Do Clock Dates Agree with the First Fossils?

"Thus, one ends up in the curious situation where the fossil record is taken seriously in providing a prior for the ages of nodes, but in retrospect is discovered to be wildly unreliable." — Graham Budd and Richard Mann, Systematic Biology, 2024¹

"Both scenarios can be distinguished using molecular-clock methods because old groups would have accumulated a substantial number of molecular substitutions, leading to long ancestral molecular branches." — Sandra Álvarez-Carretero and colleagues, Molecular Biology and Evolution, 2026²

Summary

A molecular clock dates the split between living groups from their DNA differences.  The fossil record dates the same groups by where their oldest members sit in the rocks.  This article asks whether the two agree for the great radiations:  animals, placental mammals, modern birds and flowering plants, with a look at other groups.  It is a test of two mainstream methods against each other.  All ages below are the mainstream's figures on the radiometric timescale, printed to show how far apart the two methods are on their own terms.  The article does not adopt them.  How clocks are built and calibrated is covered in the parent article, Molecular Clocks.

1. The clock dates are older than the first fossils in every great radiation. The gap is about 75 to 260 million years for crown animals,³¹ 13 to 18 for placental mammals in the newest study,⁴ about 27 to 66 for the root of modern birds, and about 14 to 126 for flowering plants, depending on the study. Budd and Mann list beetles, sand dollars and bryozoans with gaps of the same size.¹
2. The gaps have narrowed over time, but method choices move them as much. The placental gap fell from about 43 million years in 2003 to about 16 in 2022.⁵⁴ Yet inside a single 2026 paper, changing only the calibration strategy moved the flowering-plant gap from about 102 to 17,⁶ and changing only the rate model moved the animal date by about 71.² The trend across decades is mostly a record of those choices.
3. Where the two agree, the agreement is often built in. Fossils set the minimum and maximum bounds a clock may use.⁷ When the maximum is set close to the fossils, the clock lands on the fossils.¹ For flowering plants the crown age is "almost entirely conditioned by its own prior distribution".⁸
4. The order agrees even where the timing does not. Animal fossils appear in the rocks in the order the molecular tree branches,¹ and flowering-plant pollen types appear in the order the molecular tree predicts.⁹ Order matched between rock sections is evidence either side may use, and on order the two methods agree.
5. Each of the four mainstream explanations has some independent support, and none closes the gap alone. The fossil record is incomplete, but fossil-only sampling models expect gaps of only a few to about a dozen million years for placentals and flowering plants. How well a group fossilizes does not predict how large its gap is: flowering plants leave abundant, durable pollen and still show one of the largest gaps. Bursts of faster evolution during radiations are measured in arthropods¹⁰ and proposed for birds,¹¹ but not shown to account for the gaps. Calibration and prior choices are shown to move dates by tens of millions of years. Rate models are shown by simulation to inflate deep dates.¹²
6. The experts dispute the result in print. Budd and Mann conclude that for animals and placentals the clock gives no "evidence for an origin of the clade deeper than what is suggested by the fossil record."¹ The clock's authors reply that the young-origin models "have a posterior probability of ∼0."² Their test assumes rate models that do not allow a short burst, which is the alternative Budd and Mann propose.
7. The young-earth critique describes the disagreement correctly and draws more from it than it shows. Tomkins and Bergman write that "divergence dates commonly disagree with paleontology despite being calibrated by it."¹³ That is true. But both columns sit on the same timescale, so the gap between them measures how imprecise deep clock dates are, not how old the earth is. Young-earth geneticists hold that differences between created kinds were made, not accumulated, so on their own view the deep clock dates are not dates at all.¹⁴¹⁵ The agreement on order is the part of this evidence their model still has to explain.

Verdict.  At the deepest nodes, clock dates and first fossils do not agree, and the disagreement has lasted thirty years.  They agree better at shallower nodes and where the fossil record is written into the clock's priors.  The gap is real, its size depends heavily on the analyst's choices, and no one explanation has been shown to account for it.  It says that deep clock dates are imprecise and lean old.  It says nothing about the age of the earth, and the fossils' agreement with the molecular branching order is the stronger signal in this evidence.

Part 1: What is being compared

Two dates for one event.  The event is the split that produced the last common ancestor of all living members of a group, called the crown group.  A clock estimates when that ancestor lived from the DNA differences among its living descendants.  The fossil record gives the age of the oldest fossil that can be placed inside the crown group.  Because a group must exist before its first fossil can form, the fossil gives a minimum.  Some gap between the two is expected on any account.  The question is how large it is, and whether its size can be explained.

The clock is not independent of the fossils.  A clock needs a rate, and for deep nodes the rate comes from fossils.  Michael Benton and Philip Donoghue, whose rules for fossil calibration are widely used, put it this way:  "Fossils can provide good ‘‘minimum’’ age estimates for branches in the tree, but ‘‘maximum’’ constraints on those ages are poorer."⁷  A calibration, by the consensus rules of James Parham and colleagues, "uses the oldest assigned fossil of a taxon as the basis for its minimum age".¹⁶  So a clock analysis takes in the first fossils of many groups as bounds, and then estimates the ages of the nodes between them.  When the node in question carries its own fossil minimum, a clock date younger than that fossil is ruled out in advance.  The test is only how much older the date comes out.

The prior is not what the analyst typed.  Programs combine the bounds on all nodes into a joint "effective prior," which can differ greatly from each calibration taken alone.  Budd and Mann define it as "the prior probability distribution that is constructed from the calibration densities by enforcement of known topological constraints."¹  Part 7 shows how much this matters.  The parent article covers calibration and circularity in general;  this article asks only whether the dates and the first fossils agree.

A gap is also predicted by sampling alone.  Statistical models of how often fossils are found, using no DNA, put each group's origin somewhat before its oldest fossil.  Such models exist for placental mammals and flowering plants, and they give the size of gap the fossil record's own sampling leads one to expect.  How complete the record is, and how that is measured, belongs to the sibling article How Complete Is the Fossil Record?;  this article uses only the figures those models print.

Part 2: Animals and the Cambrian explosion

The clock says animals are far older than their fossils, and has said so for three decades.  The figures below are for the last common ancestor of living animals and of bilaterians, the animals with a left and right side.  Each is set against the oldest fossils that the critics of the clock accept.

Ayala and colleagues called their 1998 result "consistent with paleontological estimates,"¹⁷ and Michael Lee's 1999 recalibration of an older dataset gave animal splits "between 791 and 528 million years ago, a result broadly consistent with the palaeontological documentation of the 'Cambrian explosion.'"²¹  Since then the estimates have not moved steadily toward the fossils.  The newest figures, from the group that wrote MCMCTree, one of the two main dating programs, put crown animals in the Cryogenian, some 80 to 260 million years before the oldest accepted animal fossils.  Which rate model is chosen moves the date by about 70 million years on the same data.

The traces set a harder limit than the bodies.  Soft-bodied animals rarely fossilize, but animals that crawl and burrow leave traces, and the trace record "is far more continuous than the body-fossil record, particularly with respect to that of soft-bodied organisms," as Gabriela Mángano and Luis Buatois put it.²²  Their worldwide compilation found few, simple Ediacaran traces and a sharp rise at the base of the Cambrian.  Their verdict:  "Trace-fossil information strongly supports the Cambrian explosion, but allows for a short time of phylogenetic fuse during the terminal Ediacaran–Fortunian."²²  The numbers are in the Cambrian explosion article.  A long fuse of 80 to 260 million years therefore needs animals that left neither bodies nor traces.  Kevin Peterson and colleagues proposed exactly that in 2004:  "This cryptic precambrian history suggests that these stem-group bilaterians must have been 'micrometazoans,'"¹⁸ animals too small to leave burrows.  The proposal explains the absence.  It was made to explain the absence, and this article found no fossil that confirms it.

The critics' reanalysis, and the reply.  Budd and Mann reran the 2015 dos Reis analysis "changing a single number, the soft maximum age of the Metazoa," to 580 million years.  "The analysis accommodates this shift simply by bunching up the earliest nodes, including the bilaterians.  Once again, there is a close prior–posterior correspondence."¹  They concluded that "A fossil-based date of c. 555–540 Ma for the origin of the crown-group bilaterians would thus not be unreasonable."¹  Álvarez-Carretero and colleagues answered with a formal comparison of the two hypotheses on the same data.  Under the better-fitting rate model the late Ediacaran hypothesis has a posterior probability of 1.70 × 10⁻²⁰ against the unconstrained one,² and they conclude that "the data provide strong evidence against the hypothesis that molecular data are uninformative about the age of crown Metazoa."²

That reply shows that the sequences prefer longer branches than the young hypothesis allows, under the rate models tested.  It does not show that the longer branches mean more time rather than faster change.  The authors themselves write that the comparison is "conditioned on the rate model used."²  Michael Lee and colleagues measured arthropod evolution in the Cambrian as "~4 times faster" in form and "~5.5 times faster" in molecules than in all later periods, and found that these fast rates "do not change substantially even if the radiation of arthropods is compressed entirely into the Cambrian (~542 mega-annum [Ma]) or telescoped into the Cryogenian (~650 Ma)."¹⁰  So a burst of rate during the radiation is measured, and its size does not decide between a short fuse and a long one.  The animal case is open inside the mainstream.  For bilaterians the two sides differ by about 40 to 150 million years on the mainstream timescale.  Both sides read the same radiometric timescale, so neither brings the rocks any closer to a young-earth reading.

Part 3: Placental mammals and the end-Cretaceous extinction

Three models, named before the genomes arrived.  Mark Springer and colleagues credit Kevin Archibald and Douglas Deutschman with "three competing models for the diversification of placental mammals, which are the Explosive, Long Fuse, and Short Fuse Models."²³  On the explosive model, placental mammals and their orders arose after the extinction that ended the Cretaceous, about 66 million years ago on the mainstream scale.  On the long-fuse model the orders split in the Cretaceous and diversified within themselves afterward.  On the short-fuse model even the diversification within orders began in the Cretaceous.  No crown placental fossil is accepted before the boundary.  Maureen O'Leary and colleagues, scoring 4,541 characters on fossils and living species, named Protungulatum donnae, at 64.85 million years, "the oldest undisputed species within crown Placentalia."²⁴  Budd and Mann add that "there are thousands of stem-group fossils known from before the boundary and no uncontroversial crown-group forms."¹  The clocks have favoured the long fuse.  Robert Meredith and colleagues wrote in 2011 that "Relaxed clock analyses support the long-fuse model of diversification".²⁵  Springer and colleagues, themselves clock workers, summed up in 2019 that "The Long Fuse Model has received more support from relaxed clock studies than any of the other models, but this model is not supported by morphological cladistic studies that position Cretaceous eutherians outside of crown Placentalia."²³

The placental gap has shrunk more than any other.

The fall from about 42 million years to about 15 came as data grew from a few genes to genomes and calibrations were revised.  O'Leary's team put the crown "just younger than the K-Pg boundary, or ~36 My younger than molecular clock–based mean estimates," while warning that "Ghost lineage estimates are minimum divergence dates and may underestimate the timing of actual splits."²⁴  The fossil-only model of Emily Carlisle and colleagues splits the difference:  it puts the root in the latest Cretaceous, but finds that "the majority of orders likely originated around or after the K-Pg boundary."²⁷  Álvarez-Carretero and colleagues agree on the orders:  "the latest Bayesian analyses provide diversification timings for ordinal-level crown groups that are close to their fossil records, but not for the common ancestor of the group."²  So for placentals the remaining disagreement is about one node, and it is now roughly 10 to 15 million years on the mainstream scale.

What the remaining gap rests on.  Budd and Mann reran the 2022 analysis with the placental calibration set from the oldest fossil to the boundary and got "mean 66.47 Ma, 95% hdi 65.60–68.85 Ma" for uncorrelated rates.¹  They argue that a post-extinction origin is the better reading, because survival odds rise after a mass extinction and the clock's own tree shows lineages crossing the boundary "with no apparent perturbation," although mammals suffered "up to 95% of species extinction."¹  The reply found the post-extinction model's posterior probability "∼0," with the posterior "very narrow and concentrated at the maximum bound at 66 Ma (Fig. 5), which indicates conflict between the molecular data and the K-Pg constraint used."²  The authors' strongest physical argument is that "forcing divergence timings to equate the age of oldest fossils can result in estimated molecular rates that appear physiologically implausible."²  Matthew Phillips found the reverse risk in the clock:  rate models for placentals show "substantial misspecification associated with life history rate correlates," and "divergences deep in the tree will tend to be inflated by the interaction of rate errors and asymmetric confidence in minimum and maximum bounds."  His own tighter analysis gave dates "younger than others published recently," consistent with a radiation "near the 66 Ma Cretaceous-Paleogene extinction event."¹²  The rate question is the hinge of the placental case, and it has not been settled.

Part 4: Modern birds

Two nodes tell different stories.  The root of modern birds, the split between the ostrich line and all other birds, still dates well before the oldest crown fossil.  The great radiation of Neoaves, the branch that holds most living birds, now dates close to the end-Cretaceous boundary in some studies and well before it in others.

The oldest accepted crown bird is Asteriornis maastrichtensis, a skull from Belgium dated "66.8-66.7 million years ago," which "occupies a position close to the last common ancestor of Galloanserae," the ducks and fowl.²⁸  Below the boundary there are many birds outside the crown.  Jacob Berv and Daniel Field note that "The Mesozoic fossil record of crownward stem birds argues against the up to 100 million-year missing fossil record advocated by unconstrained divergence time analyses without a maximum age prior on the root."¹¹

The duck-and-fowl row is the one place in this article where a clock date comes out younger than a fossil placed near the node.  Kuhl's team say the 62.5 "fits well" with Asteriornis,³¹ which holds only if Asteriornis sits just outside the crown of ducks and fowl.  Field's team placed it "close to" that ancestor.  The point matters for the claim that clocks always run older:  when a node carries no fossil minimum of its own, the clock can land on either side of a fossil.

Two genome studies of the same year disagree on whether most bird orders are Cretaceous.  Josefin Stiller and colleagues, with 363 genomes, found that "Only two divergences occurred before the boundary," and wrote "In contrast to a recent study proposing a diversification of Neoaves during the Upper Cretaceous, we found that the early divergences in Neoaves were tightly associated with the K–Pg boundary."³²  The study they answer, by Shaoyuan Wu and colleagues, found that "interordinal diversification of modern birds largely occurred in the Late Cretaceous," with about 15 orders after the boundary and "the remaining 24 orders" before it.³³  Both used fossil calibrations and genome-scale data.  Wu's team bounded the root between Asteriornis at 66.7 and Ichthyornis dispar at 94.3,³³ so their root could not leave that window.

Part of the agreement was put in by hand.  Of the bird study by Richard Prum and colleagues in 2015, which put the radiation after the boundary, Berv and Field write that it "employed an informative but soft-bound prior on the root of Neornithes, reflecting the conspicuous absence of avian crown group fossils in the Late Cretaceous Niobrara Formation," and that "Applying this prior compresses the age of the neornithine MRCA by 30 Ma."¹¹  When the absence of fossils is written into the prior, an agreement with that absence is not an independent result.

The rate-burst explanation for birds.  Berv and Field propose that birds that survived the extinction were small, that small birds have faster molecular rates, and that this "hidden" rate acceleration "may result in both strict and relaxed molecular clocks over-estimating the age of the avian crown group."¹¹  Their evidence is partly independent of the clock.  Body size reconstructed for the ancestor of modern birds came out at 553 grams, "50% smaller than our median body size estimate for the pre-K-Pg assemblage," a fossil bird fauna from the last stage of the Cretaceous.¹¹  But the direct test did not run:  "Despite lengthy attempts to reanalyze the avian evolutionary timescale in Coevol 1.4b to directly test for lineage-specific rate accelerations, these analyses failed to converge."¹¹  The mechanism is plausible and partly supported.  It has not been shown to close the bird gap.

Part 5: Flowering plants

The fossil side is unusually good.  Flowering plants shed pollen in large numbers, and fossil pollen is the record this case turns on.  The oldest accepted crown angiosperm fossils are pollen grains:  "The earliest unequivocal fossil evidence of crown angiosperms dates to the Early Cretaceous (Valanginian stage; ∼135 Ma)."³⁴  Mario Coiro, James Doyle and Jason Hilton describe what follows:  "Pollen floras document a Northern Gondwanan appearance of monosulcate angiosperms in the Valanginian and subsequent poleward spread of monosulcates and tricolpate eudicots, accelerating in the Albian."  They add that "supposed pre-Cretaceous angiosperms either represent other plant groups or lack features that might confidently assign them to the angiosperms."⁹

The clock side ranges over 130 million years.

Hervé Sauquet, Santiago Ramírez-Barahona and Susana Magallón reviewed the range in 2022:  "Recent studies have reported a perplexing range of estimates for the crown-group age of angiosperms, from ~140 million years (Ma; Early Cretaceous) to 270 Ma (Permian)."  Their diagnosis is that the clock estimates "have little to do with either the amount of molecular data or the number of internal fossil calibrations included.  Instead, we argue that this age is almost entirely conditioned by its own prior distribution," so that "contrasting age estimates in the literature mostly reflect conflicting prior distributions."⁸  The 2026 study by Ruolin Wu and colleagues shows the point inside one paper:  the same 644 species and 83 genes give a crown age between 212 and 261 million years under hard bounds and about 152 under bounds drawn from a model of the fossil record.⁶

The gap shrinks when the fossils are written into the clock.  Magallón's Early Cretaceous result came after "The angiosperm crown age was bound within a confidence interval calculated with a method that considers the fossil record of the group."³⁶  Wu's Late Jurassic result came from calibration densities built from 25,685 fossil occurrences, from which "We excluded all pollen records," because pollen is "difficult to attribute consistently to family-level clades."⁶  Each is a fair method.  Neither is an independent check of the fossils, because the fossils set the bounds.

Why pollen makes this case hard to explain away.  If flowering plants lived through the Jurassic or Triassic, they shed pollen there.  Coiro and colleagues find that the order of pollen types in the rocks "agrees with molecular phylogenetic inferences on the course of pollen evolution, but it conflicts strongly with Triassic and early Jurassic molecular ages, and the discrepancy is difficult to explain by geographic or taphonomic biases."⁹  They leave one door open:  "the record may allow the Late Jurassic existence of ecologically restricted angiosperms, like those seen in the basal ANITA grade."⁹  That reconciles the record with the youngest clock dates, those near 150 to 160 million years.  It does not reconcile it with the Triassic dates that conventional calibrations still produce.  The fossil-only estimates disagree among themselves.  Silvestro's model found a pre-Cretaceous origin;  Budd, Mann, Doyle, Coiro and Hilton disputed it in 2021, arguing that the result follows from the oldest fossil and the number of fossils;³⁸ Wu's pooled run, without pollen, gives the Late Jurassic to earliest Cretaceous.⁶

Part 6: Other groups, large gaps and small

The pattern is general, but not uniform.  Budd and Mann list recent gaps "between the first recognized fossil of a clade and the 95% credible interval of molecular clock estimates":  beetles "55–134 myr older than the oldest fossil," sand dollars and sea biscuits "40–65 myr," cheilostome bryozoans "132–238 myr," angiosperms "23–121 myr" and bilaterians "50–143 myr."¹  Three more groups show the range.

Land plants are the closest case, and the agreement is partly built in.  Jennifer Morris and colleagues put crown land plants near 500 million years, against spores of land-plant type in the Middle Ordovician.  But Cambrian spores "of unclear affinity," though "not considered unequivocally embryophyte, informed our soft maximum constraint (515.5 Ma)."³⁹  The top of the clock's range is a fossil bound.  The first body fossil, Cooksonia, comes much later, in the Silurian,³⁹ so the land-plant clock agrees with the spores and not with the bodies.

Primate estimates have spanned some 30 million years.  Mario dos Reis and colleagues write that before their study "the estimated age of the living primate clade has spanned a 30 My differential, ranging from as young as 55 Ma (Koop et al. 1989) to as old as 87 Ma (Perelman et al. 2011)."⁴¹  Springer's 2012 estimate is "markedly younger than most timetree dates (Table 4), some of which are as old at 85–90 Ma."⁴⁰  The oldest well-preserved early primate skeleton, Archicebus from China, is "about 55 million years ago".⁴⁵  The primate gap is small on the newest dates and large on the older ones, the same pattern as placentals.

Eukaryotes show how the reference fossil decides the gap.  Laura Parfrey and colleagues found their eukaryote ancestor "consistent with the earliest microfossils interpreted with confidence as eukaryotic."⁴²  Holly Betts and colleagues name the red alga Bangiomorpha as "The oldest fossil remains that can be ascribed to crown-Eukaryota".⁴³  Against that fossil the gap is hundreds of millions of years, and Timothy Gibson and colleagues moved that fossil "150 m.y. younger than commonly held interpretations" in 2018.⁴⁴  Whether a gap is large depends as much on which fossil is accepted as the first as on the clock.

Part 7: How the gap moved as methods changed

From strict clocks to relaxed clocks.  Dos Reis, Donoghue and Yang describe the first genome-era conflict:  "the clock estimates were much older than the dates suggested by the fossil record, sometimes twice as old."⁴⁶  Relaxed clocks let rates differ between branches.  They narrowed the placental and bird gaps over the next two decades, as the tables in Parts 3 and 4 show, but not the animal gap.

From node dating to tip dating.  In tip dating, fossils enter the analysis as dated species with their own anatomy, rather than as bounds on nodes.  It was expected to break up long branches and bring dates toward the fossils.  For placental mammals it did the opposite.  Robin Beck and Michael Lee found that morphological clocks "retrieve implausibly ancient, Late Jurassic–Early Cretaceous estimates for the initial diversification of Placentalia," and that "These dates are much older than all recent molecular and palaeontological estimates."⁴⁷  Mark Puttick, Gavin Thomas and Michael Benton found the same with a different matrix:  "the estimated date for the origin of crown mammals is much older, ∼130–145 million years ago (Ma), than fossil and molecular clock data (∼80–90 Ma)."⁴⁸  Their title says it plainly:  morphological clocks "fail to close the molecular fossil gap."⁴⁸

Better priors.  The newest move is to build calibration bounds from statistical models of the fossil record, as Wu and colleagues did for flowering plants.  That closed most of the gap, from over 100 million years to about 17, inside one paper.⁶  The cost, as Part 5 notes, is that the clock then agrees with the fossils partly because the fossils set its bounds.

The gaps, priced.  The script research/gap-pricing.ts takes every clock interval cited in this article for the five best-studied nodes and subtracts the first-fossil age each literature prints.⁴⁹  Three results follow.

1. Every group's gap has narrowed, at 9 to 28 million years per decade.
2. But one analyst's choice inside one paper moves the gap by 9 to 85 million years, which is as much as one to four decades of that trend. The trend cannot be read as convergence on a true date until those choices are settled.
3. Most gaps exceed the clock's own precision floor. Dos Reis and Yang showed that a deep date's 95 % interval cannot shrink below about ±16 % of the date, however much DNA is added.⁵⁰ Of the 22 clock estimates in the script, 18 have a gap larger than that floor. The gap is not just noise.

Part 8: What explains the gaps, and which explanations are tested

The field offers four explanations.  Each makes a prediction that can be checked against something other than the gap it was offered to explain.

1. Incompleteness is real, and smaller than the gaps.  Ghost lineages, the stretches that a lineage must have existed because its sister group is older, give a fossil-only measure of what is missing.  Álvarez-Carretero and colleagues count from O'Leary's fossil tree "61 ghost lineages among the basal branches that they sampled in placental mammal phylogeny, equating to a total of over ∼780 Myr of evolution with no record."²  That is about 13 million years per lineage on average, close to the remaining placental gap of 13 to 18.⁵¹  For placentals, then, incompleteness of the size needed is measured independently.  The fossil-only sampling models say the same:  Carlisle's model expects a gap of about 1 to 12 million years, and Wu's pooled model for flowering plants expects 0 to 15.  The clock gaps in the newest studies are 13 to 29 for placentals and 17 to 126 for flowering plants.⁴⁹  Sampling accounts for part of each gap, and in flowering plants only the gap under fossil-derived bounds.

The other prediction fails.  If the gaps came from poor preservation, groups that fossilize badly should have the largest gaps.  Ranking the five nodes by how well their first fossils preserve, with early animals lowest and pollen highest, gives a rank correlation with the relative gap of +0.39 across all 22 clock estimates and −0.15 across the latest estimates alone.⁴⁹  Neither is the strong negative value that preservation alone predicts.  Five groups and ranks of this article's choosing make that a weak test.  The pollen case makes it a strong one.  Coiro and colleagues judge the flowering-plant discrepancy "difficult to explain by geographic or taphonomic biases."⁹  For birds, Berv and Field note a rich record of stem birds just below the crown.¹¹  For placentals, Budd and Mann count "thousands of stem-group fossils" before the boundary.¹  In each case the rocks below the first crown fossil are full of the group's close relatives, and the crown members are the ones missing.  An explanation by incompleteness has to say why the record missed only them.

2. Rate bursts are measured but not shown to close any gap.  Lee and colleagues measured about 5.5 times faster molecular evolution in Cambrian arthropods, and found it robust whether the fuse was short or long.¹⁰  That makes a burst real and, for arthropods, uninformative about the fuse.  For birds, Berv and Field's body sizes are partly independent support,¹¹ and their direct test did not converge.¹¹  For placentals, Budd and Mann argue that rates "could be much higher than the long-term mean when short intervals are considered,"¹ and the clock's authors reply that the required rates "appear physiologically implausible."²  Neither paper prints the rate it means.  The models used to reject the young-origin hypotheses, autocorrelated and independent lognormal rates, do not allow a short burst at one point in time.²  So the formal test has not yet been run against the alternative it is cited to defeat.

3. Priors are shown to move the dates by tens of millions of years.  Rachel Warnock, Ziheng Yang and Philip Donoghue found that the bounds a user types "are not the effective priors implemented in calculating the posterior time estimates," and that changing a lognormal calibration's settings can move the estimates and their intervals "by hundreds of millions of years."⁵²  Budd and Mann moved the animal date to the fossils by changing one bound.¹  Sauquet's team found that the flowering-plant crown age is set mainly by its prior,⁸ and Wu's team moved it by about 85 million years with calibration choices alone.⁶  This explanation is tested, not fitted.  It has a direction, too.  Phillips argues that "asymmetric confidence in minimum and maximum bounds" inflates deep dates,¹² and Francisco Rodríguez-Trelles, Rosa Tarrío and Francisco Ayala that the statistics of rate estimation "introduces a bias toward an overestimation of time since divergence."⁵³

4. Rate models are shown to move the dates.  On the same data and the same calibrations, the two rate models in the 2026 reply put crown animals 71 million years apart and crown placentals 9 apart.²  Phillips used fossil calibrations to test placental rate models and found "substantial misspecification associated with life history rate correlates."¹²  Beck and Lee warn that with relaxed clocks "relatively minor deviations in rate patterns can generate large effects on estimated divergence dates."⁴⁷

Fitted after the fact.  Two explanations were proposed to cover a particular absence and have no independent confirmation yet.  One is Peterson's microscopic early bilaterians, which would leave no traces.¹⁸  The other is ecologically restricted early flowering plants, which Coiro and colleagues allow for the Late Jurassic only.⁹  Both are reasonable.  Neither is evidence until its fossils are found.

Part 9: The young-earth reading

The positions, in their own words

The method critique.  Jeffrey Tomkins and Jerry Bergman list among the clock's standing problems that "divergence dates commonly disagree with paleontology despite being calibrated by it,"¹³ and judge that "the prospects of achieving any type of evolutionary concordance has been dismal."¹³  Brian Thomas of the Institute for Creation Research, writing on the Cambrian in 2008, drew the young-earth conclusion from the spread of animal dates:  "Either the molecular clocks are wrong, the rock dates are inaccurate, or both."  He added that "creation scientists would suggest that assumptions of evolution may influence both of them too strongly to allow any confidence in either."⁵⁴  Michael Oard notes that "The molecular clock often suggests that fossils should be found earlier," and that fossil ranges keep being pushed back.⁵⁵

The burial-order model.  Leonard Brand and Arthur Chadwick hold that first appearances record when habitats were destroyed in the Flood, not when groups arose:  "These changes in the fossils do not represent the appearance of newly created organisms.  They resulted from the successive final destruction of some habitats and the death and burial of organisms from each of those habitats as the catastrophe unfolds."⁵⁶  Of birds, mammals and flowering plants:  "Perhaps these groups were largely inhabitants of upland regions that were not affected until late in the geological catastrophe."⁵⁶  Kurt Wise proposes separate pre-Flood biomes:  "It is also likely that the gymnosperm-dinosaur biome was located at a lower altitude or closer to the shore of the antediluvian world.  This would explain why members of the gymnosperm-dinosaur biome are consistently buried beneath members of the other biome."⁵⁷

Created differences.  For differences between created kinds, young-earth geneticists use no clock.  Robert Carter writes that dating the great-ape lineages amounts to "uncovering the differences God built into their initial genomes."¹⁵  Nathaniel Jeanson writes of human-chimpanzee differences that "A creationist would respond by saying that these differences were created."¹⁴

An intelligent-design reading, kept separate.  Stephen Meyer, who is not a young-earth creationist, argues that "Proponents of deep divergence don't deny that the fossil evidence has come up short.  Instead, they adopt one of the versions of the artifact hypothesis to account for that missing evidence."⁵⁸

The positions, tested

The description is right.  The dates do disagree with the first fossils, by the amounts in Parts 2 to 6, and mainstream authors say so in print.  Thomas took his range of "586 to over 1200 million years" for animal origins from a 2008 review by the biologist Jeffrey Levinton.⁵⁴  Oard's remark that clocks predict earlier fossils is also correct, and it points the right way:  each fossil found below an old first appearance shrinks a gap.

"Despite being calibrated by it" needs care.  A clock is calibrated by first fossils at many nodes, as minima.  The test is whether the nodes between them come out near their own first fossils.  Where the node's own fossil sets a firm bound, agreement is built in.  Where it does not, disagreement is a finding, not a contradiction in the method.  Part 7's numbers support the critics on a narrower point:  the dates move with the analyst's choices as much as with the data.

Thomas's "or both" is the consistent young-earth position, and it ends the test.  Both columns of every table in this article rest on the same radiometric timescale.  A reader who rejects that timescale rejects both columns, and the gap between them then measures nothing about the age of the earth.  Inside the mainstream, the gap shows that deep clock dates are imprecise and lean old.  It does not show that the earth is young, and the young-earth authors quoted here do not claim it does.

The burial-order model gets the sign right and nothing more yet.  If every group lived from creation and was buried by habitat, every group is older than its first fossil, so a clock date older than the first fossil is expected.  But the model as published gives no size for any gap, and no reason why the placental gap should shrink with better data while the animal gap does not.  Its authors name its hardest case themselves:  "why is there no flowering plant pollen below the Middle Cretaceous?  Pollen is easily carried long distances by wind or water."  Brand and Chadwick add, "Our current understanding of this model leaves many unanswered questions."⁵⁶  The same pollen record is the strongest evidence against the Triassic clock dates for flowering plants,⁹ so it presses both readings at once.

The order is the harder test, and it is open to both sides.  Order matched between rock sections does not depend on radiometric dates, and young-earth geologists accept it:  Wise and Donna Richardson conclude that "the global biostratigraphic column seems to be real."⁵⁹  On that order, the clock's branching and the rocks agree.  Budd and Mann find "a compelling congruence between the order of appearance of fossils in the record and phylogeny,"¹ and Coiro's team find that the sequence of pollen types "agrees with molecular phylogenetic inferences on the course of pollen evolution."⁹  A burial-order model sorts by habitat and elevation.  Nothing in it, as published, says why habitat order should match the branching order of DNA trees.  Whether ecological zonation reproduces the order of the fossil record is tested in The Order of the Fossil Record, which finds that it does not.  Whether trees and rock order agree beyond chance is in Phylogenetic Inference and Congruence.

Created differences leave the clock dates without meaning, and leave the order to explain.  On Carter's and Jeanson's view, sequence distances between kinds are not times, so the deep clock dates in this article are neither right nor wrong.  That is a coherent position, and nothing here refutes it.  It does leave the order unexplained:  the branching order read from those differences is the order in which the groups first appear in the rocks.  Designed differences need not nest in an order that burial follows.  A created-kinds model that predicted the match would answer this article's strongest finding.  None is published.

Objections and responses

1. "The clock is calibrated on fossils, so comparing it with fossils is circular."  Partly true.  Where a node carries a firm bound from its own first fossil, agreement is built in, as Parts 1 and 7 show.  The comparison is informative only at nodes left loose, and those are the nodes where the gaps appear.  The objection explains why some agreements mean little.  It does not remove the disagreements.

2. "The fossil record is incomplete, so of course groups are older than their first fossils."  A gap is expected, and fossil-only models give its expected size:  a few million years for placentals and flowering plants.  The clock gaps are larger, and they are not largest where fossilization is worst.  Flowering plants leave abundant pollen and still have one of the biggest gaps, and the beds below the first crown birds and placentals are full of their stem relatives.

3. "The gaps are closing.  Give the method time."  They have narrowed by 9 to 28 million years per decade.  But a single paper's choice of rate model or calibration moves them by 9 to 85 million years, and some of the closing came from building the fossil record into the priors.  Closing by construction is not convergence on a true date.

4. "The 2026 model comparison settles it:  the DNA rejects young origins."  It shows that the sequences prefer longer branches than a young origin allows under the two rate models tested.  Neither model allows a short burst at the radiation, which is the alternative Budd and Mann propose and which Lee and colleagues measured in arthropods.  Until a burst model is compared, the question of time against rate is open.

5. "Clock dates miss the fossils by hundreds of millions of years, so the method is worthless and evolution is false."  The misses show that deep clock dates are imprecise and lean old, which mainstream authors concede.  They do not bear on common descent, because the same trees put the groups in the order the rocks show.  They do not bear on the age of the earth, because both dates sit on one timescale.

6. "Rate acceleration is an ad hoc rescue."  Not wholly.  The Cambrian arthropod burst was measured from anatomy and DNA without reference to the gap, and the bird body-size shift is partly independent.  The ad hoc explanations are the ones with no independent trace yet:  microscopic early bilaterians and hidden Jurassic flowering plants.

7. "If every kind existed from creation, young-earth creation predicts that every group is older than its first fossil."  It predicts the sign.  It does not predict the sizes, why they differ between groups, or why the placental gap shrank with data while the animal gap did not.  Brand and Chadwick concede that it does not yet explain the missing pollen.  And it does not yet say why burial order matches the molecular branching order.

8. "The first fossils are disputed, so the gaps mean nothing."  For placentals and birds the alternatives move the first-fossil age by a million years or two, and the gaps stand.  For eukaryotes the choice of first fossil matters greatly, as Part 6 shows, and that case is weighted accordingly here.

9. "Fossil-only models also assume deep time."  They do.  They are used here only to show what the mainstream's own sampling models expect, so that the clock and the fossils can be compared on the mainstream's terms.  They do not bear on the age of the earth either.

Issues and unanswered questions

1. No fossil-only origin model was found for animals or birds. Such models exist for placentals, primates and flowering plants. An animal model would need the trace-fossil record as well as bodies.
2. No published test compares node-age hypotheses under a rate-burst model on the animal or placental datasets. That test would separate time from rate, and both sides in the 2024 to 2026 exchange point toward it.
3. The preservation test is weak. It uses five nodes and ranks set by this article. A proper test would use a measured preservation rate per group, such as the sampling rates estimated by fossil-only models.
4. Two figures in the sources disagree with themselves. Álvarez-Carretero and colleagues give "at least 720 Myr" of placental ghost lineages in the abstract and "over ∼780 Myr" in the text.² Carlisle and colleagues report 12 of 23 clades overlapping and 10 not, which sums to 22.²⁷
5. Some primary sources were not read in full: Wray and colleagues 1996 on billion-year animal dates; Prum and colleagues 2015, whose node ages are known here only through Berv and Field; Ksepka and Phillips 2015 on bird dating; Lee 1999 beyond its abstract; Heads 2009 on primate biogeography.
6. A Jurassic flower is claimed. Brand and Chadwick report that "A perfect angiosperm flower has been found in the Jurassic."⁵⁶ Coiro and colleagues judge every supposed pre-Cretaceous angiosperm to be another plant or unassignable.⁹ That specimen is not assessed here.
7. No young-earth model predicts the size of any gap or the match between branching order and rock order. The burial-order models of Brand, Chadwick and Wise are qualitative. A model that ranked groups by pre-Flood elevation and predicted their first appearances, tested against the molecular branching order, would be the natural next step for that side.

Verdict

At the deepest nodes, clock dates and first fossils do not agree.  For crown animals the clock runs 80 to 260 million years older than the oldest accepted animal fossils, on the mainstream timescale.  For flowering plants it runs 14 to 126 older than the first pollen.  For the root of modern birds it runs 27 to 66 older than Asteriornis.  For placental mammals the gap has fallen from about 42 to about 15, and the placental orders and the main bird radiation now date close to their fossils.  Land plants and primates show small to moderate gaps.

The disagreement is real, but its size is not a stable measurement.  Within one paper, the analyst's choice of rate model or calibration strategy moves it by 9 to 85 million years.  Where the clock and the fossils agree, the fossils often set the bounds.  Of the mainstream explanations, calibration and prior effects and rate-model error are demonstrated by sensitivity tests.  Incompleteness is measured and accounts for part of each gap, but not where fossils are most abundant.  Rate bursts are measured in one group and proposed in another, and have not been shown to close any gap.  The experts most involved disagree in print, and the decisive test, time against rate, has not been run.

For the young-earth question the gap is neutral, and the order is not.  Both dates rest on the same timescale, so their disagreement says nothing about the age of the earth, as Thomas's "or both" already concedes.  What this evidence does show is that the branching order read from DNA matches the order of first appearances in the rocks.  That match is open to both sides, does not depend on radiometric dating, and is the part of the evidence a young-earth model has yet to explain.

Sources