The DOI matches its Crossref record; 46 of the 46 references it cites were found in a registry or an index; none of the 27 sources its claims cite is checked for support yet.
A General Comparison of Relaxed Molecular Clock Models
Open accessAbstract
Several models have been proposed to relax the molecular clock in order to estimate divergence times. However, it is unclear which model has the best fit to real data and should therefore be used to perform molecular dating. In particular, we do not know whether rate autocorrelation should be considered or which prior on divergence times should be used. In this work, we propose a general bench mark of alternative relaxed clock models. We have reimplemented most of the already existing models, including the popular lognormal model, as well as various prior choices for divergence times (birth-death, Dirichlet, uniform), in a common Bayesian statistical framework. We also propose a new autocorrelated model, called the "CIR" process, with well-defined stationary properties. We assess the relative fitness of these models and priors, when applied to 3 different protein data sets from eukaryotes, vertebrates, and mammals, by computing Bayes factors using a numerical method called thermodynamic integration. We find that the 2 autocorrelated models, CIR and lognormal, have a similar fit and clearly outperform uncorrelated models on all 3 data sets. In contrast, the optimal choice for the divergence time prior is more dependent on the data investigated. Altogether, our results provide useful guidelines for model choice in the field of molecular dating while opening the way to more extensive model comparisons.
Overview
AI snapshot
Written from the paper’s full text.
TL;DR
The paper benchmarks relaxed molecular clock models and divergence-time priors in a common Bayesian framework, adding a new autocorrelated CIR process. Across three eukaryotic protein data sets, autocorrelated rate models fit better than uncorrelated ones, while the best divergence-time prior depends on the data set.
Key findings
- The strict clock was strongly rejected relative to the unconstrained model on all data sets, with log Bayes factors ranging from about −157 to −3,072.
- Autocorrelated models, LOGN and CIR, had similar fit and clearly outperformed uncorrelated models (UEXP, UGAM, WN) on nearly all data sets; LOGN was slightly favored except on MITO100, where they were equivalent, and MITO150, where CIR was significantly preferred.
- The Dirichlet prior was generally close to the uniform prior, whereas the birth-death prior was rejected in most cases (ln BF < −6) but strongly favored on MITO100 and MITO150 (ln BF > 45).
- Dense taxonomic sampling was important for distinguishing autocorrelated from uncorrelated models; with only 10 taxa, the uncorrelated models were comparable to the autocorrelated ones.
Methods
- The authors reimplemented seven relaxed-clock models—STRICT, UEXP, UGAM, WN, LOGN, CIR, and an unconstrained reference—and combined them with uniform, Dirichlet, and birth-death priors on divergence times.
- Data sets were MOC (129 genes, 30,399 positions, 36 eukaryotes), MAM (3 nuclear genes, 3,700 positions, 60 mammals), and MITO (11 mitochondrial protein-coding genes, 3,220 positions, 155 osteichthyans plus 5 chondrichthyans, with ingroup subsamples of 10, 50, 100, and 150 taxa).
- Posterior sampling used MCMC in an extension of PhyloBayes under the normal approximation, and model fit was compared by Bayes factors computed with thermodynamic integration.
Limitations
- The paper states that no fossil calibrations were used, so only relative divergence times were examined and no conclusions about absolute dates or root age are provided.
- The analyses fixed the tree topology, used the normal approximation exclusively, and did not account for heterogeneity across sites or data partitions.
- The authors note that the reasons for the birth-death prior’s data-dependent performance are unclear, that sampling error in Bayes factor estimates was on the order of a few log units, and that substitution-model effects and other relaxed-clock models were not investigated.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
0 of 27 cited sources checked · 0 supported · 0 flagged
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In contrast, the optimal choice for the divergence time prior is more dependent on the data investigated.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our results provide strong evidence in favor of autocorrelated models such as the lognormal and the CIR models, in particular under rich taxonomic sampling.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We see that for all data sets, the molecular clock is overwhelmingly rejected in favor of the unconstrained model, with Bayes factors ranging from 157 to 3,072.”
Load-bearingStrength:boosted· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“For almost all data sets, the Bayes factors of the relaxed but completely uncorrelated models (WN, UGAM, UEXP) are significantly inferior to those of the autocorrelated relaxed clock models (LOGN and CIR)”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Only when the number of taxa is small (MITO10) are the uncorrelated models comparable to the autocorrelated models.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“under dense taxonomic sampling (MITO150), the UExp and the WN models were so bad that they performed even worse than the unconstrained model.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“For most data sets, the 2 autocorrelated models, LOGN and CIR, have comparable Bayes factors.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“on which it is strongly favored over the 2 alternatives (ln BF . 45).”
Load-bearingStrength:boosted· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This trend is also observed under the birth-death prior (fig. 3 b ) but is less pronounced.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesA result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We observe in both cases that, for most nodes, the posterior standard errors (SEs) are much smaller than the prior SEs”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passes
- It may state two separate claims as one. 28
- Probably not a substantial claim. 14
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We find that the 2 autocorrelated models, CIR and lognormal, have a similar fit and clearly outperform uncorrelated models on all 3 data sets.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
In the paper
“Interestingly, UGAM, the uncorrelated gamma model (Drummond et al. 2006), appears to be the best among the 3 branchwise models, although it is less fit than LOGN and CIR.”
What the sources say
- Not checked yetcites Drummond et al. 2006
Strength:hedged· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“LogN is slightly favored in all cases, except on MITO100 (where the 2 models have equivalent scores) and on MITO150 (where CIR is significantly preferred over LOGN, table 1, last row).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Nodes near the root are an exception, however, and more strikingly so under the uniform prior: for old nodes, and compared with the birth-death prior, the uniform prior appears to be both more informative and more in contradiction with posterior intervals.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This is particularly true for the nodes closer to the root (ages . 50), for which there is virtually no agreement between CIR and STRICT (fig. 2 a ), whereas some overlapcanbeseeninthecredibility intervals obtained under Strict and WN (fig. 2 d ).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“On the other hand, the number of departing nodes went up to 100 between STRICT and CIR and to 93 between STRICT and LOGN.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“In contrast, the other 2 data sets (MOC, MAM) display a tree shape characterized by an early interphyletic or interordinal diversification, combined with a low phylogenetic redundancy (few closely related taxa).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This contrasts with the linearly increasing variance of the lognormal model, which does not have a stationary behavior.”
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We also propose a new autocorrelated model, called the ''CIR'' process, with well-defined stationary properties.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
None of the sources cited for this claim have been checked yet.
In the paper
“This makes more sense as it preserves the positivity of the rate process and avoids the systematic bias (rates tending to be higher near the root) that had been pointed out in the model of Aris-Brosou and Yang (2002) by Welch et al. (2005).”
What the sources say
- Not checked yetcites Aris-Brosou and Yang (2002)
- Not checked yetcites Welch et al. (2005)
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We see here that the expected value of R ( t ) tends to 1 for large values of t . More precisely, it follows an exponential decay of parameter h ; h is, thus, the autocorrelation parameter of the CIR process (or equivalently, 1/ h is the decorrelation time of the process).”
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“There is, however, a lack of identifiability between the 3 parameters, as evaluating the prior at ( k , l , q ) gives the same value as if evaluated at k þ k ; l þ k ; kq k þ k : We avoid such problems by restraining the parameter number to 2, with p 1 5 k -l and p 2 5 kq as our new parameterization.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“We see here that the expected value of R ( t ) tends to 1 for large values of t . More precisely, it follows an exponential decay of parameter h ; h is, thus, the autocorrelation parameter of the CIR process (or equivalently, 1/ h is the decorrelation time of the process).”
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Finally, for small values of t , the variance of R ( t ) is proportional to R (0) r 2 , so that r will be designated as the ''short-term'' variance of the process.”
Strength:plain· not yet calibratedCausal:correlational· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In practice, the main difference between the WN model and the UGAM model lies in the variance of B i : whereas it is quadratic with time in the UGAM case, the WN has a linear variance, accounting for the fact that, by an averaging effect, the mean rate of an uncorrelated process is expected to have a smaller variance over longer branches.”
Strength:plain· not yet calibratedCausal:direct· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes? Load-bearing: unclear✓ Based: passesIt may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“However, the success of a model on simulated data may not always be representative of the model's relevance to the analysis of real data, which is more directly assessed by measuring the statistical fit of the model to the data at hand.”
Load-bearingStrength:hedged· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The CIR process mainly differs from the log normal in the fact that it possesses a stationary distribution.”
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
In the paper
“However, they are regarded as very controversial (see answers to Aitkin 1991).”
What the sources say
- Not checked yetcites Aitkin 1991
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
None of the sources cited for this claim have been checked yet.
In the paper
“One of these techniques, called thermodynamic integration (Ogata 1989), has been recently used in several phylogenetic studies (Lartillot and Philippe 2004, 2006; Blanquart and Lartillot 2006; Rodrigue et al. 2006).”
What the sources say
- Not checked yetcites Ogata 1989
- Not checked yetcites Lartillot and Philippe 2004, 2006
- Not checked yetcites Blanquart and Lartillot 2006
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Bayes factors present several advantages: they implicitly penalize models with higher dimensionality; furthermore, by using one single model as a global reference, they allow for a general comparison among an arbitrary number of models that are not necessarily nested.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“The latter, CIR, is similar to the Ornstein-Uhlenbeck (OU) model (Aris-Brosou and Yang 2003) but without many of the shortcomings of the OU identified by Welch et al. (2005).”
What the sources say
- Not checked yetcites Aris-Brosou and Yang 2003
- Not checked yetcites Welch et al. (2005)
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Thus, unlike WN, UGAM has a dispersion index increasing with t , which is a typical feature of autocorrelated models.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“In the present case, the theoretical dispersion index is equal to 1 þ r for WN and 1 þ r t for UGAM.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“A possible explanation is that LOGN may perform better on smaller data sets because it is simpler (it has only one parameter, controlling the autocorrelation).”
Strength:hedged· not yet calibratedCausal:conditional· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
In the paper
“In general, this absence of stationarity for a stochastic process is seen as a potential problem (Lepage et al. 2006).”
What the sources say
- Not checked yetcites Lepage et al. 2006
Strength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
The paper states this without a source or data behind it here, so there is nothing to check it against. The quote shows it is in the paper.
In the paper
“Note that, in the long term, the CIR has a well-defined behavior, that is, a stationary distribution, whereas in the LOGN model the rate has a variance that increases linearly with time.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“No fossil records were used in this study, so dates are displayed on a relative scale where all leaves have an age of zero and the root an age of 100.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“given that large discrepancies in the relative node ages between 2 models will probably imply large discrepancies in the absolute dates as well”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✗ Based: failsProbably not a substantial claim. The paper may give no reason why it holds.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“the lower fit of the birth-death prior for small data sets may be the result of a dimensional penalty (the birth-death has one more parameter than the Dirichlet).”
Strength:hedged· not yet calibratedCausal:conditional· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Alternatively, this could be due to the fact that the trees of MITO100 and 150 have long basal branches, separating densely sampled vertebrate classes, a pattern that may be particularly well fitted by a birth-death process.”
Strength:hedged· not yet calibratedCausal:conditional· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it. It may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“As a consequence, this study will not provide any information on the effect of models on absolute dates and, in particular, on the root age.”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Altogether, autocorrelation can be detected in 2 different ways: between branches, it can be directly observed. Within branches, it can only be indirectly inferred, through its net effect on the dispersion index.”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“This also makes UGam more in accordance with experimentally observed values for the dispersion index (Ohta 1995; Bickel and West 1998), which may explain the better fit we obtain for UGAM, compared with WN.”
What the sources say
- Not checked yetcites Ohta 1995
- Not checked yetcites Bickel and West 1998
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“This may be simply due to the fact that our priors on hyperparameters are more vague.”
Strength:hedged· not yet calibratedCausal:conditional· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✗ Load-bearing: fails✓ Based: passesProbably not a substantial claim. The paper’s conclusions may not depend on it.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In summary, our observations illustrate the high sensitivity of the estimated dates to the relaxation model used (fig. 2) and to a lesser extent also to the prior on divergence times (fig. 3).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
None of the sources cited for this claim have been checked yet.
In the paper
“Another point raised by Welch et al. (2005) is that the OU process, as implemented in Aris-Brosou and Yang (2003), imposes a general trend in the rate history.”
What the sources say
- Not checked yetcites Welch et al. 2005
- Not checked yetcites Aris-Brosou and Yang 2003
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
A conclusion the paper draws from its own results — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In any case, and irrespective of the exact prior used, the data seem to at least partially override the prior information as it provides age estimates that are quite robust (although not totally insensitive) to the exact choice of the prior (fig. 3 c and d ).”
Tests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesIt may state two separate claims as one.
What the paper did — it cites no source for this, so there is none to check. The quote shows it is in the paper.
In the paper
“Our conclusions are exclusively based on relative age estimates”
Load-bearingStrength:plain· not yet calibratedTests✓ Asserted: passes✓ Checkable: passes✓ Load-bearing: passes✓ Based: passesProbably not a substantial claim.
The authors state this as an assumption.
The rate process is Markovian.
“Assuming that the rate process is Markovian, the joint probability density of the values of the process at the nodes is given by:”
The paper relies on this without support.
The strict molecular clock is not biologically realistic and substitution rates change over time, requiring explicit accounting in molecular dating.
“However, it has become clear that the strict molecular clock hypothesis is not biologically realistic and that the rate at which substitutions accumulate with time is subject to changes that need to be explicitly accounted for in molecular dating techniques.”
The paper relies on this without support.
Few attempts have been made to compare the performance of available clock models.
“Remarkably, there have been few attempts thus far to compare the performances of available clock models.”
The paper relies on this without support.
The random subsampling of osteichthyans in the MITO data sets produces taxon samplings that differ only in number of taxa, so differences in model performance can be attributed to taxonomic sampling density.
“considered 4 different taxon samplings for the ingroup, obtained by randomly subsampling 10, 50, 100, and 150 osteichthyans.”
The paper relies on this without support.
The data sets used are representative of real sequence evolution, so conclusions about the strict molecular clock and autocorrelation generalize to real sequences.
“indicating that the strict molecular clock is not adapted to describing the evolution of real sequences.”
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The cited source says what the claim attributes to it.
In the citing paper
“Bayes factors have been used to decide between competing clock models (such as the IR and AR models, e.g., [55]) in phylogenetic analysis.”
From Uncertainty in the Timing of Origin of Animals and the Limits of Precision in Molecular Timescales (2015) · cites this paper as [55]
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
Each contradiction between two passages listed here was confirmed by an independent check.
Autocorrelated models do not outperform UGAM on MITO10
MinorThe abstract says both autocorrelated models clearly outperform uncorrelated models on all three data sets. For MITO10, table 1 gives UGAM a higher log Bayes factor (3.9-4.0) than either LOGN (3.2-3.7) or CIR (2.5-3.0).
“We find that the 2 autocorrelated models, CIR and lognormal, have a similar fit and clearly outperform uncorrelated models on all 3 data sets.”
“| UGAM | [26.6; 27.2] | [53.0; 55.7] | [3.9; 4.0] | [14.6;16.8] | [34.6; 36.2] | [4.5; 12.0] | | WN | [23.9; 24.6] | [48.8; 49.3] | [1.3; 1.6] | [9.0; 9.5] | [20.4; 21.3] | [ 4.3; 3.0] | | LOGN | [40.8; 43.1] | [61.1; 64.1] | [3.2; 3.7] | [23.5; 26.0] | [48.3; 56.8] | [36.7; 44.0] | | CIR | [36.7; 40.5] | [58.9; 60.6] | [2.5; 3.0] | [20.0; 22.8] | [51.6; 57.4] | [51.4; 60.2] |”
MITO100 birth-death Bayes factor is not above 45
MinorThe stated threshold holds for MITO150, but not MITO100. Table 2 reports MITO100 log Bayes factors of 13.3-17.4 for CIR and 13.3-17.1 for LOGN under the birth-death prior.
“except on MITO100 and MITO150, on which it is strongly favored over the 2 alternatives (ln BF . 45).”
“| CIR þ BD | [ 10.2; 8.1] | [ 9.6; 6.3] | [ 13.9; 12.9] | [ 8.8; 6.9] | [13.3; 17.4] | [46.4; 53.8] | | LOGN þ BD | [ 20.5; 19.0] | [ 10.3; 6.2] | [ 13.9; 13.1] | [ 8.4; 7.2] | [13.3; 17.1] | [46.6; 55.0] |”
Figure 1b uses a different prior from the cited model
MinorThe text points to figure 1b when discussing the CIR plus birth-death model, but the figure caption identifies its prior as uniform.
“We therefore checked that the optimal model (CIR þ BD) found under MITO150 (see fig. 1 b ) did not produce such patterns (fig. 4).”
“( b ) Relative divergence times under the MITO150 data set, under the CIR process and a uniform prior on divergence times.”
Citations
Cites
1 retracted
RETRACTED ARTICLE: TREEFINDER: a powerful graphical analysis environment for molecular phylogenetics
Jobb, G., von Haeseler, A., & Strimmer, K. (2004) · BMC Evolutionary Biology · doi:10.1186/1471-2148-4-18 · Cited by 1,109
RetractedEquation of State Calculations by Fast Computing Machines
Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., Teller, A. H., & Teller, E. (1953) · The Journal of Chemical Physics · doi:10.1063/1.1699114 · Cited by 37,192
MRBAYES: Bayesian inference of phylogenetic trees.
Huelsenbeck, J. P., & Ronquist, F. (2001) · Bioinformatics · doi:10.1093/bioinformatics/17.8.754 · Cited by 21,937
Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables
Owen, D. B., Abramowitz, M., & Stegun, I. A. (1965) · Technometrics · doi:10.2307/1266136 · Cited by 16,379
Monte Carlo sampling methods using Markov chains and their applications.
Hastings, W. K. (1970) · Biometrika · doi:10.1093/biomet/57.1.97 · Cited by 15,195
Evolutionary trees from DNA sequences: A maximum likelihood approach.
Felsenstein, J. (1981) · Journal of Molecular Evolution · doi:10.1007/bf01734359 · Cited by 14,571
A Theory of the Term Structure of Interest Rates
Cox, J. C., Ingersoll, J. E., & Ross, S. A. (1985) · Econometrica · doi:10.2307/1911242 · Cited by 8,591
The rapid generation of mutation data matrices from protein sequences
Jones, D. T., Taylor, W. R., & Thornton, J. M. (1992) · Computer applications in the biosciences · doi:10.1093/bioinformatics/8.3.275 · Cited by 7,003
Relaxed Phylogenetics and Dating with Confidence
Drummond, A. J., Ho, S. Y. W., Phillips, M. J., & Rambaut, A. (2006) · PLoS Biology · doi:10.1371/journal.pbio.0040088 · Cited by 6,463
Reversible Jump Markov Chain Monte Carlo Computation and Bayesian Model Determination
Green, P. J. (1995) · Biometrika · doi:10.1093/biomet/82.4.711 · Cited by 5,941
Sanderson, M. J. (2002) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a003974 · Cited by 1,977
A Bayesian Mixture Model for Across-Site Heterogeneities in the Amino-Acid Replacement Process
Lartillot, N., & Philippe, H. (2004) · Molecular Biology and Evolution · doi:10.1093/molbev/msh112 · Cited by 1,516
Self-citationEstimating the rate of evolution of the rate of molecular evolution
Thorne, J. L., Kishino, H., & Painter, I. S. (1998) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a025892 · Cited by 1,317
Bayesian phylogenetic inference using DNA sequences: a Markov Chain Monte Carlo Method
Yang, Z., & Rannala, B. (1997) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a025811 · Cited by 1,274
The causes of molecular evolution
Gillespie, J. H. (1992) · Choice Reviews Online · doi:10.5860/choice.30-2064 · Cited by 1,214
A nonparametric approach to estimating divergence times in the absence of rate constancy.
Sanderson, M. J. (1997) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a025731 · Cited by 1,156 · cited 2×
“approaches ( Sanderson, 1997, 2002 ), local clocks (Yoder and Yang, 2000), and Bayesian parametric”
MethodDivergence time and evolutionary rate estimation with multilocus data.
Thorne, J. L., & Kishino, H. (2002) · Systematic Biology · doi:10.1080/10635150290102456 · Cited by 962
Yang, Z., & Rannala, B. H. (2005) · Molecular Biology and Evolution · doi:10.1093/molbev/msj024 · Cited by 917
Placental mammal diversification and the Cretaceous–Tertiary boundary
Springer, M. S., Murphy, W. J., Eizirik, E., & O’Brien, S. J. (2003) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.0334222100 · Cited by 846
Phylogenetic test of the molecular clock and linearized trees.
Takezaki, N., Rzhetsky, A., & Nei, M. (1995) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a040259 · Cited by 826
Bayesian Selection of Continuous-Time Markov Chain Evolutionary Models
Suchard, M. A., Weiss, R. E., & Sinsheimer, J. S. (2001) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a003872 · Cited by 791
Computing Bayes factors using thermodynamic integration.
Lartillot, N., & Philippe, H. (2006) · Systematic Biology · doi:10.1080/10635150500433722 · Cited by 668
Self-citationThe timing of eukaryotic evolution: Does a relaxed molecular clock reconcile proteins and fossils?
Douzery, E. J. P., Snell, E. A., Bapteste, E., Delsuc, F., & Philippe, H. (2004) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.0403984101 · Cited by 638
Self-citationPerformance of a divergence time estimation method under a probabilistic model of rate evolution.
Kishino, H., Thorne, J. L., & Bruno, W. J. (2001) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a003811 · Cited by 633
Estimation of primate speciation dates using local molecular clocks.
Yoder, A. D., & Yang, Z. (2000) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a026389 · Cited by 494
Estimating metazoan divergence times with a molecular clock
Peterson, K. J., Lyons, J. B., Nowak, K. S., Takacs, C. M., Wargo, M. J., & McPeek, M. A. (2004) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.0401670101 · Cited by 461
Some Tests of Significance, Treated by the Theory of Probability
Jeffreys, H. (1935) · Mathematical Proceedings of the Cambridge Philosophical Society · doi:10.1017/s030500410001330x · Cited by 432
Aitkin, M. A. (1991) · Journal of the Royal Statistical Society Series B (Statistical Methodology) · doi:10.1111/j.2517-6161.1991.tb01812.x · Cited by 362
A Compound Poisson Process for Relaxing the Molecular Clock
Huelsenbeck, J. P., Larget, B. R., & Swofford, D. L. (2000) · Genetics · doi:10.1093/genetics/154.4.1879 · Cited by 348
Synonymous and nonsynonymous substitutions in mammalian genes and the nearly neutral theory
Ohta, T. (1995) · Journal of Molecular Evolution · doi:10.1007/bf00166595 · Cited by 304
Arrival and Diversification of Caviomorph Rodents and Platyrrhine Primates in South America
Poux, C., Chevret, P., Huchon, D., JONG, W. W. D., & Douzery, E. (2006) · Systematic Biology · doi:10.1080/10635150500481390 · Cited by 240
Aris-Brosou, S., & Yang, Z. (2002) · Systematic Biology · doi:10.1080/10635150290102375 · Cited by 230 · cited 2×
“This may be a true mathematical independence between the instantaneous values taken by the process and its integral over finite time intervals (e.g., the white-noise model described below) or, alternatively, just a matter of practical convenience (Aris-Brosou and Yang 2002; Drummond et al. 2006).”
BackgroundSmith, A. B., Pisani, D., Mackenzie‐Dodds, J. A., Stockley, B., Webster, B. L., & Littlewood, D. T. J. (2006) · Molecular Biology and Evolution · doi:10.1093/molbev/msl039 · Cited by 202
Dating the Time of Origin of Major Clades: Molecular Clocks and the Fossil Record
Smith, A. B., & Peterson, K. J. (2002) · Annual Review of Earth and Planetary Sciences · doi:10.1146/annurev.earth.30.091201.140057 · Cited by 193
Ho, S. Y. W. (2005) · Molecular Biology and Evolution · doi:10.1093/molbev/msi125 · Cited by 190 · cited 1×
“In some cases, models were validated by their ability to properly retrieve already known rates and dates from simulated data (Thorne and Kishino 2002; Ho et al. 2005; Drummond et al. 2006).”
MethodAris-Brosou, S. (2003) · Molecular Biology and Evolution · doi:10.1093/molbev/msg226 · Cited by 164
A Monte Carlo method for high dimensional integration
Ogata, Y. (1989) · Numerische Mathematik · doi:10.1007/bf01406511 · Cited by 161
Blanquart, S. (2006) · Molecular Biology and Evolution · doi:10.1093/molbev/msl091 · Cited by 142
Growing up with dinosaurs: molecular dates and the mammalian radiation
Bromham, L., Phillips, M. J., & Penny, D. (1999) · Trends in Ecology & Evolution · doi:10.1016/s0169-5347(98)01507-9 · Cited by 127
Pérez‐Losada, M., Høeg, J. T., & Crandall, K. A. (2004) · Systematic Biology · doi:10.1080/10635150490423458 · Cited by 125
Estimating Divergence Times in the Presence of an Overdispersed Molecular Clock
Cutler, D. J. (2000) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a026264 · Cited by 86
Assessing Site-Interdependent Phylogenetic Models of Sequence Evolution
Rodrigue, N., Philippe, H., & Lartillot, N. (2006) · Molecular Biology and Evolution · doi:10.1093/molbev/msl041 · Cited by 74
Self-citationMolecular Clocks Do Not Support the Cambrian Explosion
Blair, J. E., & Hedges, S. B. (2005) · Molecular Biology and Evolution · doi:10.1093/molbev/msi039 · Cited by 73
Continuous and tractable models for the variation of evolutionary rates
Lepage, T., Lawi, S., Tupper, P. F., & Bryant, D. (2006) · Mathematical Biosciences · doi:10.1016/j.mbs.2005.11.002 · Cited by 57
Self-citationMolecular dates for the “cambrian explosion”: The influence of prior assumptions.
Welch, J. J., Fontanillas, E., & Bromham, L. (2005) · Systematic Biology · doi:10.1080/10635150590947212 · Cited by 53 · cited 13×
“Another point raised by Welch et al. (2005) is that the OU process, as implemented in Aris-Brosou and Yang (2003), imposes a general trend in the rate history.”
BackgroundResultInfluentialBickel, D. R., & West, B. J. (1998) · Molecular Biology and Evolution · doi:10.1093/oxfordjournals.molbev.a026012 · Cited by 15
Flagged first, then by citations
Research works printed but not linked: 1
- Kimura M. 1983. The neutral allele theory of molecular evolution. Cambridge: Cambridge University Press.
Cited by
Sorted by citations
MrBayes 3.2: Efficient Bayesian Phylogenetic Inference and Model Choice Across a Large Model Space
Ronquist, F., Teslenko, M., van der Mark, P., Ayres, D. L., Darling, A. E., Höhna, S., Larget, B. R., Liu, L., Suchard, M. A., & Huelsenbeck, J. P. (2012) · Systematic Biology · doi:10.1093/sysbio/sys029 · Cited by 28,399
“Three different relaxed clock models are available: the Compound Poisson Process (CPP; Huelsenbeck et al. 2000), the Thorne– Kishino 2002 (TK02; Thorne and Kishino 2002), and the Independent Gamma Rate (IGR; Lepage et al. 2007) models.”
MethodPhyloBayes 3: a Bayesian software package for phylogenetic reconstruction and molecular dating
Lartillot, N., Lepage, T., & Blanquart, S. (2009) · Bioinformatics · doi:10.1093/bioinformatics/btp368 · Cited by 1,297
“Currently available molecular dating software programs propose either branch i.i.d. models of rate variations (Akerborg et al., 2008; Drummond et al., 2006), or autocorrelated model of rate variations (Kishino et al., 2001; Lepage et al., 2007; Rannala and Yang, 2007).”
Self-citationErwin, D. H., Laflamme, M., Tweedt, S. M., Sperling, E. A., Pisani, D., & Peterson, K. J. (2011) · Science · doi:10.1126/science.1206375 · Cited by 1,336
Upham, N. S., Esselstyn, J. A., & Jetz, W. (2019) · PLoS Biology · doi:10.1371/journal.pbio.3000494 · Cited by 1,235
Baele, G., Lemey, P., Bedford, T., Rambaut, A., Suchard, M. A., & Alekseyenko, A. V. (2012) · Molecular Biology and Evolution · doi:10.1093/molbev/mss084 · Cited by 1,145
“Hence, the approaches of Lepage et al. (2007) and Friel and Petitt (2008) both place most of the power values at points where the power posterior is changing rapidly.”
BackgroundMethod
Metadata
- Authors
- Lepage, T., Bryant, D., Philippe, H., & Lartillot, N.
- Affiliations
- McGill University (CA)
- University of Auckland (NZ)
- Université de Montréal (CA)
- Centre National de la Recherche Scientifique (FR)
- Venue
- Molecular Biology and Evolution
- Publisher
- Oxford University Press (OUP)
- Published
- 2007-06-29
- Volume
- 24
- Issue
- 12
- Pages
- 2669-2680
- ISSN
- 0737-4038 (ISSN-L)
- 1537-1719
- OpenAlex
- W2116474217
- Cited by
- 546
- References
- 46
- Citations by year
- 2026: 27 · 2025: 28 · 2024: 24 · 2023: 33 · 2022: 39 · 2021: 34 · 2020: 38 · 2019: 29 · 2018: 43 · 2017: 33 · 2016: 32 · 2015: 38 · 2014: 50 · 2013: 18 · 2012: 19
- FWCI
- 4.98
- Type
- journal article
- Language
- en
- Open access
- bronze
- doi.org
- Locations
- Molecular Biology and Evolution (publishedVersion)
- PubMed (publishedVersion)
- HAL (submittedVersion, cc-by)
- Funders
- Génome Québec
- Centre National de la Recherche Scientifique
- Topics
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Genomics and Phylogenetic Studies
- Physical Sciences › Earth and Planetary Sciences › Paleontology › Evolution and Paleontology Studies
- Life Sciences › Biochemistry, Genetics and Molecular Biology › Genetics › Evolution and Genetic Dynamics
- Keywords
- relaxed molecular clock, divergence time estimation, CIR process, Dirichlet prior, model comparison, Bayes factor, thermodynamic integration, protein sequence data
- MeSH terms
- Bayes Theorem, Models, Molecular, Phylogeny, Time Factors, Genetic Variation
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
A language model drew the claims from the paper on 2026-10-05, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-05 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
