A comparison of methods for estimating substitution rates from ancient DNA sequence data

Table 2 Results from analyses of six time-structured mitogenomic data sets

Species	Clock model^a	Tree prior^a	Phylo-temporal clustering^b (P-value)	Date-randomization test^c
Species	Clock model^a	Tree prior^a	Phylo-temporal clustering^b (P-value)	CR1	CR2
Adélie penguin	Strict	Constant size	0.079	Fail	Fail
Brown/polar bear	Strict	Constant size	0.168	Pass	Pass
Dog	Relaxed	Constant size	0.006	Pass	Pass
Horse	Strict	Constant size	< 0.001	Pass	Pass
Modern human	Strict	Skyride	0.166	Pass	Pass
Woolly mammoth	Relaxed	Constant size	0.075	Pass	Pass

^aClock models and tree priors were compared using marginal likelihoods estimated by stepping-stone sampling. Marginal likelihoods are given in Additional file 2: Table S5
^bP-values below 0.05 indicate that sequences with similar ages tend to be clustered together in the phylogenetic tree
^cWe considered two criteria that have been proposed for the date-randomization test, CR1 and CR2 [27]. These criteria are described in the Methods. Our results are based on 20 date-randomized replicates

ISSN: 2730-7182