Limits...
The evolution of methods for establishing evolutionary timescales.

Donoghue PC, Yang Z - Philos. Trans. R. Soc. Lond., B, Biol. Sci. (2016)

Bottom Line: While the molecular clock can be used to extend the time estimates from fossil species to lineages not represented in the fossil record, fossils are the only source of information concerning absolute (geological) times in molecular dating analysis.While node-calibrations are often constructed by a crude assessment of the fossil evidence and thus involves arbitrariness, tip-calibrations may be too sensitive to the prior on divergence times or the branching process and influenced unduly affected by well-known problems of morphological character evolution, such as environmental influence on morphological phenotypes, correlation among traits, and convergent evolution in disparate species.We discuss the utility of time information from fossils in phylogeny estimation and the search for ancestors in the fossil record.This article is part of the themed issue 'Dating species divergences using rocks and clocks'.

View Article: PubMed Central - PubMed

Affiliation: School of Earth Sciences, University of Bristol, Life Sciences Building, Tyndall Avenue, Bristol BS8 1TQ, UK phil.donoghue@bristol.ac.uk.

No MeSH data available.


(a) Estimation of the absolute time (t) and rate (r) using the human and orangutan 12S rRNA genes from the mitochondrial genome. The data are summarized as x = 90 differences out of n = 948 aligned sites. The likelihood, calculated under the JC69 substitution model [86], depends on the distance d = 2tr only, but not t and r individually. The maximum-likelihood estimate of d under the JC69 model is 0.1015, with the 95% confidence (likelihood) interval to be (0.0817, 0.1245). All points on the red dashed line in (b) correspond to the same likelihood value and are maximum-likelihood estimates of t and r. To generate the posterior of t and r, we assign the prior t ∼ G(40, 40/15), with the prior mean to be 15 Myr and the 95% equal-tail interval to be (10.7, 20.0) Myr, and the rate prior r ∼ G(4, 800), with the mean to be 0.005 substitution per million years and the 95% interval to be (0.14, 1.10). (c) Relatively, our prior knowledge of the rate is less certain than that for time. Note that to obtain sensible posterior time estimates, it is important to constrain the time from both below and above in the prior (in this case, the time is weakly constrained to be in the range 10–20 Myr). (Online version in colour.)
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4920342&req=5

RSTB20160020F4: (a) Estimation of the absolute time (t) and rate (r) using the human and orangutan 12S rRNA genes from the mitochondrial genome. The data are summarized as x = 90 differences out of n = 948 aligned sites. The likelihood, calculated under the JC69 substitution model [86], depends on the distance d = 2tr only, but not t and r individually. The maximum-likelihood estimate of d under the JC69 model is 0.1015, with the 95% confidence (likelihood) interval to be (0.0817, 0.1245). All points on the red dashed line in (b) correspond to the same likelihood value and are maximum-likelihood estimates of t and r. To generate the posterior of t and r, we assign the prior t ∼ G(40, 40/15), with the prior mean to be 15 Myr and the 95% equal-tail interval to be (10.7, 20.0) Myr, and the rate prior r ∼ G(4, 800), with the mean to be 0.005 substitution per million years and the 95% interval to be (0.14, 1.10). (c) Relatively, our prior knowledge of the rate is less certain than that for time. Note that to obtain sensible posterior time estimates, it is important to constrain the time from both below and above in the prior (in this case, the time is weakly constrained to be in the range 10–20 Myr). (Online version in colour.)

Mentions: The most serious problem facing tip-calibration may be the extreme sensitivity of the posterior time estimates to the prior of divergence times specified by the branching process. Because the sequence data provide information about distances only, resolution of the sequence distance into absolute time and rate relies entirely on the priors on time and rate (figure 4). Most tip-calibration methods require a bound or prior on the age of the root for extant species, e.g. [16], but no calibrations are applied on the ages of other internal nodes. Thus, node ages are bounded by the ages of the fossil tips, because ancestral nodes cannot be younger than their descendent fossil tips, while there is otherwise effectively no constraint on the ages of clades except for the prior on the root age. In other words, there are multiple forces pushing up the node ages, but almost no force pushing them down. It is left to the divergence time prior or the branching-process model to keep the node ages on the tree within reasonable bounds, and that may prove to be too much burden on the time prior. A dozen or so initial studies applying the total-evidence dating approach have produced ancient time estimates, older even than those derived from the use of node-calibrations [59]. This is remedied by incorporating the FBD model into total-evidence dating, replacing the original uniform tree prior and making use of morphological character data in resolving the affinity of fossil taxa [81,82]. However, details of the FBD prior, such as the assumed sampling regime, can have a strong influence on divergence time estimates [81].Figure 4.


The evolution of methods for establishing evolutionary timescales.

Donoghue PC, Yang Z - Philos. Trans. R. Soc. Lond., B, Biol. Sci. (2016)

(a) Estimation of the absolute time (t) and rate (r) using the human and orangutan 12S rRNA genes from the mitochondrial genome. The data are summarized as x = 90 differences out of n = 948 aligned sites. The likelihood, calculated under the JC69 substitution model [86], depends on the distance d = 2tr only, but not t and r individually. The maximum-likelihood estimate of d under the JC69 model is 0.1015, with the 95% confidence (likelihood) interval to be (0.0817, 0.1245). All points on the red dashed line in (b) correspond to the same likelihood value and are maximum-likelihood estimates of t and r. To generate the posterior of t and r, we assign the prior t ∼ G(40, 40/15), with the prior mean to be 15 Myr and the 95% equal-tail interval to be (10.7, 20.0) Myr, and the rate prior r ∼ G(4, 800), with the mean to be 0.005 substitution per million years and the 95% interval to be (0.14, 1.10). (c) Relatively, our prior knowledge of the rate is less certain than that for time. Note that to obtain sensible posterior time estimates, it is important to constrain the time from both below and above in the prior (in this case, the time is weakly constrained to be in the range 10–20 Myr). (Online version in colour.)
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4920342&req=5

RSTB20160020F4: (a) Estimation of the absolute time (t) and rate (r) using the human and orangutan 12S rRNA genes from the mitochondrial genome. The data are summarized as x = 90 differences out of n = 948 aligned sites. The likelihood, calculated under the JC69 substitution model [86], depends on the distance d = 2tr only, but not t and r individually. The maximum-likelihood estimate of d under the JC69 model is 0.1015, with the 95% confidence (likelihood) interval to be (0.0817, 0.1245). All points on the red dashed line in (b) correspond to the same likelihood value and are maximum-likelihood estimates of t and r. To generate the posterior of t and r, we assign the prior t ∼ G(40, 40/15), with the prior mean to be 15 Myr and the 95% equal-tail interval to be (10.7, 20.0) Myr, and the rate prior r ∼ G(4, 800), with the mean to be 0.005 substitution per million years and the 95% interval to be (0.14, 1.10). (c) Relatively, our prior knowledge of the rate is less certain than that for time. Note that to obtain sensible posterior time estimates, it is important to constrain the time from both below and above in the prior (in this case, the time is weakly constrained to be in the range 10–20 Myr). (Online version in colour.)
Mentions: The most serious problem facing tip-calibration may be the extreme sensitivity of the posterior time estimates to the prior of divergence times specified by the branching process. Because the sequence data provide information about distances only, resolution of the sequence distance into absolute time and rate relies entirely on the priors on time and rate (figure 4). Most tip-calibration methods require a bound or prior on the age of the root for extant species, e.g. [16], but no calibrations are applied on the ages of other internal nodes. Thus, node ages are bounded by the ages of the fossil tips, because ancestral nodes cannot be younger than their descendent fossil tips, while there is otherwise effectively no constraint on the ages of clades except for the prior on the root age. In other words, there are multiple forces pushing up the node ages, but almost no force pushing them down. It is left to the divergence time prior or the branching-process model to keep the node ages on the tree within reasonable bounds, and that may prove to be too much burden on the time prior. A dozen or so initial studies applying the total-evidence dating approach have produced ancient time estimates, older even than those derived from the use of node-calibrations [59]. This is remedied by incorporating the FBD model into total-evidence dating, replacing the original uniform tree prior and making use of morphological character data in resolving the affinity of fossil taxa [81,82]. However, details of the FBD prior, such as the assumed sampling regime, can have a strong influence on divergence time estimates [81].Figure 4.

Bottom Line: While the molecular clock can be used to extend the time estimates from fossil species to lineages not represented in the fossil record, fossils are the only source of information concerning absolute (geological) times in molecular dating analysis.While node-calibrations are often constructed by a crude assessment of the fossil evidence and thus involves arbitrariness, tip-calibrations may be too sensitive to the prior on divergence times or the branching process and influenced unduly affected by well-known problems of morphological character evolution, such as environmental influence on morphological phenotypes, correlation among traits, and convergent evolution in disparate species.We discuss the utility of time information from fossils in phylogeny estimation and the search for ancestors in the fossil record.This article is part of the themed issue 'Dating species divergences using rocks and clocks'.

View Article: PubMed Central - PubMed

Affiliation: School of Earth Sciences, University of Bristol, Life Sciences Building, Tyndall Avenue, Bristol BS8 1TQ, UK phil.donoghue@bristol.ac.uk.

No MeSH data available.