069	0.000552	0.000198
448	0.248	0.041	0.000710	0.000118
449	1.318	0.129	0.003777	0.000369
464a	0.447	0.080	0.001280	0.000229
464b	0.468	0.058	0.001340	0.000166
464c	0.424	0.036	0.001215	0.000104
464d	0.370	0.040	0.001061	0.000115
460	0.248	0.020	0.000710	0.000057
H4	0.221	0.026	0.000633	0.000075
YCAIIa	0.557	0.251	0.001595	0.000718
YCAiib	0.497	0.077	0.001423	0.000221
456	0.737	0.146	0.002111	0.000419
607	0.446	0.083	0.001279	0.000237
576	1.114	0.091	0.003190	0.000260
570	1.025	0.100	0.002936	0.000286
CDYa	2.204	0.582	0.006313	0.001667
CDYb	2.003	0.252	0.005737	0.000723
442	0.727	0.375	0.002084	0.001075
438	0.113	0.017	0.000322	0.000048

References

Stumpf MP, Goldstein DB. 2001. Genealogical and evolutionary inference with the human Y chromosome.

Science 291:1738-42.

Takezaki N, Nei M. 1996. Genetic distances and reconstruction of phylogenetic trees from microsatellite DNA.

Genetics. 144:389-99

Zhivotovsky LA, Underhill PA, Cinnioglu C, Kayser M, Morar B, Kivisild T, Scozzari R, Cruciani F, Destro-Bisol G, Spedini G, Chambers GK, Herrera RJ, Yong KK, Gresham D, Tournev I, Feldman MW, Kalaydjieva L. 2004. The effective mutation rate at Y chromosome short tandem repeats, with application to human population-divergence time. Am J Hum Genet. 74:50-61