Mass Extinctions

"Creation scientists do not consider these as true “extinction” events.  Instead, these horizons are interpreted as major shifts in the burial pattern of fossils during the Flood." — Jeffrey Tomkins and Tim Clarey, Institute for Creation Research, 2023¹

"A creation/Flood framework would have much more difficulty dealing with a global pattern of uniform sharp 'cut-off' at the K-T boundary, rather than a blurry picture, which appears to be the reality." — Carl Wieland, Creation Ministries International, 1998²

Summary

A mass extinction is a level in the rock where many kinds of fossils stop at once.  Five such levels stand out in the marine record:  the end of the Ordovician, the Late Devonian, the end of the Permian, the end of the Triassic and the end of the Cretaceous.  These are the "Big Five."  Smaller ones sit between them.

This article asks four questions of each.  What died?  How abruptly, measured in the rock of a single section?  What came back above it, and in what order?  And where do young-earth creationist models place it:  inside Noah's Flood, at its end, or after it?

The article works from the order of the layers alone.  Ages in years appear only when quoting what mainstream geologists claim;  nothing here depends on them.

Eight conclusions, stated first.

1. The extinctions are real patterns in the rock, not artefacts of how the timescale is divided. Each shows up in single sections, where the order of the beds is not in dispute. At Meishan in China, the classic end-Permian section, most species stop in one bed. In the Hell Creek badlands of Montana, the youngest dinosaur bone found in place sits 13 cm below the layer where Cretaceous pollen disappears and fern spores take over.³ A database check run for this article agrees with the published curves: of 602 marine genera that enter the last Permian stage and can be placed in single stages, 387 have no record above it, by far the largest such loss in the record.⁴
2. The order above each boundary is the order of a recovery, and it repeats. Above the end-Permian come a few holdovers that soon vanish, a handful of hardy survivors, dwarfed shells and mats of microbes. Swimmers such as ammonoids and conodonts rebound within the Lower Triassic; animals fixed to the sea floor stay scarce until the base of the Middle Triassic, and reefs and coal are missing throughout. About two-thirds of the genera found both below and above the Lower Triassic are missing from the Lower Triassic itself, and so is every one of the 53 demosponge genera among them, most of them reef builders.⁵ Above the end-Cretaceous come a fern spike on land and, in the sea, one surviving plankton species followed by the first new ones, then burrowing animals.
3. Every active young-earth model puts four of the five inside the Flood year. The Institute for Creation Research (ICR) assigns the Cambrian to Mississippian rocks to the first 40 days and the end-Permian and end-Triassic to Days 40 to 90.⁶ No active camp puts the end of the Flood below the Cretaceous; the 1996 group that did withdrew its model in 2000.⁷
4. The end-Cretaceous splits the young-earth camps. One camp, founded by Steven Austin and colleagues in 1994 and carried by John Whitmore, Paul Garner and Kurt Wise, ends the Flood at or just above it. ICR's Tim Clarey and Davis Werner put it at the Flood's high-water mark, Day 150, with months of Flood still to come; Andrew Snelling and Michael Oard also place the end of the Flood well above it.
5. The camps also split on whether the extinctions happened at all. ICR says there are "no real extinction events in the rock record," only the last level at which each group was buried.⁸ Most other young-earth writers, including the founders of the end-Cretaceous camp, treat the losses as real deaths caused by the Flood.⁹
6. The "only burial order" reading fails a test it sets itself. If trilobites, rugose and tabulate corals, fusulinid forams and productid brachiopods were still alive in the sea after the end-Permian level, the Flood went on burying sea creatures in great numbers: 404,784 later Mesozoic marine occurrences against 124,185 from the whole Permian. At even a hundredth of their last-Permian share, those five groups should have left about 9, 53, 6, 91 and 584 fossils in that later rock. It holds 0, 3, 0, 1 and 8 outside the first Triassic substage.¹⁰ A companion prediction, that extinctions fall at the Flood's megasequence boundaries, fails for three of the five on ICR's own figure.¹
7. The "real deaths" reading survives that test but leaves the recoveries unexplained. On every model the whole Early Triassic recovery, with its dwarfs, its missing reefs, its missing coal and its stepwise rebuilding of food webs, was laid down in days. On the end-Cretaceous boundary, the whole Cenozoic recovery has to fit into the few centuries before the Ice Age, which the founders of that camp put at about five centuries after the Flood.¹¹ No young-earth source found for this article addresses the dwarfing, the reef gap or the stepwise rebuilding at all.
8. The critics of the mainstream picture have real points. A sampling effect named after Signor and Lipps makes sudden extinctions look gradual, and it can also blur the reverse. Two of the Big Five are as much failures of new forms to appear as surges of loss. The timing of the land turnover in South Africa is in open dispute, and land plants may not have suffered a mass extinction at the end of the Permian at all.

Verdict.  Against the reading of mass extinctions as artefacts of burial order:  that reading predicts fossils that are not there.  For the young-earth view that the Flood itself was the killer, the position is unfinished:  the deaths fit a catastrophe, but the ordered recoveries above them have no published young-earth account.

Part 1: What counts as a mass extinction, and which ones stand out

How the five were found

The "Big Five" come from a 1982 paper by David Raup and Jack Sepkoski of the University of Chicago.  They compiled the first and last appearances of thousands of marine animal families and asked which intervals lost far more families than the long-run trend predicts.  Their answer was four, not five:  "four mass extinctions in the marine realm are statistically distinct from background extinction levels.  These four occurred late in the Ordovician, Permian, Triassic, and Cretaceous periods.  A fifth extinction event in the Devonian stands out from the background but is not statistically significant in these data."¹²  The Late Devonian joined the list by convention.

Later work with genera instead of families sharpened the picture and also blurred it.  Richard Bambach, Andrew Knoll and Steve Wang ranked the extinction intensity of every Phanerozoic substage and found "a continuous distribution, with the five traditionally recognized mass extinctions located in the upper tail."  Compared with their neighbours in time, "only the end-Ordovician, end-Permian, and end-Cretaceous extinction intensities appear as outliers."¹³  The other two lost diversity mainly because few new genera appeared:  "about two-thirds of the diversity loss in the late Frasnian and almost 60% in the end-Triassic can be ascribed to origination failure, not elevated extinction."¹³  Bambach's 2006 review found 18 intervals that fit Sepkoski's definition of a mass extinction and concluded that "mass extinctions are not a homogeneous group of events."¹⁴  John Alroy, working with occurrence data rather than published ranges, found "no sharp distinction between background and mass extinctions" across the whole record, while confirming that the end-Permian and "several" others "were particularly severe."¹⁵

So the young-earth complaint that mass extinctions are "arbitrarily defined," made by Kurt Wise and Matthew Richardson, is half right.¹⁶  The threshold for calling an interval a mass extinction is a convention.  The end-Permian is not, and young-earth writers have said so:  Oard called it "the most singular extinction event in the supposed history of life."¹⁷  It is the largest loss on every measure, and Wise and Richardson's own data show it most clearly of all:  of 2,556 species they found in the last Permian stage, only 4 occur on both sides of the boundary within 2 degrees of each other, 0.2 %, where ordinary stage boundaries run at about 3 to 17 %.¹⁶

How big they were

The size of each loss depends on whether species, genera or families are counted, and on how the counts are corrected.  The most careful recent estimates are smaller than the figures usually quoted.  Steven Stanley of the University of Hawaii, correcting for the way rare fossils make extinctions look earlier than they were, put the end-Permian at "only ∼81% of marine species," not the "90–96%" often quoted, and wrote that "Life did not nearly disappear at the end of the Permian, as has often been claimed."¹⁸  For genera he estimated 42 to 43 % for the end-Ordovician, 16 to 20 % for the Frasnian event of the Late Devonian, and 38 to 40 % for the end-Cretaceous.¹⁸  He left out the end-Triassic because "its timing and list of marine victims remain controversial."¹⁸

Creationist writers quote the larger figures.  Emil Silvestru repeats National Geographic's "some 90 percent of the planet’s species" for the end-Permian;¹⁹  Clarey's table gives 96 % of species.²⁰  Neither number is wrong as a citation, but both are older estimates that later work has revised downward.

A check against the whole database

The published curves rest on compilations that a reader cannot easily rerun.  So this article ran its own check on the Paleobiology Database, the largest open compilation of fossil occurrences.  For every animal genus with fossils of known age, it took PBDB's first and last appearance, placed each in one of the 101 named stages, and asked, stage by stage, what share of the genera that entered a stage from below have no record above it.⁴  Only the order of the stages matters here.  The stage ages serve as labels.

Three things stand out, and all three match the published work.  The end-Permian is far ahead of every other well-sampled stage.  The Late Devonian does not show as one great loss, and its Frasnian stage made few new genera, which is Bambach's "origination failure."  And above each event the share of new genera climbs back:  after the end-Permian it runs 10.5 % in the last Permian stage, about 26 % in each of the two Lower Triassic stages, and 45.9 % in the first Middle Triassic stage, the Anisian.  Cambrian stages score as high as the end-Permian on these measures, but on samples of a few dozen genera.  The numbers are raw.  They are not corrected for how much each stage has been collected, so they check the published curves rather than replace them.

The sampling effect that cuts both ways

A fossil's last appearance is almost never the moment its species died.  A rare animal is found less often, so its last fossil usually sits some way below the true extinction level.  When many species die at once, their last fossils therefore scatter below the kill horizon, and a sudden event looks gradual.  This is the Signor-Lipps effect, named for Philip Signor and Jere Lipps, who described it in 1982:  "the incorrect assignment of extinctions that occurred during a crisis to an interval preceding the crisis because of the incompleteness of the fossil record," in Stanley's definition.¹⁸

Charles Marshall showed it in a single section.  On Seymour Island, Antarctica, ammonites "disappeared gradually over a stratigraphic interval 10–50 m below the boundary" if the record is read literally, yet the pattern "is consistent with a sudden mass extinction" at the end-Cretaceous, and the iridium layer falls where his method predicts the extinction should be.  He was careful to add that gradual scenarios spread over as much as 20 m also fit, so that "without saturation collecting near the K-T boundary it will be impossible to distinguish between gradual and sudden extinction scenarios" from the ammonites alone.²¹

The effect also runs the other way.  Steven Holland and Mark Patzkowsky simulated how fossils accumulate in real sedimentary sequences and found that last appearances "tend to cluster at subaerial unconformities, surfaces of forced regression, flooding surfaces and intervals of stratigraphical condensation."  A long period of raised extinction can therefore look like a sudden pulse.  "With the possible exception of the end‐Cretaceous," they wrote, the mass extinctions show exactly such clusters, which "may represent brief pulses of elevated extinction" but "are equally likely to form by stratigraphical processes."²²  This is the mainstream version of a point Clarey makes from a young-earth survey:  that many sudden-looking extinctions sit at sudden changes of rock type.²⁰  Both are right that the shape of an extinction in one section can be an artefact.  Neither shows that the extinction itself is one.

Part 2: The end-Permian, bed by bed

The kill in one section

The reference section for the end of the Permian is a quarry at Meishan in Zhejiang, China.  Its beds are numbered from the bottom up.  The boundary is defined by the first appearance of a small conodont, Hindeodus parvus, at the base of Bed 27c.  The extinction lies a little lower.  Ian Metcalfe and colleagues place "the main mass extinction level in Bed 25," with another conodont restricted "to the narrow boundary interval immediately above," in Beds 26 to 29.  Bed 28 lies "just 8 cm above the GSSP boundary."²³  A study of 162 genera and 333 species through the section "confirms a sudden extinction event."²⁴

Meishan is thin at this level, and its students say so.  At Shangsi, another Chinese section, the boundary conodont first appears "about 5 m above the highest level from which a typical Permian fauna is recovered," which, Metcalfe's team wrote, "may suggest that that some section may be missing at Meishan."²³  Where the sequence is condensed, a loss spread over some thickness of rock elsewhere is squeezed into a few beds.  That is Holland and Patzkowsky's warning from Part 1, made by the people who measured the section.

Looking across seven Chinese sections and 537 species, Haijun Song and colleagues found two steps, not one.  The first, in the latest Permian, removed "57% of species, namely all plankton and some benthic groups, including algae, rugose corals, and fusulinids."  The second, in the earliest Triassic, removed "71% of the remaining species."²⁵  Between them lie the first Triassic beds, where a few Permian holdovers persist.  In the Paleobiology Database, 107 of the 115 productid brachiopod records above the Permian come from the Griesbachian, the first Triassic substage, among them Paryphella, Fusichonetes and Spinomarginifera.  Above it there are essentially none.¹⁰

On land

The best land record is the Karoo Basin of South Africa, where the Permian Daptocephalus fauna gives way to the Triassic Lystrosaurus fauna.  The mass-death beds of the Karoo are treated in Fossil Graveyards;  what matters here is the order.  Roger Smith and Jennifer Botha found "a 69% mass extinction of Late Permian terrestrial vertebrates" followed by "a lesser extinction event (31%)" involving four taxa that crossed the boundary, and a "main extinction zone" about 40 m thick at two separate sections.²⁶  Pia Viglietti and colleagues, with a larger database, found extinction spread "through a protracted interval," "initially co-occurring with low origination rates," with the peak "near the Daptocephalus–Lystrosaurus declivis Assemblage Zone boundary."²⁷

Two disputes run through this record, and both favour caution.  Robert Gastaldo's team accepts "an abrupt turnover" between the two faunas but argues that it happened before the marine extinction.²⁸  That argument rests on radiometric dates, which this series sets aside, so it is noted and not weighed here.  And Hendrik Nowak and colleagues found that "evidence for a mass extinction among plants in the latest Permian is not robust," because the plant record is so unevenly sampled.²⁹  Tomkins and Clarey make a related observation from the young-earth side:  land plants show "a more tiered extinction, with many of their fossils extending well into the Triassic."¹  On this they are right.  The end-Permian was a marine catastrophe first.

What came back, and in what order

Above the kill, the same sequence turns up section after section.  It is the core of this article's question, so it is set out in order.

1. Hardy survivors take over. The aftermath is dominated by what Song and colleagues call "opportunistic organisms," naming "cyanobacteria, small foraminifers, Claraia bivalves, and linguloid brachiopods."³⁰ On land the earliest Triassic is dominated by Lystrosaurus. Sean Modesto has argued that the textbook label "disaster taxon" does not fit it, and that at most its species "could be described as opportunistic species."³¹
2. Everything is small. Jonathan Payne's global database of snail shells found that "no specimens of unambiguous Early Triassic age larger than 2.6 cm have been reported, in contrast to common 5–10-cm specimens of both Permian and Middle Triassic age." The shrinking was sudden and the return slow: "The loss of large gastropods is abrupt even at a fine scale of stratigraphic resolution, whereas the return of larger individuals in the Middle Triassic appears gradual when finely resolved."³² Palaeontologists call this the Lilliput effect.
3. Much is missing. Of the 570 animal genera that the Paleobiology Database records both in the Permian or earlier and in the Middle Triassic or later, 387, or 68 %, have no record in the Lower Triassic between. All 53 demosponge genera among them are missing, most of them reef builders such as Amblysiphonella and Girtycoelia.⁵ Such "Lazarus taxa" could be an artefact of poor collecting. Paul Wignall and Michael Benton tested that and found "no link with the abundance of fossiliferous sites and the proportion of Lazarus taxa," concluding that the gaps reflect "the extreme rarity of organisms at this time."³³ The PBDB numbers agree on the sampling point: the whole Lower Triassic holds 20,730 marine occurrences, about the same as the last Permian stage alone. But preservation matters too. An exceptionally preserved Lower Triassic fauna from Idaho includes leptomitid sponges known otherwise from the Cambrian and Ordovician, "a 200–million year Lazarus taxon."³⁴
4. The sea floor stays quiet, then stirs in steps. In the Werfen Formation of the Italian Dolomites, 250 to 700 m of Lower Triassic shelf sediment,³⁵ Richard Hofmann reports that in the lowest Triassic member "bioturbation virtually vanished," that "a first significant rebound in ichnodiversity" comes in the lower Siusi Member, that the next two members "again record a drop," and that the highest member "signifies the most advanced recovery stages."³⁶ William Foster and colleagues, sampling 328 fossil beds in six Werfen sections, found two further bursts of extinction within the recovery, "not only in the Dienerian, as previously recognised, but also around the Smithian/Spathian boundary."³⁵ In Utah and Nevada, the last Lower Triassic stage still shows "persistent small size, low average ichnofabric index, low bedding-plane coverage, and reduced tiering" of burrows.³⁷
5. Swimmers return first; animals fixed to the sea floor return last. Song, Wignall and Alexander Dunhill found that non-motile animals, those fixed to or lying on the sea floor, fell "from more than 500 genera in the Late Permian to fewer than 100 genera in the Early Triassic" and reached pre-extinction levels only "after a rapid rebound in the initial Middle Triassic."³⁰ By contrast, recovery "was rapid for some nektonic-pelagic groups, such as ammonoids and conodonts."³⁴ Marine genus diversity as a whole "attained pre-extinction levels in the early Anisian," the first Middle Triassic stage.³⁰ That is where the PBDB check finds the peak of new genera.
6. No reefs. Reefs built by animals and algae vanish at the boundary. Brian Kelley and colleagues summarise earlier work that "metazoan-algal reefs were absent for 8−10 m.y. following extinction," and show that when they returned, at the top of the Lower Triassic, "calcareous sponges, calcareous algae, and early scleractinian corals occurred progressively."³⁸ The corals of the returning reefs were a new group, the scleractinians, not the rugose and tabulate corals of the Paleozoic.
7. No coal. Gregory Retallack and colleagues: "Early Triassic coals are unknown, and Middle Triassic coals are rare and thin." They found the sedimentary settings that hold coal elsewhere, with "fluctuating sea levels and sedimentary facies and paleosols," present in Lower Triassic rocks without the coal.³⁹ Their explanation, the extinction of peat-forming plants, is weakened by Nowak's finding; the gap itself is not.

How the food web was rebuilt is disputed among mainstream workers, and the disagreement should be reported.  Zhong-Qiang Chen and Benton read the record as rebuilt "stepwise from low to high trophic levels," with "a stable, complex ecosystem" not back "until the beginning of the Middle Triassic."⁴⁰  Song and colleagues found the opposite direction:  "animals at high trophic levels recovered faster than those at lower levels."³⁰  Arnaud Brayard's team described an Idaho fauna from the upper Lower Triassic that is already "trophically multileveled ... from primary producers up to top predators."³⁴  And the Middle Triassic Luoping fauna of China added "new trophic levels ... most notably among top predators in the form of the diverse marine reptiles."⁴¹  What all four agree on is the frame:  a crash at one level, a long interval of impoverished, small-bodied, opportunist-dominated communities, and a return of full complexity some distance up the section.

Part 3: The end-Cretaceous, centimetre by centimetre

The crater left by the Chicxulub impact, and the worldwide layer of ejecta at the boundary, are covered in Impact Craters and the Flood.  The chalk below the boundary and the bryozoan limestone above it at Stevns Klint in Denmark are covered in Chalk, Limestone, and the Flood.  This part asks only what died, how abruptly, and what came back.

On land: the last dinosaurs and the fern spike

In the Western Interior of North America the boundary is drawn where Cretaceous pollen types disappear.  Robert Tschudy and colleagues found that it "occurs at the top of an iridium-rich clay layer" and is marked "by the abrupt disappearance of certain pollen species, immediately followed by a pronounced, geologically brief change in the ratio of fern spores to angiosperm pollen."⁴²  That jump in fern spores, the "fern spike," is the expected first stage of plant succession on devastated ground.  Vivi Vajda and colleagues found the same thing, with an iridium layer, in New Zealand:  "a diverse flora being abruptly replaced by one dominated by a few species of fern."⁴³

Whether the dinosaurs died out suddenly or dwindled was argued for decades, and the argument turned on statistics.  Peter Sheehan's team surveyed the Hell Creek Formation for three years and found "no evidence (probability P < 0.05) of a gradual decline of dinosaurs."⁴⁴  Stuart Hurlbert and David Archibald replied that the methods "provide no basis for choosing between scenarios of abrupt extinction and gradual decline."⁴⁵  Their reply cut both ways, and its title says so:  "No statistical support for sudden (or gradual) extinction of dinosaurs."

Field work since then has narrowed the gap.  The Hell Creek "3-m gap," a barren interval below the boundary once read as a decline, proved partly a problem of where the boundary was drawn:  it "actually lies anywhere from –5 to +3 m relative to the contact" between the formations.  At Camel Butte, Montana, Tyler Lyson and colleagues fixed the boundary by pollen samples taken centimetres apart and found a horned dinosaur's brow horn "in a poorly rooted, silty, mudstone floodplain deposit located no more than 13 cm below the palynologically defined boundary."  Just above it the fern spike begins, with "as much as 80% ferns," and Cretaceous pollen types drop "from 14% in sample 190-1 to 1–3% in the overlying samples."³

Other land animals tell the same story.  Nicholas Longrich and colleagues counted 30 lizard and snake species in the last Cretaceous beds of North America and found "only five" in the Paleocene:  "Species-level extinction was 83%," and "Survival was associated with small body size."⁴⁶  Of birds, "at least four separate lineages of archaic birds persisted up to the K–Pg boundary," and none has been found above it.⁴⁷  The bird record is covered in Archaeopteryx and Bird Origins.

The recovery on land has its own order.  At Corral Bluffs near Colorado Springs, Lyson's team found that above the boundary, mammal diversity doubled first;  then the largest mammal body size tripled, together with a rise in the number of plant species;  and then more large mammals arrived together with "the first appearance of Leguminosae (the bean family)."⁴⁸  The same team's pollen work at Corral Bluffs found the fern spike followed by "low diversity assemblages often dominated by palm pollen."⁴⁹  The order is ferns, then palms and a few flowering plants, then a richer forest, then legumes, with the mammals growing in size and number alongside.

In the sea: plankton, ammonites and the first burrows

The open-ocean plankton give the cleanest record, because they lived everywhere and sank everywhere.  The standard zones for the first Paleocene stage were set out in 1999 by Richard Olsson and colleagues.  The lowest, P0, is the range of a single Cretaceous survivor, Guembelitria cretacea, between the last appearance of the Cretaceous species "at the K/P boundary as delineated by the essentially global iridium spike" and the first appearance of the first new species, Parvularugoglobigerina eugubina.⁵⁰  Then come, in fixed order, the first appearances of Subbotina triloculinoides, of Globanomalina compressa or Praemurica inconstans, and of Praemurica uncinata.⁵⁰  The new microperforate species "evolved from the survivor-species Guembelitria cretacea."⁵⁰

At El Kef in Tunisia, the reference section for the base of the Paleocene, the iridium shows "a millimetre thick pulse-like feature occurring in perfect coincidence with the beginning of the biological crisis."  Ammonites there end below it:  "A rich assemblage, consisting of at least 17 taxa, occurs in the interval from meters 7 to 2 below the K/Pg boundary."⁵¹  That two-metre gap in the very section that defines the boundary is the Signor-Lipps effect at work.

Inside the Chicxulub crater itself, Christopher Lowery and colleagues read the first Paleocene sediment from a drill core.  Its numbers are depths in the core, so they decrease upward.  Burrows of Planolites and Chondrites "characterize the upper 20 cm of the transitional unit (above 54 cm)," evidence "for benthic life in the crater within years of the impact," on the authors' reading.⁵²  Above that, the lowest sample of the overlying limestone, at 34 cm, holds the first new plankton species, P. eugubina, together with the survivor G. cretacea, and "many other species that originate within Zone Pα first appear a few cm higher in the section (31–32 cm)."⁵²

A few ammonites outlived the boundary for a short time.  At Stevns Klint, Marcin Machalski and colleagues found ammonites in the Cerithium Limestone 10 and 37 cm above the base of the 5 to 10 cm boundary clay, the Fiskeler, and concluded that "our results confirm ammonite survival into the Danian for the bulk of the Cerithium Limestone fauna."⁵³  In the whole Paleobiology Database there are 9 ammonite records from the first Paleocene stage against 2,622 from the last Cretaceous stage, and essentially none above.¹⁰

Not every group fell at the line.  Charles Marshall and Peter Ward's statistics on the molluscs of the European Tethys found "a major extinction at or near the Cretaceous-Tertiary (K-T) boundary," but also "gradual extinction of most inoceramid bivalves well before the K-T boundary."⁵⁴  Mass extinctions are often a sharp kill laid over losses that were already under way.

Part 4: The other three, and the smaller events

End-Ordovician: two pulses

The first of the five came in two steps.  David Harper and colleagues summarise it as "two discrete pulses, both linked in different ways to an intense but short-lived glaciation at the South Pole."  The first, "at, or just below, the Normalograptus extraordinarius graptolite Biozone, mainly affected nektonic and planktonic species together with those living on the shallow shelf and in deeper water."  The second, "within the N. persculptus graptolite Biozone, was less focused."  In all, "about 85% of marine species were removed."⁵⁵  Seth Finnegan and colleagues put the loss at "an estimated 61% of marine genera globally" and found that falling sea level and tropical cooling "played important roles in the first pulse" in North America.⁵⁶  The PBDB check shows the same two steps:  47 % of boundary-crossing marine genera last seen in the Katian stage, then 24 % in the Hirnantian.

The order is fixed by graptolite zones that follow one another in the same sequence wherever they are found.  On ICR's timetable both pulses fall inside the Tippecanoe megasequence, within the Flood's first 40 days.⁶  That places the glaciation Harper describes, and a two-step extinction tied to it, inside the opening weeks of the Flood.  No young-earth source found for this article discusses the end-Ordovician extinction in particular.

Late Devonian: a crisis of missing newcomers

The Late Devonian is the event that Raup and Sepkoski found "not statistically significant."  It was real, but it worked differently.  Alycia Stigall's study of the brachiopods and other shallow-sea animals of the time concluded that "the primary cause of biodiversity decline was reduced speciation during the crisis interval, not elevated extinction rates," so that "The Frasnian-Famennian event is, therefore, better termed a “biodiversity crisis” than a “mass extinction.”"⁵⁷  Bambach's team put about two-thirds of the late Frasnian diversity loss down to origination failure.¹³  The PBDB check agrees:  no single Late Devonian stage shows an outsized loss, but the Frasnian made unusually few new genera.  What did collapse was the reef.  Stigall notes that the Middle Devonian held "the most geographically widespread metazoan reef ecosystem in Earth’s history, but its extent was reduced by a factor of 5000 following the crisis interval."⁵⁷  The stromatoporoid-and-coral reefs did not return in that form.

End-Triassic: dinosaurs above, crocodile relatives below

On land, the end-Triassic is best known from the rift basins of eastern North America.  Paul Olsen and colleagues analysed "tetrapod footprints and skeletal material from more than 70 localities" and found that "large theropod dinosaurs appeared" just above the boundary, "synchronous with a terrestrial mass extinction," which was "associated with an iridium anomaly ... and a fern spore spike."⁵⁸  The pattern is the end-Cretaceous pattern in miniature:  iridium, ferns, then a new set of dominant animals.  In the sea the picture is less clear.  Stanley left the end-Triassic out of his estimates because "its timing and list of marine victims remain controversial,"¹⁸ and Bambach's team attribute almost 60 % of its diversity loss to origination failure.¹³  On ICR's timetable the end-Triassic falls late in the Absaroka, close to Day 90;  Tomkins and Clarey note that "crocodylomorphs, pterosaurs, and dinosaurs somehow selectively avoided extinction" at this level.¹

The smaller events

Bambach counted 18 intervals that meet Sepkoski's definition of a mass extinction.¹⁴  The best known of the smaller ones are the end of the Middle Permian, the early Jurassic, the middle Cretaceous and the end of the Eocene.  The PBDB check gives their last stages, the Capitanian, Toarcian, Cenomanian and Priabonian, marine extinction shares of 20, 21, 22 and 15 %.  The Capitanian also made only 4.5 % new genera, the lowest origination of any well-sampled stage.  Stanley points out that the frequently quoted 90 to 96 % for the end-Permian "were incorrectly based on combined data for the Middle and Late Permian mass extinctions," that is, by merging the Capitanian event with the end-Permian one.¹⁸

One more extinction belongs in any list but not in this argument.  The large mammals of the last Ice Age, mammoths, ground sloths, sabre-toothed cats and many others, died out in rocks that every young-earth model places after the Flood.  Austin and colleagues already allowed that the Ice Age "contributed to the megafaunal extinctions of the Pleistocene."¹¹  That event is treated with the post-Flood world in The Ice Age.

Part 5: Where young-earth models put each extinction

Every young-earth model reads most of the fossil-bearing rock as the record of Noah's Flood.  So every model must say where each extinction level falls in the Flood year, and whether it records a real death or only the order of burial.  The answers below come from the models' own primary papers.

Four of the five are inside the Flood on every live model

The Institute for Creation Research (ICR) gives the most detailed timetable.  Tim Clarey and Davis Werner tie each of Sloss's six megasequences, the continent-wide packages of sediment that frame their model, to days of the Flood.  The Sauk, Tippecanoe and Kaskaskia, which hold the Cambrian to Mississippian rocks and the end-Ordovician and Late Devonian extinction levels, "may represent the first 40 days of the Flood."  "We interpret Days 40 to about 90 of the Flood as corresponding to the Absaroka megasequence," which holds the end-Permian and end-Triassic levels.  The Zuni runs "from about Day 90 of the Flood to Day 150," and its top, at the end-Cretaceous, is "the highest point of the Flood."⁶  The rocks above, the Tejas megasequence, are "the receding phase of the Flood."¹

Andrew Snelling of Answers in Genesis puts the end of the Flood "certainly" above the end-Cretaceous, so all five extinctions are Flood events on his model too.⁵⁹  Kurt Wise and Matthew Richardson place the start of the Flood in the Precambrian-Cambrian rocks and its end "just below the Danian/Selandian boundary," the first stage boundary above the end-Cretaceous.¹⁶  On their reading the end-Cretaceous extinction and most of the first Paleocene stage, with its first recovery faunas, were laid down in the closing days of the Flood.

The only published young-earth model that made the end-Permian post-Flood was the 1996 "recolonisation" proposal of four British creationists.  In 2000 they wrote:  "None of us is satisfied with the model that we proposed in 1996."⁷  The Flood/Post-Flood Boundary article records that no active camp now puts the end of the Flood below the Cretaceous.  So the end-Permian extinction and the whole Early Triassic recovery above it are Flood rock on every model now defended.

The end-Cretaceous divides the camps

The end-Cretaceous is where the boundary fight sits, and the two camps read it in opposite ways.

• The end of the Flood. Steven Austin and five colleagues founded catastrophic plate tectonics in 1994 and wrote: "we tentatively place the Flood/post-Flood boundary at approximately the Cretaceous/Tertiary (K/T) boundary."¹¹ John Whitmore and Paul Garner's criteria put it "somewhere in the Lance or Fort Union Formation" in Wyoming, the two formations that straddle the end-Cretaceous.⁹ On this reading the extinction is the Flood's last kill, and everything above it, including the whole Paleocene recovery, is post-Flood.
• The high-water mark. ICR puts the end-Cretaceous at Day 150, "when all land animals finally succumbed and drowned," in the words Clarey uses for this view.²⁰ Michael Oard, who places the end of the Flood late in the Cenozoic, wrote that the dinosaurs "simply died in the Genesis Flood (except those dinosaurs likely taken on the Ark, which probably died soon after the Flood)."¹⁷ On this reading the Paleocene rocks above the extinction are not a recovery at all. They were laid down by the receding Flood, and their fossils are the next communities it reached.

The full boundary debate, including its arithmetic for the mammals, is in The Flood/Post-Flood Boundary.  This article takes only what each camp's placement implies for the extinction and recovery pattern.

Real deaths, or only the order of burial?

Here the young-earth literature splits in a way it rarely acknowledges.

ICR says the extinctions are not extinctions.  Jeffrey Tomkins and Clarey:  "There are no real extinction events in the rock record, only the last appearance of organisms as they were encapsulated by the tsunami-like floodwaters."⁸  In their 2023 conference paper:  "Creation scientists do not consider these as true “extinction” events. … So-called extinctions are merely the level at which certain fossils were no longer being actively buried."¹  Clarey and Werner explain how:  "rapid changes in water level that buried completely new types of organisms from new biozones," "the simultaneous inundation of a complete ecological environment at the same approximate elevation globally."⁶  Clarey called the end-Cretaceous extinction "probably just the last appearance of many organisms in Flood-deposited strata,"²⁰ and he applies the idea to ammonites, which "never really went extinct—they just didn’t get fossilized after a certain level," and "might have survived the Flood."⁶⁰  Carl Wieland of Creation Ministries International (CMI) put the same view in 1998:  in Flood rock, "'extinctions' simply mean points above which there are no more such fossils found buried."²

ICR does not deny that groups died out.  Tomkins and Clarey grant that the Flood destroyed pre-Flood environments, so that "a lot of marine animals" went extinct "during or shortly after the Flood," trilobites and many Paleozoic brachiopods and corals among them.  "The exact reason for this is unclear."¹  The claim is about the rock:  the level where a group stops is set by burial, not by the death of its last member.

Most other young-earth writers treat the losses as real deaths in the Flood.  Whitmore and Garner, founders of the end-Cretaceous camp, use extinctions as one of their criteria for Flood rock:  "most marine extinctions are likely to have occurred directly as a result of the global Flood, due to the non-representation of these organisms on the Ark."⁹  Snelling calls the extinction levels places where a great variety of organisms "were buried on a global scale," by processes that "transcended specific ecological, morphological, and taxonomic groupings," and he accepts impacts and volcanic outpourings as part of the Flood.^{59 59}  Emil Silvestru, then a geologist with CMI, put it most compactly:  "The Flood can wrap the whole nine extinctions in one 400-day event."¹⁹  Leonard Brand and Arthur Chadwick propose that each crisis came "when the geological catastrophe reached some critical point and the remaining members of certain groups of organisms were all killed and buried.  A life zone may have been destroyed."  They add, frankly:  "Those who accept a short-age theory also do not know exactly what happened."⁶¹  Wise explains the end of the Paleozoic land flora and fauna as the destruction of a floating forest that "residual catastrophism" kept from re-forming after the Flood.⁶²  CMI's 2006 answer to a PBS documentary splits the losses by habitat:  marine extinctions in the Flood, land extinctions "in subsequent centuries," since every land kind was on the Ark.⁶³

The table sets the placements side by side.

Two of these positions make predictions about the rock that can be checked with numbers.  Snelling's is that the extinctions "would be expected to coincide with" the megasequence unconformities.⁵⁹  ICR's is that the groups which stop at an extinction level were still alive while the next levels were buried.  Part 6 tests both, and then prices what every model has to fit above the boundaries.

Part 6: Testing and pricing the young-earth readings

Test 1: did the "extinct" groups go on living, unburied?

ICR's reading makes a prediction that can be counted.  If a group "just didn’t get fossilized after a certain level," it was still alive in the sea while the Flood went on burying sea creatures above that level.  Then the question is how much burying went on, and whether a living group could have escaped all of it.

The Paleobiology Database answers the first part.  It holds 124,185 marine fossil occurrences from the whole Permian, 19,697 from its last stage, the Changhsingian, and 404,784 from the Mesozoic rocks above the Permian.  So on ICR's own timetable, the Flood buried more than three times as many sea creatures after the end-Permian level as in all the Permian rocks below it.  These were the same kinds of shelf, reef and open-sea settings:  brachiopods, bivalves, snails, corals, sponges, ammonoids.

Five large Paleozoic marine groups stop at or near the end-Permian level.  The table asks how many fossils each should have left in that later Mesozoic rock if it had lived on at only one hundredth of its share of the last Permian stage.¹⁰

The few survivors that do appear sit where a real extinction puts them.  Of the 115 productid records above the Permian, 107 are in the Griesbachian, the first substage of the Triassic:  the "mixed fauna" of South China, in which Permian-type brachiopods such as Paryphella and Fusichonetes persist into the first Triassic beds and then vanish.  The pattern at the end-Cretaceous is the same.  Ammonites make up 6.5 % of the marine fossils of the last Cretaceous stage.  At that share they should appear about 849 times in the first Paleogene stage, and 41,202 times in all the marine rock after it.  They appear 9 and 11 times, the nine being the Danian survivors discussed in Part 3.  Mosasaurs, plesiosaurs, inoceramid clams and rudist clams, 2,262 occurrences in the last Cretaceous stage between them, have none in the first Paleogene stage.

ICR's own paper supplies the fact that closes the obvious escape.  If these groups lived on in some other sea, mixing from that sea should have carried a few of their shells into later Flood deposits.  Clarey and Werner report the opposite:  "we do not observe significant numbers of reworked early Paleozoic fossils and mixed fossil debris in younger, Absaroka, Zuni and Tejas strata."⁶  Tomkins and Clarey raise the escape themselves for the end-Permian:  "Did larger tsunami-like waves bring in distinctly different marine fossils at the same level?  Or were waves coming from different directions with different marine fossils?"¹  For that to work, every wave after the early Absaroka, some time after Day 40 on ICR's timetable, had to come from seas that held none of these five groups, on every continent, for the rest of the year.

The land-elevation mechanism does not reach this problem at all.  ICR explains extinction levels as water reaching "the next ecological zone" at "higher elevations."⁶⁰  Trilobites, rugose corals and ammonites did not live at elevations.  They lived in the sea, and the sea went on being buried.

Verdict on Test 1.  The reading of the end-Permian and end-Cretaceous as burial order alone predicts thousands of fossils that are not there.  The groups stop because they died.  This does not touch the young-earth writers who say the Flood killed them;  it refutes only the claim that they were never extinct at those levels.

Test 2: do extinctions fall at the Flood's big unconformities?

Snelling's prediction was that "the so-called mass extinctions would be expected to coincide with these major unconformities," the erosion surfaces that bound the megasequences, because each new wave of sediment carried new creatures.⁵⁹

ICR checked this against its own megasequence chart and found that it holds for two of the five.  The end-Cretaceous sits at the top of the Zuni and the end-Triassic near the Absaroka-Zuni boundary;  "the other three major extinction horizons fall within the middle or toward the top of the megasequence boundaries."¹  The largest of all sits in the middle of the Absaroka, and Tomkins and Clarey ask the question themselves:  "why did this event occur in the middle of a global megasequence (Absaroka) and not at one of its boundaries?"¹

Wise and Richardson, testing the same idea against the Paleobiology Database, go further.  They see "no connection" between the megasequences and the biostratigraphic divisions, and the end-Permian "has no special relation to the Absaroka megasequence in which it is found."  Their conclusion:  "It would seem that the sequence of burial of organisms was not at all affected by the changing energy levels of Flood waters."¹⁶  That sentence is a creationist's, and it removes the mechanism by which Snelling expected the waves to sort the dead.

Verdict on Test 2.  The prediction fails on the young-earth side's own reckoning.  The order of the fossils does not follow the rhythm of the Flood model's sediment.

Test 3: can the Early Triassic recovery fit inside the Flood year?

This test applies to every live young-earth model, because all of them put the end-Permian and the whole Triassic inside the Flood.  On ICR's timetable the Absaroka, which runs from the Upper Carboniferous to the Lower Jurassic, took Days 40 to about 90.  The Lower and Middle Triassic recovery therefore had, at the very most, those 50 days, and in fact a fraction of them.

Three features of that recovery are hard to place inside 50 days, in rising order of difficulty.

The rebuilding is recorded by animals living where they are found.  Burrows cannot be carried in by a current;  they are dug in place, in sediment that was the sea floor at the time.  Snelling allows that each burrowed layer represents "the passing of at least a few hours."⁵⁹  The Werfen Formation records not one burrowed surface but a sequence of them:  burrowing nearly absent through the lowest member, a rebound, a fall, and a final recovery with larger burrows, across 250 to 700 m of rock.  At 50 days for the whole Absaroka, that sequence was laid down while the same Flood was depositing the Permian and Jurassic rocks above and below it on other continents.  The hours are not the problem.  The order is:  why should the burrowers of the lowest Triassic all be small and few, everywhere, and the burrowers a few days later be larger and more varied, everywhere?

The same genera vanish and return.  Two-thirds of the genera found below and above the Lower Triassic are missing from it.  On a burial-order model, every wave that reached the sea floor during the Lower Triassic days carried none of them, in every basin, and then the waves of the Middle Triassic carried them again.  Tomkins and Clarey's suggestion, "waves coming from different directions with different marine fossils," would have to switch the source of every wave on Earth at once, twice.

A reef grew in place, and its builders arrived in order.  At the Great Bank of Guizhou in southern China, a platform-margin reef began in the uppermost Lower Triassic and "persisted in situ" into the Middle Triassic, "reaching a stratigraphic thickness of ∼900 m."  Its framework is a boundstone, rock bound together as it grew, with early marine cement.  Its builders enter in sequence up the section:  microbial Tubiphytes at the base, the first reef-building calcareous sponges 365 m above it, the first framework-building calcareous algae at 665 m, and the first scleractinian corals at 895 m.³⁸  The young-earth answer to fossil reefs in general, that they are broken debris or reefs that grew before the Flood, is weighed in Fossil Preservation;  neither answer fits here.  This reef sits on top of Lower Triassic rock that every young-earth model calls Flood sediment, so it could not have grown before the Flood, and its authors describe it as a reef that grew in place.  Modern reef margins in sheltered settings build upward at "1 to 25 mm year⁻¹, with a mode of 9 mm year⁻¹."⁶⁴  Building 900 m inside the 50 days that ICR allows for the entire Absaroka requires 18 m a day, about 6,600 m a year, which is roughly 260,000 times the fastest of those averages.⁶⁵

Verdict on Test 3.  No young-earth publication found for this article addresses the Early Triassic recovery:  not the dwarfing, the missing genera, the reef gap, the coal gap or the stepwise return of burrowers and reef builders.  The deaths at the end-Permian can be fitted to a catastrophe.  The ordered recovery above it cannot yet be fitted to anything inside a single year, and nobody on the young-earth side has tried.  This is an unsolved problem for every live model.

Test 4: the end-Cretaceous recovery, read by each camp

The two camps face different bills here.

The end-of-the-Flood camp predicts the right kind of pattern.  If the Flood ended at or near the end-Cretaceous, the beds just above should record life coming back onto bare ground and into emptied seas.  That is what they show:  a fern spike over an iridium clay, then palms, then a richer forest and the first legumes, with mammals multiplying and growing larger;  and in the sea one surviving plankton species followed by new ones in fixed order.  On this point the camp of Austin, Whitmore and Garner reads the boundary more naturally than the late-boundary camp does.  Their bill is time.  Austin and colleagues fit the whole Cenozoic, with its cooling oceans and its turnover from woodland to grassland faunas, into the centuries before an Ice Age that began "about five centuries after the Flood."¹¹  The cost of that, a doubling of mammal genera roughly every two decades, is priced in The Flood/Post-Flood Boundary.

The camp also faces a consistency problem it has not addressed.  The end-Triassic of eastern North America shows the same signature as the end-Cretaceous:  an iridium anomaly, a fern spike, and a new set of dominant animals just above.⁵⁸  The end-Permian shows the same aftermath of survivors, dwarfs and slow return.  On the end-Cretaceous camp's own model those two events happened in the middle of the Flood, near Days 90 and 40 on ICR's timetable or their equivalents.  If an iridium layer topped by a fern spike and a recovery fauna means "the Flood is over" at the end-Cretaceous, the camp needs a reason why the same sequence means something else twice lower down.

The high-water camp has to read a recovery as a burial order.  On ICR's model the Paleocene beds were laid down by the receding Flood after Day 150, and their fossils are the next communities it reached.  The land plants might be read that way.  The sea cannot.  Planktic foraminifera floated in the open ocean, which has no "higher elevations" for a rising flood to reach.  Yet the same sequence, a single surviving species first and then Parvularugoglobigerina eugubina and the later species in a fixed order, is found from Tunisia to the Chicxulub crater.  Inside the crater, burrowing animals lived on the sea floor before the first new plankton species arrived in the sediment above them.⁵²  A receding flood would have had to sort the world's plankton by species into the same order in every ocean, and to pause for burrowers in between.  Tomkins and Clarey acknowledge that their model "still has a bit of difficulty explaining the trackways and footprints found in some layers of the Cenozoic."¹  The ordered return of the plankton is a larger version of the same difficulty.

Verdict on Test 4.  The end-Cretaceous recovery fits the end-of-the-Flood camp in kind and fails it on time.  It fits the high-water camp in neither.

Objections and responses

"The stage boundaries were drawn where the fossils change, so of course extinctions fall at stage boundaries.  The Big Five are an artefact of the timescale."  Half true.  Many boundaries were indeed drawn at faunal changes, and Wise and Richardson are right that the threshold for a "mass" extinction is a convention.  But every boundary was drawn the same way, and they are not all alike.  On Wise and Richardson's own count, the end-Permian keeps 0.2 % of its species across the boundary where ordinary boundaries keep 3 to 17 %.¹⁶  And the strongest evidence does not use the timescale at all:  a dinosaur horn 13 cm below a pollen change in one Montana hillside, a single extinction bed at Meishan, a 900-m reef whose builders appear in order.

"A last appearance is not an extinction.  The coelacanth vanished from the record and is alive today."  True, and the young-earth writers who make this point are right that the record misses things.  Brand and Chadwick use exactly this example.⁶¹  Lazarus taxa are common:  two-thirds of the genera that range through the Lower Triassic are missing from it.  But the argument proves too little.  A rare fish can hide in the deep sea for a long time.  Productid brachiopods, fusulinid forams and rugose corals were among the commonest fossils of the last Permian stage, with 2,842, 445 and 256 database occurrences there, and outside the first Triassic substage they are all but absent from a later marine record three times the size of the whole Permian.¹⁰  Trilobites, with only 46 occurrences in that last stage, carry less weight on their own, since the coelacanth survived a longer stretch of missing record.  The coelacanth is the subject of the article on Living Fossils.

"The suddenness is an effect of changing rock type, not of a sudden kill."  Clarey reports a survey in which 72 % of sudden extinctions in 110 papers coincided with a "rapid shift in rock type."²⁰  The mainstream literature agrees that rock architecture shapes the pattern:  Holland and Patzkowsky find last appearances clustering at unconformities and flooding surfaces in every mass extinction "with the possible exception of the end‐Cretaceous."²²  So the shape of a loss in one section, sharp or smeared, can mislead.  Two things survive the objection.  Studies built to control for it still find the loss:  Sheehan's team compared "results only from similar facies" and still found no gradual decline of dinosaurs.⁴⁴  And the size of the loss is not a matter of shape.  Clarey's survey has not been published in a form that could be checked:  his reference list gives it as a two-page abstract in the Journal of Creation Theology and Science, Series C, volume 2, which this article could not obtain.

"Mainstream scientists cannot agree what caused any of these extinctions."  True.  Bambach's review says flatly that "No consensus has been reached on the kill mechanism for any marine mass extinction."¹⁴  Impact, volcanism, cooling, warming and anoxia are all argued.  But the dispute is about cause, and this article is about pattern.  No side of the cause dispute questions that the losses happened, in the order the rocks show, or that recovery took far longer than a year.  The disagreement among mainstream workers lies entirely inside a range that is millions of times longer than a Flood year.

"The famous 90 to 96 % figure for the end-Permian is exaggerated."  It is.  Stanley's corrected figure is about 81 % of marine species, and he notes that "About 90 orders and more than 220 families of marine animals survived."¹⁸  This cuts against popular writing on both sides.  It does not change the order of events or the size of the recovery problem.

"Recovery was fast:  burrowers were back in the Chicxulub crater within years, and a complex fauna lived in Idaho early in the Triassic."  These findings are real and the article reports them.  Mainstream authors call them fast in mainstream terms:  years to hundreds of thousands of years.  None of the tests above depends on how many years passed.  They depend on order, on animals living where they are found, and on a reef that grew in place.

"Hydrodynamic sorting puts small shells on top.  The Lilliput effect is sorting."  Sorting by settling makes each event bed fine at the top and coarse at the base.  It does not make every snail in 250 to 700 m of Lower Triassic rock smaller than 2.6 cm, worldwide, and then let 5 to 10 cm snails return gradually in the beds above.³²  And Wise and Richardson found that the order of burial "was not at all affected by the changing energy levels of Flood waters,"¹⁶ which removes sorting by flow strength as an explanation of the large-scale pattern.

"The Flood killed these animals.  Of course they are not found higher up:  they were dead."  That is the real-death reading of Whitmore and Garner, Snelling, Wise and Brand.  It passes Test 1.  Its difficulty is Test 3:  whatever lived through the end-Permian level had to rebuild ordered communities, with dwarfs, absences, a stepwise return of burrowers and a reef 900 m thick, within the days the model allows before the Middle Triassic.

From the skeptic's side:  "Calling the real-death reading 'unfinished' is too kind.  It is simply wrong."  A model can be shown wrong only once it says something specific.  No young-earth publication found for this article says anything about the Early Triassic recovery.  The reef arithmetic shows how large the gap is, and a skeptic may judge that no model will close it.  But the series' rule is to call a position unfinished when the pattern is real, the standard answers fail, and no answer has yet been offered.

From the defender's side:  "Microbial reefs cemented on the sea floor may grow much faster than modern coral reefs, so the 260,000-fold figure is unfair."  This is a fair point, and Issue 3 below records it.  The Guizhou reef's lower part is microbial boundstone with marine cement, not coral.  No published growth rate for such a reef was found.  To close the gap, it would have to build about 260,000 times faster than the fastest modern coral reef margins, while its builders entered in sequence.

Issues and unanswered questions

1. Several key papers were read only as abstracts. Jin and colleagues' bed-by-bed counts at Meishan, Song's two-pulse data, Chen and Benton's review, Lyson's metre-by-metre recovery at Corral Bluffs, Twitchett's Lilliput study and Signor and Lipps's original paper were not obtained in full. The claims made from them here are the claims in their abstracts.
2. The database check is raw. It is not corrected for sampling intensity, PBDB's environment tags are incomplete, and some "Lazarus" genera may be later look-alikes given an old name. The published sampling-corrected studies agree with its ranking, which is why it is offered as a check and not as a finding of its own.
3. The reef rate compares unlike reefs. Modern reef-growth figures come from coral reefs; the lower Guizhou reef is microbial and cement boundstone. A measured growth rate for a modern or ancient microbial-cement reef would sharpen or weaken Test 3.
4. Nobody has counted the burrowed surfaces. No paper read here counts the burrowed horizons in the Werfen Formation or any other Lower Triassic section. With that count, Snelling's "few hours" per surface could be turned into a minimum time.
5. The South African land record is disputed. Whether the Karoo turnover is the same event as the marine extinction rests on radiometric dates, which this series does not use. The order of the Karoo faunas is not in dispute; their correlation with Meishan is.
6. The end-Ordovician glaciation sits in the Flood's first 40 days on ICR's timetable. No young-earth source found for this article discusses it.
7. A young-earth model of the Early Triassic would need to say where the small-bodied, low-diversity Lower Triassic communities lived before the Flood, why the same genera reappear above them, and how a reef grew 900 m in place during the Flood. A pre-Flood "stressed habitat" zone might answer the first question; it would not answer the other two.
8. A citation defect in the young-earth literature. Oard's 1997 list of objections to the impact theory includes "no statistical support for a sudden extinction of dinosaurs," citing Hurlbert and Archibald 1995, whose title reads "No statistical support for sudden (or gradual) extinction of dinosaurs."¹⁷ The paper found the data unable to decide either way.
9. The Flood/post-Flood boundary is not settled by this article. Test 4 shows that the end-Cretaceous recovery suits the end-of-the-Flood camp in kind; the rate problem that camp faces is in The Flood/Post-Flood Boundary.

Verdict

The Big Five, and the smaller events between them, are not artefacts of how geologists divide the column.  In single sections they appear as losses within centimetres or a few metres, followed by a recognisable aftermath:  short-lived holdovers, opportunists, dwarfs, missing groups, and a stepwise rebuilding of communities and reefs.  The end-Permian is by far the largest on every measure, including a young-earth one.  The Late Devonian and end-Triassic were as much failures of new forms to appear as surges of loss, and the size of the end-Permian has been overstated in popular writing.

Every live young-earth model places the four oldest events inside Noah's Flood and splits over whether the end-Cretaceous is its end or its high-water mark.  The ICR reading, that extinctions are only the levels where each group stopped being buried, fails on its own terms:  even at a hundredth of their last-Permian share it predicts about 740 trilobites, Paleozoic corals, fusulinids and productids in the later marine record, and outside the first Triassic substage there are 12.  Its companion prediction, that extinctions fall at the megasequence boundaries, fails for three of five on ICR's own chart.  The reading in which the Flood itself did the killing survives those tests.  But it has no account of the recoveries:  the Early Triassic, with its dwarfs, its absences and a reef 900 m thick built in place by builders arriving in sequence, has to fit inside a few weeks, and no young-earth author has attempted it.  That position is unfinished, and the size of the gap is now measured.

Sources