Mass Extinctions
in a Hurry

The Whole Thing in One Page

In the Bottaccione Gorge outside Gubbio, Italy, a narrow seam of clay divides two worlds. Below it lie rocks formed while ammonites filled the seas and non-avian dinosaurs ruled the land. Above it, both are gone. The seam contains an unusual concentration of iridium, one clue that led researchers towards an asteroid impact sixty-six million years ago. Rock has compressed catastrophe into something you could cover with a thumb.

That image is accurate for one famous crisis and misleading as a general model. Most mass extinctions were not single blows. Several unfolded in pulses across tens or hundreds of thousands of years. Asteroids explain one member of the traditional Big Five. The others are tied chiefly to rapid changes in climate, sea level, ocean oxygen and chemistry, often associated with immense volcanic provinces. Life did not approach literal sterilisation. Microbes, plants and many animal lineages survived. What ended were living worlds: dominant groups, food webs, reefs and ecological arrangements built over millions of years.

A mass extinction is more than many species dying. Extinction continues between catastrophes. A mass extinction is a geologically sharp, geographically broad rise in loss that cuts across groups and environments, outpaces replacement and changes the structure of life afterwards. The famous five are the largest such crises recognised in the best-sampled marine fossil record since the early Ordovician. They are useful landmarks, not five events separated from every lesser crisis by a natural border. Their species percentages are inferred from an archive that favours hard-bodied marine organisms and records some intervals far better than others.

Each crisis asks two linked questions: what removed the old order, and why did the survivors build something else?

The sequence begins about 443 million years ago, when cooling, glaciation and falling seas were followed by deglaciation and rapidly changing ocean oxygen. The Late Devonian brought repeated crises and the collapse of ancient reef systems. At the end of the Permian, Siberian magmatism drove the largest recognised marine loss through carbon release, warming, oxygen limitation and chemical disruption. End-Triassic volcanism accompanied another pulsed carbon crisis. Then Chicxulub struck at the end of the Cretaceous, darkening the world and breaking food chains based on sunlight.

The trigger is never the whole explanation. A volcanic province does not kill a shellfish merely by existing, nor an impact a distant forest merely by making a crater. Triggers alter air, water, temperature, light, nutrients and habitat; those changes pass through bodies and ecological dependencies. Survival is selective, but the rule changes with the stress, its sequence and geography. A useful trait can become a liability. Luck and timing matter beside physiology.

Recovery is slower than collapse and does not reverse it. Survivors expand into vacancies, but they do so in altered environments and with missing partners, predators and competitors. New worlds emerge because the old cast cannot be reassembled.

The possible sixth extinction has the same structure and an unusual driver: human systems changing land, seas, climate and the movement of organisms. Rates in well-studied groups are exceptional, and pressures are broad, while the realised fraction recorded as extinct remains far below a completed Big Five-scale event. That is neither reassurance nor proof that the label is settled. The trajectory is dangerous and much of its final magnitude remains undecided.

That is the book.

Why You Should Care

The asteroid that ended the Cretaceous did not merely remove dinosaurs. It changed which mammal lineages had room to expand, which plants rebuilt forests, which predators could exist and which bodies would inherit the next sixty-six million years. Your presence is not the intended result of that impact, but it is downstream of it. Without the extinction, there is no reason to expect the same ecological openings, the same mammalian radiation or the same primate history. Mass extinctions are among the moments when contingency stops being a philosophical word and becomes ancestry. They explain why the tree of life has its present shape, but they also expose how many other shapes were once possible and then permanently removed.

That is the first reason to care. Evolution is often presented as a steady branching tree. Mass extinction takes a saw to the tree, but not at random and not according to the rules that shaped ordinary growth. Entire branches disappear. Surviving twigs become the starting material for later forests. To understand why the modern world contains birds rather than pterosaurs, mammals rather than non-avian dinosaurs, and reefs built by one set of organisms rather than another, you need the interruptions as much as the long process between them.

The second reason is that these crises reveal how planetary systems connect. A flood-basalt province erupts on land. Carbon enters the atmosphere. Temperature rises. Weathering and runoff change. The ocean warms, holds less oxygen and may stratify more strongly. Carbon dioxide alters seawater chemistry. Nutrient cycles shift. The biological losses appear in shells and bones far from the lava. Mass extinction turns geology, climate, ocean chemistry and ecology into one causal chain. It punishes explanations that stop at the spectacular first link.

The third reason is practical. Public arguments about present biodiversity loss often collapse into one of two claims: that a sixth mass extinction is already complete, or that the phrase is exaggerated because fewer than three quarters of named species have vanished. Both confuse rate, magnitude and timing. A crisis can remain far below the final palaeontological threshold while losses accumulate unusually quickly. The useful questions are which pressures are operating, how fast they are eroding populations and ranges, and how much of the eventual loss remains alterable.

The fossil record also teaches intellectual discipline. It is incomplete, biased towards hard parts and marine sediments, and blurred by the fact that the last fossil of a species usually predates its true last individual. Yet independent clues can converge: fossil ranges, isotopes, volcanic ash dates, mercury, iridium, shocked minerals, crater rocks and models of climate and ocean circulation. Deep time cannot be watched, but neither is it a licence for any story. The evidence is patchy and still restrictive.

There is a darker attraction too. These are the largest biological disasters recognised in the fossil record, and the details resist cinematic neatness. The end-Permian crisis produced the greatest marine fossil loss and a more prolonged ecological breakdown than the K-Pg event, yet it left no single crater. The Late Devonian may be less an event than a long failure of recovery between repeated blows. The end-Ordovician combined cold, ice and falling seas with a later reversal towards warmth and oxygen stress. Nature has more than one way to dismantle a world.

The final reason is the most useful. Every past mass extinction was beyond prevention by its victims. The present biodiversity crisis is different because its dominant pressures are generated by human choices, institutions and technologies. That does not make rescue easy, and this book is not a conservation manual. It does make fatalism scientifically wrong. The rocks show how quickly losses can compound and how slowly complex systems return. They also show why recognising an extinction before it has reached full magnitude is the only recognition that matters.

The Core Ideas

Extinction Has a Background and a Pulse

A species is a temporary lineage. It originates, persists for some span of time and eventually ends; it may also give rise to descendant lineages before it disappears. Even in long intervals without a famous catastrophe, extinctions continue. This ordinary turnover is called background extinction, though the word background should not make it sound constant. Rates differ among groups, places and periods, and the fossil record adds its own unevenness.

A mass extinction is a departure from that moving baseline. Loss becomes unusually intense, geographically broad and taxonomically widespread. Origination cannot replace diversity fast enough, and the biological structure after the crisis differs from the structure before it. This is why a local die-off, however appalling, is not a mass extinction. Nor is the disappearance of one dominant group by itself. Scale, rate, breadth and consequence all matter. A crisis can therefore be severe without entering the traditional five, and a sharp regional loss can remain biologically different from a global turnover. The category is comparative: it asks how far extinction departed from the range a particular record normally contains.

The arithmetic needs a numerator, a denominator and a clock. Fifty extinctions mean little unless you know whether they came from sixty species or six million, and whether they occurred in a century or ten million years. Palaeontologists often measure the fraction of genera disappearing across a stratigraphic interval because fossil species are hard to identify consistently across the whole world. A genus is a broader unit, so converting genus losses into exact global species percentages requires modelling and assumptions. The familiar definition of a mass extinction as more than 75 per cent of species lost in a geologically short interval is a useful convention, not a test the rocks perform for us.

The Big Five arose from attempts to count these changes systematically. In 1982 David Raup and Jack Sepkoski analysed a large compilation of marine animal families and identified five unusually strong extinction peaks. Later databases added genera, more localities and methods that correct for unequal sampling. The same five remain among the largest marine crises since the early Ordovician: end-Ordovician, Late Devonian, end-Permian, end-Triassic and end-Cretaceous.

Yet modern work also weakens the impression of five separate mountains rising from a flat plain. Extinction intensity forms a continuum. Other crises were severe. Different statistical corrections change the relative heights, and the Late Devonian looks more like an extended interval of elevated loss than one clean spike. The end-Permian is the clearest giant. The lower four can change order depending on data and method. Origination matters as well: a crisis becomes biologically deeper when losses remain elevated while new lineages fail to appear, prolonging the fall in standing diversity.

This does not make the Big Five arbitrary. Landmarks remain useful even when borders are fuzzy. The error is to treat the list as a natural law, as though Earth contains a counter that clicks only when a crisis qualifies. A mass extinction is a biological regime shift inferred from rates and consequences. Five is the traditional number because five peaks dominate the best global marine record. Six is the warning that a new trajectory may be forming before its final magnitude is known.

The Fossil Record Is an Unequal Archive

Imagine comparing two ancient communities. One lived on a shallow sea floor and contained animals with thick shells. The other lived in a tropical forest and contained soft-bodied worms, insects, fungi and small vertebrates. The first community is much more likely to leave a crowded page in the geological archive. The second may vanish with barely a mark. A global extinction history built from fossils therefore begins with unequal witnesses.

Preservation favours hard parts, rapid burial and environments where sediment accumulates. Marine invertebrates are especially useful because they are abundant, widespread and often preserved in layered rocks that can be matched across regions. Terrestrial organisms enter a patchier record. Mountain environments erode. Tropical soils recycle bodies quickly. Small, soft and rare organisms are easy to miss. The traditional Big Five are consequently defined chiefly through marine animal data, even when land ecosystems also suffered.

Time is distorted too. A fossil's last appearance is almost never the death of the last individual. Rare species become harder to find as they decline, and collectors cannot search every bed. Their observed ranges tend to stop below the true extinction horizon, an effect named after Philip Signor and Jere Lipps. A sudden extinction can therefore look gradual. The reverse error is possible when a gap in sediment removes part of the record and forces many ranges to end at the same surviving surface. A missing page can masquerade as a simultaneous ending. Fossils can also be reworked from older sediment into younger rock, making a taxon appear to survive beyond its true range. Apparent gradualism and apparent suddenness both require sedimentary checking before they become biological claims.

Rock availability changes across time and geography. Sea level determines how much shallow marine sediment is deposited and later exposed. Some intervals are represented by thick, fossil-rich sequences; others by erosion or chemically altered rock. Intensive study creates its own map: formations near roads, quarries and long-established universities are sampled more heavily than inaccessible regions. A raw count of fossil genera can therefore rise when more rock is available, even if living diversity did not.

Researchers respond with several controls. They compare well-sampled groups, standardise the amount of fossil material drawn from each interval, model preservation and sampling, test whether losses cluster at independently dated boundaries, and seek physical evidence of environmental change. The strongest cases do not rest on a single diversity curve. At the Cretaceous-Palaeogene boundary, fossil disappearances coincide with a global iridium-rich layer, shocked minerals, impact ejecta and the Chicxulub crater. At volcanic crises, ash dates, mercury anomalies, carbon-isotope shifts and climate proxies can be placed beside the biological turnover.

The archive still refuses an exact body count. Claims that a given event killed a precise percentage of every species on Earth are usually more confident than the data. Even the end-Permian, the largest event, has estimates that depend on how genus losses translate into species losses and how background turnover is removed. The old figure of 96 per cent is an upper estimate that became a slogan. One statistical reanalysis estimated that about 81 per cent of marine species disappeared. That is a model-based correction, not a census or a universally accepted replacement figure for all life.

The right response is neither credulity nor surrender. The fossil record is biased, but its biases can be studied. It cannot tell us every victim. It can tell us when loss rose far above ordinary turnover, which groups suffered, how environments changed and whether the living world that followed was structurally new.

Causes Kill Through Chains

The most famous trigger in extinction history is a rock from space. That fame encourages a bad explanatory habit: naming the initiating event and treating the biology as solved. Chicxulub made a crater about 180 kilometres wide, but distance from the crater alone did not determine survival. Organisms died because the impact altered heat, light, air, water, food and habitat. The trigger began a chain. The chain did the killing.

Every mass extinction must be separated into at least four levels. There is the trigger, such as an impact, flood-basalt volcanism or rapid glaciation. There are physical and chemical stressors, such as cooling, warming, darkness, acidification, sea-level change and oxygen loss. There are biological mechanisms, including starvation, failed calcification, heat stress, habitat removal and the collapse of ecological partners. Then there is the final pattern of extinction and survival. Evidence for one level does not automatically establish the next.

Large igneous provinces show why the distinction matters. They are not single volcanoes with one explosive afternoon. They are regions where enormous volumes of magma erupt or intrude in pulses, often across hundreds of thousands of years. Lava destroys habitat locally, but global extinction is tied more closely to gases and feedbacks. Sulphur aerosols can cool climate over short periods. Carbon dioxide and thermogenic gases can drive longer warming. Intrusions into carbon-rich sediments may release additional greenhouse gases. The rate and pulsing of release can matter as much as the total because oceans, soils and weathering need time to absorb disturbance.

Warming then propagates. Warmer seawater holds less dissolved oxygen. Stronger stratification can reduce the mixing that replenishes deep water. Metabolism speeds up, raising animals' oxygen demand while supply falls. Nutrient runoff and changing circulation can increase productivity in some surface waters, followed by decomposition that consumes oxygen below. Anoxia means severe oxygen depletion; euxinia adds dissolved sulphide, which is toxic to many organisms. A carbon pulse can also lower ocean pH and carbonate saturation, making shell and skeleton formation harder for some calcifiers. These pressures interact rather than queue politely. Buffering also has a speed limit. Oceans can absorb carbon and rocks can consume it through weathering, but a rapid pulse can outrun those processes. The same total disturbance spread over longer time may permit migration, adaptation or chemical compensation that a faster release denies.

Cooling can be as destructive. The end-Ordovician crisis coincided with glaciation and a major sea-level fall. Much of animal diversity lived in shallow continental seas. When water was locked in ice and shorelines retreated, habitat area contracted. A later shift towards warming and deglaciation brought a different set of stresses, including widespread oxygen loss in some basins. One crisis could therefore select against organisms first under cold, shrinking seas and then under warmer, poorly oxygenated water.

Chicxulub followed another route. Dust, soot and sulphate aerosols reduced sunlight after the impact. Photosynthesis fell, primary production crashed and food chains were starved from the bottom. Rapid cooling was followed by longer adjustments, while acid rain and ocean acidification added regional and taxon-specific stress. The violence at ground zero matters, but the global extinction pattern requires the atmospheric and ecological aftermath.

No single kill mechanism explains every victim, and no environmental proxy proves a complete causal chain by itself. Good extinction science asks whether timing matches, whether the proposed stress reached the right places, whether vulnerable traits predict the losses, and whether models can reproduce the scale. The question is not what dramatic thing happened. It is how that thing reached living bodies.

Disaster Selects, but the Rules Change

Mass extinction is selective. If it were wholly random, losses would mirror the composition of the pre-crisis world and survival would be a lottery alone. Instead, some habitats, body plans and ecological strategies repeatedly fare better than others within a particular event. The difficult part is that the rule differs among crises, and ordinary evolutionary success may offer little protection when conditions change abruptly.

Geographical range often helps. A lineage spread across several regions and environments is less likely to lose every population to one local failure. Abundance can help because rare species have fewer chances to persist. Flexible diets, broad tolerances, dormancy and rapid reproduction may matter when food webs or climates shift. Small bodies can lower absolute food demand, though smallness is not a universal shield. Freshwater systems, deep water, high latitudes or burrows can act as refugia in one event and traps in another.

Ecological dependence is often the hidden variable. A predator can survive the first physical shock and still disappear when its prey collapses. A plant may endure while its pollinator does not. Reef organisms lose both living partners and the three-dimensional habitat the reef constructed. Specialists can be efficient in stable worlds and fragile when one required resource vanishes. Generalists may persist through scarcity, but that advantage also depends on what remains available.

The end-Cretaceous event makes the change in rules visible. Large non-avian dinosaurs had dominated terrestrial ecosystems for more than 150 million years. That history did not protect them from a crisis in which food supply and habitat were abruptly reorganised. Birds survived, but many bird lineages died too. The survivors were not a representative sample of all birds, and researchers still debate the contribution of body size, diet, habitat and developmental strategy. Crocodilians, turtles and some freshwater groups did relatively well, possibly because detritus-based food chains and aquatic refuges were less tightly coupled to immediate surface photosynthesis. That is a conditional account, not a universal freshwater exemption.

At the end-Permian, heat and oxygen stress linked physiology to geography. Marine animals already living near their thermal limits, or with high oxygen demands, could lose viable habitat as warm, oxygen-poor conditions expanded. Tropical losses and movements towards cooler regions fit part of this model. Yet acidification, toxic chemistry, habitat change and regional variation also mattered. One physiological axis can explain a large pattern without explaining every fossil.

A trait can reverse value within one crisis. Calcifying shells can become costly under acidification, yet mineral type, physiology and local water chemistry alter the response. A wide range can buffer regional loss until a pressure becomes global. Dormancy can outlast one bad season but not a century of unsuitable habitat. Selectivity is a change in probability, not a complete sorting rule.

Chance remains. A species may possess useful traits but occur only where the stress is worst. A small founding population may survive the crisis and then disappear through bad reproduction or another pulse. Fossil sampling can hide the populations that nearly made it. Survival therefore combines properties of the organism, its relationships, its location and the exact sequence of disturbance.

This is why calling survivors the fittest is empty unless fitness is tied to a defined environment. The ammonite was not poorly adapted to the Cretaceous ocean. It was adapted to an ocean that ended. Mass extinction changes the examination while every species is already sitting it, and it does not distribute the new paper evenly.

Collapse and Recovery Run on Different Clocks

A forest can burn in a day and take centuries to regain old trees. A mass extinction stretches that asymmetry across a planet. The main loss may occupy a geological instant, which can still mean thousands of years, while ecological and evolutionary recovery takes millions. Removing species is faster than rebuilding interactions, habitats and lineages.

The word recovery hides several different measurements. Diversity can rise because new species originate. Abundance can return while the community remains dominated by a few opportunists. Food webs can regain several trophic levels. Reefs can resume building large structures. Geographic ranges can expand. These processes need not move together. A post-crisis sea may contain many species yet lack the functional complexity, body sizes or habitat engineering of the former system.

Immediately after severe extinction, simple and tolerant organisms can become abundant. Palaeontologists sometimes call them disaster taxa when they spread through disturbed environments. Their success is informative: abundance after a crisis may signal ecological simplification rather than health. A microbial mat covering a sea floor has captured available space, but it has not replaced a reef. Thin, small shells can reflect harsh conditions or slowed growth, though the so-called Lilliput effect has several possible causes and is not uniform.

Recovery cannot begin in earnest while the forcing continues. The end-Permian crisis was followed by repeated carbon-cycle disturbance, extreme warmth and widespread oxygen stress. Some lineages survived the boundary yet failed to diversify for a long interval. Others reappeared after seeming absent, either because they persisted in unsampled refuges or because the record is incomplete. The biological bottleneck was followed by an unstable recovery environment, so the clock did not reset cleanly on the boundary date.

Interactions create further delay. A reef builder needs suitable chemistry, light, substrate and partners. A large predator needs enough prey distributed across a functioning landscape. Plants, herbivores, decomposers and soil systems have to establish reciprocal flows. Evolution can generate new forms, but each branch starts from surviving variation. Lost developmental architectures and whole clades cannot be ordered back into existence.

This makes recovery path-dependent. The first organisms to expand can alter environments and occupy niches, affecting who arrives next. Geography channels dispersal. Continental positions, currents and climate differ from those before the crisis. A group that survives in one refugium may become the source of a later radiation, while an equally capable group isolated elsewhere remains minor. The rebuilt system records both environmental opportunity and historical accident.

The fossil record adds uncertainty to the timetable. Low diversity immediately after a boundary may partly reflect poor rock or preservation. High origination may partly reflect the naming of short-lived forms. Researchers therefore compare taxonomic counts with body size, ecological roles, reef volume, burrowing depth and other signs of function. No single line marks recovery complete. Evolutionary recovery and ecological recovery can also separate. A new species may differ enough to count taxonomically while performing a familiar role, or a surviving lineage may evolve a new role before total diversity returns. Counting branches and rebuilding a working network are related but different histories.

The central fact survives these complications. A mass extinction can remove evolutionary work accumulated across tens of millions of years, and a few million years can pass before comparable structural complexity returns. Geological time makes that interval look small. For any species living inside it, it is the whole future.

Survivors Inherit Vacancies, Not Victory

The usual story ends the crisis and begins a triumph. Dinosaurs vanish; mammals take over. Triassic competitors disappear; dinosaurs rise. Extinction clears the stage and the next cast walks on. The sequence is broadly real, but the word takeover smuggles purpose into a process built from vacancies, surviving lineages and conditions that keep changing.

Many supposed winners were present before the extinction. Mammals lived alongside non-avian dinosaurs for well over a hundred million years. Dinosaurs existed before the end-Triassic crisis. What changed was the ecological field around them. Competitors, predators and habitat-forming groups disappeared. Resources and spaces became available, but only lineages that had survived, reproduced and reached those spaces could use them.

Vacancy is not an instruction. A missing large herbivore does not compel a surviving species to become large. Variation must arise, populations must persist, and the environment must reward the change. Ecological opportunity can increase the range of viable experiments, leading to adaptive radiations in which lineages diversify into different forms and ways of living. The direction and speed depend on inherited bodies, developmental constraints, geography and further disturbance.

The K-Pg aftermath illustrates the distinction. Mammalian diversity and ecological variety expanded in the Palaeogene, including repeated increases in body size and the emergence of new feeding strategies. Yet the radiation was not a straight march towards humans or even towards large placental mammals. Several mammal groups flourished and later declined. Birds diversified too. Flowering plants, insects and marine plankton followed their own regional and temporal patterns. The post-impact world was assembled across many branches, not transferred as one throne.

Extinction can also reduce future innovation. Every lost clade removes genetic, developmental and ecological possibilities. The end-Permian crisis did not improve life by clearing away old forms. It destroyed reefs, shortened food chains and eliminated lineages that had survived earlier disturbances. Later diversity is impressive because evolution continued, not because the loss was beneficial. Calling mass extinction creative destruction turns the survival of descendants into a justification for the deaths that made their world.

Incumbency matters between crises. Once a lineage occupies an ecological role, it can make entry difficult for alternatives through competition, predation or prior adaptation. A mass extinction weakens that historical advantage. This helps explain why large evolutionary changes can cluster after crises without requiring the extinction itself to invent the traits. It changed access to opportunity. The removal of incumbents can also release surviving prey from predators, plants from herbivores or smaller competitors from dominant habitat users. These indirect effects mean a radiation can begin because relationships vanished, even where physical resources changed little. Vacancy exists in a network rather than in empty space alone.

The outcome remains contingent. Replay the same physical event with slightly different seasons, ocean circulation, continental positions or survivor populations and the later world could differ. That does not make history unknowable. It means causes operate through a starting state. The same pressure applied to a different biosphere will not produce the same cast.

Survivors therefore inherit three things: empty roles, damaged systems and the burden of being all that remains. Their descendants may radiate, stagnate or vanish in the next pulse. Victory is a word supplied by observers who know which branches reached the present. At the boundary itself, survival was permission to continue, and nothing more.

The Sixth Is a Trajectory, Not a Date

The present biodiversity crisis creates a naming problem. Extinction rates measured over recent centuries appear far above commonly estimated background rates for well-studied groups. Populations and ranges are shrinking, habitats are being transformed, and many species face interacting pressures. Yet the proportion of all species already recorded as extinct is nowhere near the conventional 75 per cent mass-extinction threshold. Is this the sixth mass extinction or not?

The dispute often treats a process as a certificate. Past events receive names after their biological magnitude is visible in rock. We live near the beginning of the proposed sixth, with incomplete knowledge of both the starting diversity and the losses. A strict statement is therefore possible: a full Big Five-scale magnitude has not yet occurred. Another strict statement is also possible: rates, pressures and projected losses are consistent with an incipient mass-extinction trajectory if they persist.

Rate and magnitude answer different questions. Rate asks how rapidly lineages are disappearing relative to the number exposed and the time observed. Magnitude asks what fraction has gone by the end. A short interval can have a high rate while total loss remains limited. If the rate continues, magnitude grows. If pressure is reduced and populations recover, the eventual total changes. The distinction is the reason early recognition has value.

Measurement is difficult. Many species, especially insects, fungi and marine invertebrates, remain undescribed, and many known species have not been assessed. Declines can go unnoticed, while proving extinction requires confidence that no population survives. Background rates inferred from fossils vary by group and method. Recorded losses can therefore miss extinctions, especially in poorly studied groups. The opposite error appears when threatened species or modelled future losses are counted as though they had already vanished. Undercounting and premature counting point in different directions, and neither can be repaired by choosing the more dramatic total.

The causes are clearer than the final count. Human land and sea use removes and fragments habitat. Direct exploitation takes organisms faster than some populations replace themselves. Climate change shifts temperature, water, chemistry and extremes. Pollution alters bodies and habitats. Introduced species and pathogens transform ecological relationships. These pressures interact and move quickly across a world already connected by trade and transport. The distinctive trigger is not humanity as an abstract species but systems of production, consumption, movement and governance that distribute pressure unevenly.

Population loss comes before species extinction. A species can remain officially alive after disappearing from most of its range and losing much of its ecological function. Small, isolated populations carry greater risks from chance, inbreeding, local disasters and disrupted relationships. Waiting for the last individual produces a clean statistic at the point when options are worst. Extinction debt makes the delay more dangerous: populations can persist for a time after habitat has become too small or fragmented to support them indefinitely. Present survival may therefore conceal a future loss already loaded into demography, unless conditions change.

Calling the crisis a sixth mass extinction can communicate scale, but it can also imply that the event is complete or unstoppable. Rejecting the phrase because the final threshold has not been crossed can imply the opposite error: that nothing exceptional is occurring. The scientifically useful formulation is conditional. We are driving a broad elevation in extinction risk and, in well-studied groups, extinction rate. If those pressures persist long enough, the resulting loss could approach Big Five scale. The endpoint is not yet fixed.

This closes the loop. Mass extinction was defined as a trajectory of loss against background, not as one cinematic day. The present supplies the one case in which the agent capable of measuring the trajectory can also alter it. A threshold useful for classifying ancient rock is a grotesque target for deciding when to care.

How It Actually Works

The invention of catastrophe

In the late eighteenth century, workers in the quarries around Paris exposed bones that did not fit any living animal. Georges Cuvier compared them with modern anatomy and argued that mammoths, mastodons and other fossil forms were extinct. This sounds ordinary now. At the time, many naturalists still preferred to believe that unfamiliar creatures survived in unexplored regions or that nature did not permit whole species to vanish.

Cuvier went further. Different rock layers contained different assemblages, and the breaks between them suggested repeated upheavals. His catastrophism later became entangled with scriptural flood stories that were not his scientific method, while Charles Lyell made gradual change intellectually fashionable. Darwin placed extinction inside evolution, as the ordinary failure of lineages competing and changing through time. For much of the twentieth century, sudden global catastrophe looked like a relic of bad geology.

A thin Italian clay layer changed the mood. Walter Alvarez was studying the boundary between Cretaceous and Palaeogene rocks at Gubbio and wanted to estimate how long the clay had taken to form. His father Luis Alvarez, a physicist, suggested using iridium, rare in Earth's crust but more abundant in meteorites. The boundary contained far more than expected. In 1980 the Alvarez team proposed a large asteroid impact. Evidence then accumulated across the world: the same chemical anomaly, shocked quartz, glassy spherules, tsunami deposits and, finally, a buried crater at Chicxulub in the Yucatán.

Two years after the impact paper, David Raup and Jack Sepkoski used a large fossil database to identify five major marine extinction peaks. Catastrophe had become measurable rather than rhetorical. The impact claim was attacked hard, especially before Chicxulub was securely identified, and that resistance improved the case. Researchers had to connect a global boundary layer to a dated crater, distinguish impact effects from Deccan volcanism and explain selective survival. The episode established a standard the other extinctions would also face: a trigger must match the clock, the environmental disturbance and the victims. The list Raup and Sepkoski helped establish now supplies the route through deep time.

Ice at the end of the Ordovician

About 443 million years ago, complex animal life was overwhelmingly marine. Brachiopods, trilobites, bryozoans, corals and graptolites occupied warm shallow seas spread across flooded continents. The preceding Ordovician had been a long expansion of marine diversity. Then the climate moved hard in the opposite direction.

The end-Ordovician extinction came in at least two main pulses. Gondwana lay across the South Pole, and ice sheets expanded. Cooling altered the distribution of climate zones. Water locked into ice lowered sea level, draining broad continental shelves where much marine diversity lived. Species adapted to warm, shallow water lost both temperature range and habitat area. The first pulse is strongly associated with this glacial interval and regression.

The second pulse arrived as ice retreated and climate shifted again. Rising seas did not restore the old conditions cleanly. Warming, changing circulation and oxygen-poor water spread stress through marine basins. Recent geochemical work emphasises rapid variability in marine oxygen, while other studies continue to test the roles of volcanism, nutrient supply and the exact tempo of climate change. The organising account is therefore a sequence of climatic reversal, sea-level movement and oxygen stress, not one settled switch.

Selectivity supports the climatic account. Tropical and warm-water taxa were hit strongly during cooling, while widespread forms and lineages able to track shifting climate zones often fared better. Yet the second pulse changed the test, catching survivors in a different ocean. An organism able to endure cold and retreating seas could then face warmth, rising water and reduced oxygen. The extinction was not a single filter repeated twice.

The taxonomic loss was enormous, yet its long-run ecological effect appears less radical than the end-Permian or end-Cretaceous crises. Many major groups survived and resumed familiar roles. Recovery continued during the Silurian, and reefs and marine communities rebuilt without the prolonged suppression of complex ecosystems seen after the end-Permian crisis. The number of lineages lost and the degree to which ecological architecture is replaced are related, but they are not the same measurement. The event supplies an early warning against ranking crises by one percentage alone.

The long failure of the Late Devonian

The Late Devonian resists the word event. It contains at least two major crises separated by roughly thirteen million years: the Kellwasser pulses near the Frasnian-Famennian boundary around 372 million years ago, and the Hangenberg crisis close to the Devonian-Carboniferous boundary around 359 million years ago. Loss also remained elevated through parts of the interval. Treating all of this as one moment makes the Big Five timeline neater and the science worse.

The Kellwasser crises were chiefly marine. Devonian seas contained enormous reef systems built by stromatoporoid sponges and corals, and these habitats suffered a collapse from which the same reef world never returned. Brachiopods, trilobites, ammonoids and many other groups lost lineages. Black shales and geochemical proxies show widespread oxygen stress, but the route into anoxia remains contested and probably differed among basins. Volcanism, climate change, sea-level movement and nutrient supply are all implicated.

The expansion of forests supplies one bounded mechanism. Deeper roots and altered soils could increase weathering and the delivery of phosphorus to rivers and seas. More nutrients can raise surface productivity; decomposition then consumes oxygen below. A 2023 modelling study found that phosphorus from land-plant expansion, combined with volcanic input, could reproduce important features of the Kellwasser oxygen crisis. The result makes terrestrial change a credible contributor. It does not show that trees caused every Kellwasser loss, still less the whole Late Devonian interval.

The Hangenberg crisis was later and should not inherit that model by default. It affected marine groups near the period's end, including the last placoderms, and its terrestrial record includes malformed plant spores interpreted as ultraviolet-B damage after disruption of the ozone shield. That evidence comes from a particular boundary and supports a proposed kill mechanism for terrestrial vegetation there. It does not prove one global ozone history or explain the earlier Kellwasser pulses. Other evidence around Hangenberg points to climate and oxygen change as well.

The biological consequences also differ across the interval. Reef construction was reduced for far longer than the last appearance of any one taxon. Fish communities were reorganised, while the early history of limbed vertebrates passes through a terrestrial record too patchy for a clean winner's tale. Percentages become especially unstable when the crisis interval is long enough for background loss, poor recovery and new pulses to blur together. The honest synthesis is plural: several environmental disturbances struck different systems, and uneven recovery between pulses makes it misleading to treat them as either wholly independent crises or one undifferentiated event.

The Great Dying

About 251.9 million years ago, the end-Permian crisis produced the largest known loss in the marine fossil record and a severe terrestrial collapse. The world before it was already strange by modern standards: all continents were joined into Pangaea, vast interiors were seasonal and dry, and the ocean wrapped around one supercontinent. Then northern Pangaea began to erupt on a continental scale.

The Siberian Traps are a large igneous province covering an immense region with volcanic and intrusive rock. High-precision dates place major magmatic activity before, during and after the extinction interval. The close timing is only the first link. Magma and heat interacted with carbon-rich sediments as well as releasing volcanic gases directly. Carbon-isotope records show a major disturbance to the carbon cycle. The result was rapid greenhouse forcing rather than lava flowing around the globe. A 2026 modelling study tests an additional feedback: vegetation loss and erosion may have increased weathering and phosphorus runoff, delaying peak warming while preconditioning parts of the ocean for anoxia. The sequence is physically plausible within the model and relevant to the observed proxies. It remains a proposed reconstruction rather than a measured global history.

Warmer oceans held less oxygen while animals required more of it. Circulation and stratification helped expand low-oxygen water. Models combining temperature and oxygen reproduce much of the geographical and physiological selectivity seen in marine fossils. Acidification added pressure on calcifying organisms, though its severity varied among settings and through time. In some waters, anoxia became euxinia, adding toxic sulphide. On land, heat, aridity, fire, vegetation loss and soil erosion interacted with the same carbon disruption.

The popular claim that 96 per cent of all species died is too exact and probably too high. Fossil data are strongest for marine genera, and translating them into species losses is model-dependent. One 2016 reanalysis estimated marine species loss near 81 per cent after correcting for background turnover and the relation between genera and species. The interval around that number remains wide enough to forbid a census-like claim. The event's rank as the greatest recognised marine loss of the Phanerozoic does not depend on pretending otherwise.

Precise dating has narrowed parts of the main marine extinction to a short interval on geological scales, but the magmatism and environmental disturbance began before it and continued afterwards. This matters for causation. A close boundary coincidence supports volcanism, while the longer sequence explains why survivors entered an Early Triassic world still exposed to repeated stress. It also explains why one mechanism need not do every job: heat and oxygen demand can organise broad marine selectivity while acidification, sulphide, habitat loss and food-web failure account for additional regional and biological differences.

The boundary did not end the crisis. Early Triassic climates remained unstable, carbon-cycle disturbances recurred and complex ecosystems recovered slowly. Some hardy forms became regionally abundant, including the synapsid Lystrosaurus in parts of Gondwana, but dominance by a few survivors was evidence of a narrowed world. The Great Dying was not one bad day. It was a planetary system driven beyond the conditions under which its existing biological networks could continue.

Carbon at the end of the Triassic

By 201.4 million years ago, Pangaea had begun to pull apart. Rifting that would open the Atlantic was accompanied by the Central Atlantic Magmatic Province, or CAMP, one of the largest volcanic episodes known. Its rocks now lie on several continents because the continents later separated around the newborn ocean. That scattered geography once obscured their unity. Matching ages, chemistry and structure across eastern North America, South America, Africa and Europe turned separate basalt provinces into evidence for one rifting system.

Zircon dates link the first major CAMP pulses closely to the end-Triassic extinction. Mercury enrichments in boundary sediments provide another sign of pulsed volcanic activity. Carbon-isotope shifts and evidence for carbon dioxide released from depth support a link between the eruptions and rapid carbon-cycle change. A 2025 study adds direct, but setting-specific, chemical evidence. Boron isotopes in fossil oysters from Lavernock Point in Wales record a fall of at least 0.29 pH units during the main carbon-isotope excursion after the extinction horizon. The oyster sequence does not span the initial excursion at the extinction itself. Modelling in the same study suggests that the earlier carbon pulse could have caused stronger acidification if its source and release rate were comparable. The first result is a local reconstruction; the second is a conditional inference.

The biological losses included conodonts, many ammonoids, reef organisms and several major terrestrial lineages. Heat, acidification, oxygen stress and ecological disruption provide a plausible linked account, while local records differ and the exact contribution of each pulse remains under study. CAMP did not need to cover every habitat in lava. As at the end of the Permian, gases and ocean response carried the crisis far beyond the province.

The pulsed chronology is especially valuable. A smooth average release can hide short episodes that outpaced chemical and ecological adjustment, followed by intervals in which systems partly responded. The 2025 oyster record strengthens the case for prolonged carbon-cycle disturbance and delayed chemical recovery after the extinction. It does not directly record several global pH falls across the boundary, and it does not make every marine loss a response to acidity. Temperature, oxygen, calcification and food webs changed together. The strongest explanation remains a connected carbon crisis with uneven local expression, not a single global aquarium experiment.

Dinosaurs survived, but the event did not create them. They had existed for tens of millions of years and shared land ecosystems with other archosaurs and large amphibians. Extinction removed many competitors and altered the ecological field. During the Jurassic, surviving dinosaur lineages expanded into a wider range of large-bodied roles. Opportunity came from loss; the forms that used it came from pre-existing history.

The day the light failed

Sixty-six million years ago, an asteroid roughly ten kilometres across struck shallow water near what is now the Yucatán Peninsula. The energy excavated Chicxulub, melted and shattered rock, generated earthquakes and tsunamis, and sent ejecta through the atmosphere. Near the impact, destruction was immediate. Globally, the decisive pressure came from the sky afterwards.

The target rocks contained sulphur-bearing minerals and carbonates. Fine dust, soot and sulphate aerosols reduced sunlight. Temperatures fell sharply for a period, photosynthesis collapsed and food webs lost their energy supply. Acid deposition and rapid ocean acidification added stress, while the severity differed among regions and organisms. The exact duration and temperature of the impact winter depend on modelling assumptions, but the evidence for a dark, cold interval and a primary-production crisis is strong.

Non-avian dinosaurs disappeared, along with ammonites, pterosaurs and many marine plankton. Some birds, mammals, crocodilians, turtles, amphibians, fishes and plants survived, though each group also lost lineages. Seeds, detritus, freshwater food chains, burrows, small body size and dietary flexibility have all been proposed as parts of the survival pattern. None works as a master key.

The Deccan Traps in India were erupting around the same boundary, and their gases affected climate and the carbon cycle. That coincidence produced a long argument over impact versus volcanism. High-resolution dating, global boundary evidence and climate-ecological modelling now support Chicxulub as the principal trigger of the abrupt extinction. Deccan volcanism may have shaped conditions before the impact and recovery afterwards, but it does not explain the boundary pattern as well on its own.

Forests returned, plankton communities reassembled, and surviving birds and mammals diversified. The first post-impact landscapes could be dominated by ferns in some regions, a pattern consistent with rapid colonisation after extensive vegetation loss. In the oceans, the collapse and later rebuilding of plankton altered the biological pump that transfers carbon downward. Recovery rates differed among regions and functions, and some groups expanded long before ecosystems regained former complexity. The old Cretaceous world did not.

Chicxulub is therefore the sharpest natural experiment in the book: a well-dated trigger, a global atmospheric disturbance, selective biological loss and a later radiation connected across one narrow boundary. It is also the exception that distorted public expectations. Four other members of the Big Five need no crater.

The unfinished sixth

Human-driven extinction did not begin with factories. Hunting, introduced species and habitat change accompanied earlier losses, especially on islands and among large animals. The dodo, great auk and thylacine belong to different places and pressures, but each shows how a lineage can move from abundance or local security to absence within historical time.

Industrial expansion increased the scale and speed of land conversion, extraction, pollution, transport and greenhouse-gas release. Modern trade moves predators, competitors and pathogens between biogeographic regions that had been separated for millions of years. Roads and farms divide ranges. Fishing removes target species and changes food webs. Warming and acidification alter the background conditions under which every other pressure acts.

For well-studied groups, observed recent extinction rates exceed fossil-derived background estimates, sometimes by large margins. The exact multiplier depends on the species counted, the baseline chosen and how unrecorded extinctions are estimated. Documented loss is also uneven among taxa, habitats and centuries; island species and some animal groups dominate the recorded toll, and several recent datasets do not show uniform acceleration. Realised global species loss remains well below Big Five magnitude, while declines in abundance, range and population structure are much further advanced than final extinction totals show.

That difference is visible below the species level. A species can retain its name while losing most populations, local adaptations and interactions. Range contraction reduces the places from which recovery could begin, and fragmented remnants are more exposed to fire, disease, drought and demographic chance. These losses are harder to headline than a final extinction, but they are the mechanism by which a broad crisis accumulates.

The modern record also carries its own sampling bias. Conspicuous vertebrates are watched more closely than fungi, insects and many marine invertebrates. Some extinctions are discovered after the fact, while others remain uncertain for decades. Uncertainty about the total should widen the interval, not reverse the direction of well-documented pressure.

The phrase sixth mass extinction therefore describes an unfinished path. It is strongest when used conditionally: continuing present pressures for long enough could drive a loss comparable in scale to the ancient crises. It is weakest when used as a declaration that the threshold has already been crossed or that the ending is fixed. Past victims could not reduce Siberian volcanism or deflect Chicxulub. The current forcing arises from human systems, so the future curve depends on what those systems do.

How we know

Mass extinctions are reconstructed from converging records. Fossil ranges show which taxa disappear and originate through successive layers. Sampling-standardised databases test whether peaks remain after differences in rock volume and collection effort are reduced. Radiometric dates from volcanic ash and igneous minerals constrain timing. Carbon, oxygen, sulphur and boron isotopes record changes in climate, carbon cycling, oxygen and seawater chemistry. Mercury can track large volcanic episodes. Iridium, shocked quartz, spherules and crater rocks identify impact ejecta.

Each line has limits. Hard-bodied marine organisms dominate the fossil record. A taxon's last fossil usually predates its final population, creating the Signor-Lipps effect. Erosion and non-deposition remove time. Isotopes can have more than one cause, and model results depend on boundary conditions and biological assumptions. Correlation in age does not by itself prove the kill mechanism.

Confidence is strongest when independent clocks, environmental proxies, predicted selectivity and biological turnover align. The five crises are secure as major losses. Their exact percentages, durations and regional sequences remain less certain, especially in the Late Devonian. Boundary ages can shift slightly as dating improves. New dates and databases can change the height or shape of a peak without making the vanished world reappear.

What People Get Wrong

"A mass extinction kills almost everything"

The phrase nearly ended life invites a sterilised planet: empty seas, dead continents and a few microbes clinging on. None of the Big Five came close to that. Even the end-Permian crisis left many marine and terrestrial lineages alive. Microbial life persisted widely, and no evidence approaches planetary sterilisation.

The misconception survives because percentages are quoted without a denominator. Losing four fifths of species would be biologically enormous while still leaving millions of species if modern diversity were the starting point. Fossil estimates also concern groups preserved well enough to count, chiefly marine animals, rather than every bacterium, fungus and soft-bodied organism.

What nearly ended was a particular organisation of life. Reefs collapsed. Dominant clades vanished. Food webs shortened. Ecological engineers, predators and partners disappeared together. Survivors entered environments stripped of relationships accumulated over millions of years.

That correction makes the events more interesting, not less. Mass extinction is not failed planetary sterilisation. It is biological replacement on a scale large enough to redirect evolution. Life continues, but the world it was making does not. Survival at the level of life says almost nothing about continuity for any particular lineage.

"The Big Five were five versions of the same disaster"

A row of five skulls on a timeline encourages one category of event: catastrophe arrives, diversity crashes, recovery begins. The shared outcome hides different forcing, tempo and biological reach.

The end-Ordovician involved glaciation, falling seas and a later reversal towards warmer, oxygen-poor conditions. The Late Devonian was prolonged and pulsed, with reef collapse and several plausible interacting causes. The end-Permian and end-Triassic crises are linked to enormous volcanic provinces and carbon-cycle disruption, but they differed in starting world, magnitude and aftermath. The end-Cretaceous was dominated by an asteroid impact and a rapid light and food crisis.

Even within one event, the stress changed through time. Aerosols can cool before greenhouse gases warm. Sea-level fall can remove shallow habitat before a later rise changes oxygen conditions. A second pulse can strike survivors selected by the first.

The correction matters because labels do not explain mechanisms. Past resemblance does not prove that one modern pressure will produce an ancient outcome. Compare causal chains, rates and starting conditions, not the fact that each story ends with fewer fossils. The shared label is an outcome class, not one reusable recipe.

"The asteroid killed every dinosaur at once"

Chicxulub did not send one global blast wave that struck every dinosaur in the same afternoon. Near the crater, heat, ejecta, earthquakes and tsunamis caused immediate destruction. Farther away, the main crisis unfolded through darkness, cooling, failed photosynthesis, starvation and chemical disturbance.

The word dinosaur creates another error. Birds are living dinosaurs. Many Cretaceous bird lineages vanished, but some survived and produced every bird alive now. The extinction removed non-avian dinosaurs, not Dinosauria as a whole.

Nor does survival prove that one trait supplied a universal key. Small size, seeds, detritus, freshwater habitats, burrowing and flexible diets may each have helped particular groups. Geography, habitat and the timing of disturbance affected exposure. Some organisms endured the first shock and then failed as food webs collapsed.

The cinematic version is persuasive because the crater is visible and the boundary is thin. Geological compression turns years into a line. The better model keeps the impact as the principal trigger while following its effects through atmosphere and ecology. That is how one collision became a global biological event. It also explains why distance from Yucatán was an incomplete predictor of survival.

"The Permian extinction killed 96 per cent of all species"

The figure appears everywhere because precision feels authoritative and 96 per cent sounds close enough to total annihilation to carry the story alone. It came from an upper estimate built from fossil losses and assumptions about how genera translate into species. Repetition then removed the assumptions and widened the claim from marine animals to all life.

The rocks do not provide a census of every species. Global fossil compilations are strongest for marine genera. Species are harder to distinguish consistently, and background extinction must be separated from the crisis interval. Different models produce different conversions. Steven Stanley's 2016 reassessment estimated marine species loss near 81 per cent after correcting for background turnover and the relation between genera and species.

That estimate is one reanalysis, not a replacement commandment. Terrestrial magnitude is harder to quantify, and no method can recover every soft-bodied or geographically unrecorded lineage. None of this makes the end-Permian event modest. It remains the largest recognised Phanerozoic marine extinction, paired with severe ecosystem collapse and a long, unstable recovery.

The correction matters because false exactness weakens strong science. The event does not need an inflated slogan. Its rank, causal evidence and biological consequences survive a more honest interval. Uncertainty belongs around the magnitude, not around whether the crisis was exceptional.

"The strongest organisms survive"

Strength has no general biological meaning. An ammonite could be abundant, widespread and exquisitely adapted to Cretaceous seas, then lose the planktonic food base or reproductive conditions it required. A small, unglamorous generalist could persist because it needed less food and used a refuge the crisis spared.

The phrase survival of the fittest is persuasive because it converts hindsight into merit. Survivors are visible, so their traits look like answers designed for the disaster. Yet fitness belongs to an environment. Mass extinction changes that environment rapidly and unevenly. Traits that dominated ordinary conditions can become liabilities, while formerly marginal traits gain value.

Range, abundance, body size, physiology, diet and reproductive speed can influence survival, but none wins every event. The same habitat can shelter organisms during one crisis and expose them during another. Chance matters when disturbance intersects with place, season and small populations.

The correction removes moral language from contingency. Survivors were not the best life had made. They were the lineages whose inherited traits, relationships and locations remained workable through that particular sequence of stress. Some then failed during recovery, which hindsight tends to hide.

"Recovery brings the old world back"

Diversity curves often fall and rise, which makes recovery look like refilling a container. Count returns; crisis over. The same number of taxa can conceal a different ecology.

After severe extinctions, a few tolerant forms may become abundant while food webs remain short and habitats simple. New species can originate before large body sizes, deep burrowing, reefs or complex predator-prey structures return. Some surviving groups persist without regaining their former range or importance. This is survival without recovery.

The old world cannot be reconstructed because extinct clades take their developmental possibilities with them. Environments also change. Continents move, climates settle into new states, ocean chemistry shifts and early colonists affect later opportunities. Recovery begins from the survivors, not from a stored blueprint.

The misconception is comforting because the fossil record proves that life continued after every previous crisis. It does not prove that losses are rapidly reversible. Evolution will make something from what remains. It gives no promise about timescale, resemblance or value to the species living through the gap. Geological recovery is not restoration on a human clock.

"The sixth mass extinction has either happened or it has not"

Ancient mass extinctions are named after their outcome. The present is observed while it unfolds, so demands for a yes-or-no verdict confuse classification with forecasting.

The binary survives because two tests are presented as rival answers. Palaeontological classification asks whether a completed event has removed roughly three quarters of species; by that test the modern total is far short. Conservation diagnosis asks whether current rates and pressures could create such an outcome if sustained; for several well-studied groups, the warning is defensible. One test is retrospective, the other conditional.

Both sides can misuse the ambiguity. Declaring the event complete can suggest inevitability. Rejecting the label can hide severe loss because the final threshold has not been crossed. Threatened species are not dead, but that distinction is a reason to act before extinction, not evidence that the threat is fictional.

The useful question is conditional: what magnitude follows from this rate and these pressures over time? Unlike an ancient boundary, the answer is still being produced. The label should sharpen that fact, not replace it. A warning is useful before the category becomes unanswerable. Geological naming should clarify a preventable process, not postpone attention until prevention has failed.

Use It

Follow the whole causal chain

When a disaster is explained with one noun, keep going. Impact, volcanism, warming and habitat loss identify different levels of a process. Ask what changed physically, how that change reached organisms, which biological dependence failed and what pattern followed.

The discipline transfers whenever one visible event is blamed for a distributed outcome. A trigger can be real and still explain little until exposure, vulnerability and dependence are traced. The name at the top of the chain rarely predicts the victims by itself. Naming the trigger can therefore create the illusion that mechanism is obvious when mechanism is the part that determines who is harmed.

Also test order and duration. An early stress can narrow populations, ranges and ecological options before a later one arrives, so the organisms meeting the second pressure are not the community that entered the first. A cause that fits the boundary date but cannot produce the observed selectivity is incomplete.

The useful question is therefore: what are the links between the dramatic event and the final loss, and which link has direct evidence? A good explanation should survive being unpacked.

Demand a denominator and a clock

A claim that extinction is ten times faster, that half of species are threatened or that diversity fell sharply cannot be judged without knowing what was counted, against which baseline and over what interval.

Begin with taxonomic level. Species, genera and families are not interchangeable. Ask whether the sample covers marine animals, vertebrates, plants or all assessed organisms. Then ask about geography and observation. A global database assembled from well-sampled European and North American rocks may still be informative, but it should not pretend to have equal sight everywhere.

Time changes meaning. Losing ten species in a century from a group of a thousand is different from losing ten across a million years. A brief, high rate need not yet produce a large total, while a moderate rate sustained for long enough can. Background extinction is itself an estimate, not a universal constant.

The same habit improves modern statistics. Before reacting to a percentage, reconstruct the fraction: numerator, denominator, time window and inclusion rule. Precision without those parts is decoration. Mass-extinction research is unusually good training because its most famous numbers are often stronger as orders of magnitude and rankings than as exact global censuses.

Look for selectivity before declaring a winner

After any shock, attention moves to the survivors. The tempting story assigns them a superior trait and calls the result inevitable. Resist that until you know which groups failed, where they lived and what the stress demanded.

Map exposure first. A lineage spread across several regions may persist because one population escaped the worst conditions. Then examine dependence. A predator may be physically tolerant yet lose its prey. A reef animal may lose the structure it inhabits. A specialist can be superb under stable conditions and helpless after one required partner disappears.

Next ask whether the apparent trait survives comparison. If small body size helped, did all small forms do well? If freshwater habitat acted as a refuge, which freshwater groups still vanished? If generalist diets mattered, can the available food sources be shown? A plausible story becomes a mechanism only when it predicts both survivors and victims better than rival accounts.

Finally leave room for chance. Location, season, population size and the order of pulses can divide similar organisms. Outcomes can be selective without being deterministic. The right conclusion is often that a trait shifted odds under defined conditions, not that nature chose the strongest.

Separate survival from recovery

An organisation, population or ecosystem can remain present while losing the capacity it once had. Mass extinctions force this distinction because the fossil record contains lineages that cross a boundary but never regain their former abundance, range or ecological role.

Use four questions. Did it persist at all? Did its numbers recover? Did its function return? Did the surrounding system regain complexity and stability? Each can produce a different answer. A company that survives by dismissing most staff is not operating as before. A woodland that retains tree cover but loses pollinators, seed dispersers and age structure has not recovered merely because it remains green.

Time matters here too. Early post-crisis abundance can belong to opportunists thriving in damaged conditions. Rapid numerical growth may conceal low diversity or dependence on continuing support. Recovery should be tied to the property you care about rather than treated as a mood.

This also prevents a common mistake in personal and institutional decisions: using non-collapse as proof of health. Crossing the boundary is one test. Rebuilding options, relationships and function is another, and it is usually slower.

Protect options before the threshold

The ancient record is full of thresholds recognised after the losses are complete. Present decisions cannot use that standard. By the time three quarters of species are gone, the classification question has become easy and almost every useful intervention has become harder.

Watch leading indicators: shrinking ranges, fragmented populations, loss of abundance, disappearing ecological interactions and dependence on a few refuges. None is identical to extinction. Together they describe how close a lineage or system is moving towards a state from which chance and further stress can finish the work.

The transferable principle is to preserve option value. Connected populations can recolonise. Genetic and ecological diversity spread risk. Multiple suppliers, skills, habitats or routes prevent one failure from becoming terminal. Redundancy looks inefficient while conditions are stable because its benefit is paid only when the ordinary world stops.

Avoid converting this into a rule that every system should maximise spare capacity. Protection has costs and priorities differ. The lesson is narrower: do not wait for a final binary loss when the preceding erosion is visible and cheaper to reverse. A threshold is often the last measurable point in a long causal sequence, not the first moment at which the sequence mattered.

The limits

Mass extinction is a powerful lens for abrupt, broad loss. It is a poor metaphor for every decline. A business failure, cultural change or local ecological disturbance does not become clearer merely by borrowing geological drama. The concept requires a defined population, baseline, rate, spatial scale and consequence.

The past also offers no exact forecast for the present. Continents, species and starting climates differ. Ancient carbon releases were not measured with modern instruments, and their rates remain uncertain. A mechanism observed at the end-Permian boundary cannot be transferred to the twenty-first century without checking magnitude, tempo and geography. Similar ingredients do not guarantee equal outcomes.

Finally, survival history contains moral silence. Evolution records what persisted, not what deserved to. Later radiations do not compensate extinct lineages, and eventual recovery does not make catastrophe acceptable to organisms living through it. The rocks can explain consequences. They cannot choose values or policy.

The one thing to keep

Keep the difference between an event and a change of world.

The spectacular trigger attracts attention because it has a shape: a crater, an eruption, an ice sheet. Mass extinction begins to make sense only when you follow the disturbance through air, water, bodies and relationships, then continue past the last death into the altered recovery. The defining outcome is not a pile of victims. It is that the old system can no longer be assembled from what remains.

That view changes the five ancient crises. The end-Ordovician was climatic reversal across marine habitats. The Late Devonian was repeated failure before recovery could settle. The end-Permian was carbon, heat, oxygen and chemistry acting together. The end-Triassic opened ecological space without inventing the dinosaurs that used it. Chicxulub mattered globally because it interrupted light and food. Each was a different route to the same higher-order result: history lost branches and continued from a narrowed remainder.

It also changes the possible sixth. The decisive question is not whether a committee has awarded the present a geological label. It is whether rates, ranges, populations and ecological functions are moving towards irreversible loss, and whether the forces moving them can be reduced. A completed mass extinction is easy to identify and useless to prevent.

Life has survived every crisis in this book. That sentence is often offered as comfort. It should be heard as a warning about scale. Life survived by becoming different, after millions of years in which no victim returned and no observer was promised a place. Continuation is not restoration. What remains determines what can ever come next.

Terms

Background extinction. The ordinary disappearance of lineages between major crises. It is not one fixed rate: it varies among groups, environments, intervals and methods, so every comparison needs a stated baseline.

Mass extinction. A geologically rapid, geographically broad and taxonomically widespread rise in extinction that overwhelms ordinary turnover and substantially changes the composition or structure of life.

Big Five. The traditional five exceptionally large extinction crises in the well-sampled marine record since the early Ordovician: end-Ordovician, Late Devonian, end-Permian, end-Triassic and end-Cretaceous.

Biodiversity. Variation within genes, among species and across ecosystems, including abundance, functions and relationships. Species counts are one important measure, not the whole of biological variety; mass extinction can therefore damage function before the final species count records the loss.

Species. A separately evolving lineage or population cluster. Different species concepts emphasise reproduction, ancestry, ecology or diagnosable traits, which makes global fossil species counts difficult.

Genus. A taxonomic category grouping one or more closely related species. Palaeontologists often analyse genera because they can be recognised more consistently across incomplete fossil records.

Clade. An ancestor and all its descendants. Extinction can trim species from a clade or remove the entire branch, ending every evolutionary possibility unique to it.

Taxon. Any named biological unit, such as a species, genus, family or clade. Taxonomic rank must be stated because loss at one rank cannot be converted mechanically into another.

Extinction rate. The number or proportion of lineages disappearing per unit of time, often normalised by the diversity exposed. It is distinct from realised magnitude and can rise well before a large total loss.

Origination rate. The rate at which new taxa appear in the record. Diversity falls when extinction exceeds origination and rises when origination remains higher, although ecological function may follow another timetable.

Turnover. Replacement through extinction and origination. High turnover can transform composition without a large net diversity change, while mass extinction produces an exceptional imbalance towards loss.

Standing diversity. The number of taxa present at a given time. Extinction percentages depend on this denominator, which fossil sampling and taxonomic practice can alter.

Boundary. A formally defined division between intervals of geological time, usually fixed at a reference rock section. A boundary may coincide with a crisis without recording every stage, and its numerical age can be revised.

Stratigraphy. The study of layered rocks, their order, correlation and meaning. It supplies the framework that places fossil and chemical changes into a sequence through time.

Biostratigraphy. Correlating rock layers through their fossil contents. First and last appearances can identify relative position, though reworking and incomplete sampling can blur true biological timing.

Radiometric dating. Estimating age from predictable radioactive decay in minerals. Dates from ash beds and igneous rocks help test whether eruptions, impacts and extinctions overlap closely enough for causation.

Proxy. An indirect measurement of an ancient condition. Fossil chemistry, sediments or isotopes can indicate temperature, oxygen, productivity or carbon cycling, but each interpretation needs calibration and context.

Stable isotope. A non-radioactive form of an element distinguished by neutron number. Changes in isotope ratios can help trace carbon-cycle disturbance, temperature, weathering and biological productivity.

Signor-Lipps effect. The tendency for a taxon's last fossil to appear below its true extinction because rare final populations are unlikely to be sampled. Sudden loss can look gradual, especially in sparse sections.

Anoxia. Severe depletion or absence of dissolved oxygen. Warming, stratification and decomposition can expand anoxic waters, excluding animals that require oxygen and altering nutrient and carbon cycles.

Euxinia. Anoxic water containing dissolved hydrogen sulphide. It combines oxygen loss with chemical toxicity and can leave distinctive sulphur, metal and organic signatures in sediment.

Ocean acidification. A fall in seawater pH caused by added carbon dioxide, usually accompanied by lower carbonate availability. Effects differ among organisms, minerals, regions and rates; lower pH is not a universal death sentence.

Large igneous province. A region formed by exceptionally voluminous magmatism over a geologically short interval. Global effects arise chiefly through gases, intrusions and Earth-system feedbacks; timing and pulse rate are central.

Flood basalt. Thick, extensive sequences of basaltic lava produced by repeated eruptions. The Siberian Traps and CAMP include flood basalts but also intrusive rocks and associated gas release.

Extinction debt. Future extinction caused by damage already sustained, such as habitat loss or fragmentation that leaves populations too small or isolated to persist indefinitely. The debt is not fate if conditions can still be restored.

Impact winter. A dark, cold interval after a large impact injects dust, soot or sulphate aerosols into the atmosphere. Reduced sunlight suppresses primary production and carries the impact through distant food webs.

Extinction selectivity. Non-random differences in loss among taxa associated with traits, habitats, ranges, physiology or ecology. Selectivity changes among events because the stress sequence changes.

Refugium. A place where populations persist while surrounding conditions become unsuitable. Refugia can preserve lineages, but isolation and small numbers may still prevent later recovery.

Ecological recovery. The return of abundance, interactions, functions and structural complexity after disturbance. It is not identical to taxonomic recovery or the reappearance of the pre-crisis community.

Adaptive radiation. Rapid diversification of a lineage into different ecological forms, often when new habitats or vacancies become available. Opportunity permits radiation, while inherited variation, geography and competition constrain its outcome.

Go Deeper

Michael J. Benton, Extinctions: How Life Survives, Adapts and Evolves

Thames & Hudson, 2023. Start here for a modern, accessible tour from early animal crises through the Big Five and the present. Benton is a vertebrate palaeontologist who keeps causes, victims and recoveries in the same frame, with enough geological method to show how the claims are built. The coverage is wider than this book and includes additional proposed mass extinctions, which is useful for seeing why the famous five are landmarks within a longer continuum. It is illustrated, current and the easiest next step. Use its broader event list to test where the Big Five framing clarifies the subject and where it narrows attention too aggressively.

Douglas H. Erwin, Extinction: How Life on Earth Nearly Ended 250 Million Years Ago

Princeton University Press, 2006. Read this for the end-Permian crisis in depth. Erwin follows the evidence, competing kill mechanisms and difficult recovery rather than treating the Great Dying as a large percentage with Siberian volcanoes attached. Some dating and geochemical evidence has advanced since publication, so pair it with recent papers for exact chronology. Its lasting value is explanatory: it shows how a specialist reasons from incomplete rocks across several linked Earth systems without pretending that one clue closes the case. The book also shows why recovery belongs inside the extinction story rather than as an optimistic paragraph added after the deaths.

David M. Raup, Extinction: Bad Genes or Bad Luck?

W. W. Norton, 1991. This is the conceptual classic. Raup asks whether extinction mainly removes poorly adapted organisms or whether contingency and catastrophe can erase successful lineages. The examples and datasets predate three decades of improved dating and sampling methods, but the central challenge remains sharp. Read it to understand why extinction changed evolutionary thought: survival cannot always be read backwards as proof of superiority, and the history of life contains filters that ordinary natural selection did not design. Its sceptical tone is useful whenever a tidy survivor story starts to sound like proof of inevitability.

Elizabeth Kolbert, The Sixth Extinction: An Unnatural History

Henry Holt and Company, 2014. Read this for the present crisis through reporting rather than a survey of statistics. Kolbert moves among amphibian disease, ocean chemistry, habitat fragmentation, invasive species and the history of extinction science, making the human-driven pressures concrete. It argues forcefully for a sixth extinction and should be read beside the rate-versus-magnitude distinction in this book. Its strength is moral and observational immediacy. Its limitation is that a narrative of cases cannot by itself settle the eventual global palaeontological classification. Read it for observed mechanisms and human stakes, then return to denominators, baselines and timescales before making a numerical claim.

Notes and Sources

Dates, names and scale

Geological ages in the body are rounded for readability. The reference timescale was the International Commission on Stratigraphy's International Chronostratigraphic Chart, version 2026/06. The book uses 443 to 444 million years ago for the end-Ordovician interval, 372 to 359 million years ago for the main Late Devonian crises, 251.9 million years ago for the Permian-Triassic boundary, 201.4 million years ago for the Triassic-Jurassic boundary and 66.0 million years ago for the Cretaceous-Palaeogene boundary. Boundary ages can change modestly as dating and formal ratification improve. Publication date, observation interval and dataset or chart vintage were checked separately; a new paper can analyse ancient samples, while a new chart can revise the numerical age assigned to the same boundary.

"K-Pg" is the modern abbreviation for the Cretaceous-Palaeogene boundary. Older papers use "K-T", for Cretaceous-Tertiary. "End-Cretaceous" and "K-Pg" are used interchangeably here when referring to the extinction. "End-Ordovician", "end-Permian" and "end-Triassic" identify boundary crises; "Late Devonian" identifies a longer interval containing several pulses rather than one boundary event.

Statements about percentages require caution. The global fossil record is strongest for marine animals and is usually analysed at genus or family level. Species-loss percentages are inferred rather than counted directly. The body therefore gives few precise magnitudes and treats the familiar 75 per cent threshold as a convention rather than a natural switch.

The Whole Thing and Why You Should Care

The Gubbio boundary clay and the iridium anomaly come from Alvarez et al. (1980). The impact interpretation and global evidence are synthesised by Schulte et al. (2010). The statement that Chicxulub affected later mammalian opportunity is a contingency claim, not a claim that the impact made humans inevitable. Jablonski (2001), Erwin (2001) and Hull (2015) support the wider account of mass extinction as an evolutionary filter followed by contingent ecological and evolutionary assembly.

The claim that a trigger must be followed through air, water, bodies and food webs is an organising synthesis across the event literature. It does not imply one universal cascade. Each chronological section identifies the evidence most relevant to its event and states where causal attribution remains open.

Measurement and the Big Five

Raup and Sepkoski (1982) supplied the foundational statistical identification of five unusually large marine extinction peaks. Alroy et al. (2008) and Fan et al. (2020) show how expanded occurrence databases and sampling controls change the shape of long-term diversity and extinction curves. Marshall (2023) is the main source for the book's current framing: the Big Five remain the five largest events since the early Ordovician in corrected marine analyses, but they are not statistically isolated from every lesser extinction and the Late Devonian is unusually extended.

Stanley (2016) is the source for the reassessment of major marine species-loss magnitudes, including the estimate near 81 per cent for the end-Permian event. The book does not treat that model result as a direct census. The 75 per cent convention and the distinction between realised magnitude and current rate are discussed by Barnosky et al. (2011), Marshall (2023), Cowie, Bouchet and Fontaine (2022), and Wiens and Saban (2025).

The fossil archive

The Signor-Lipps effect is named for Signor and Lipps (1982): incomplete sampling tends to place a taxon's observed last appearance below its true extinction. Alroy et al. (2008), Fan et al. (2020) and Marshall (2023) support the discussion of uneven sampling, rock availability, taxonomic level and marine bias. Reworking, non-deposition and erosion can make extinction patterns appear either more gradual or more abrupt than the biological event. The book's confidence language reflects convergence among fossil occurrences, dated strata and independent physical or chemical evidence rather than any claim that sampling problems have been removed.

Cause, selectivity and recovery

Jablonski (2001) is the main source for the evolutionary selectivity and macroevolutionary consequences of mass extinction. Erwin (2001) and Hull (2015) support the distinction among survival, taxonomic recovery and ecological recovery, and the point that recovery is assembly rather than refilling a fixed set of niches. Penn et al. (2018) supply the clearest event-specific example of physiology, temperature, oxygen supply and geography combining to predict end-Permian marine losses.

Claims about broad range, abundance, diet, dormancy, body size and habitat are presented as conditional influences rather than universal survivor traits. The body deliberately rejects a single master rule because selectivity changes with the order, rate and geography of stress. "Disaster taxon", "Lilliput effect" and "refugium" are interpretive terms whose application varies among records; no argument rests on one of them alone.

Extinction science and the impact turn

The historical sketch of Georges Cuvier follows Rudwick (1997), which translates and interprets the primary texts on extinction and geological catastrophe. The Alvarez hypothesis is sourced to Alvarez et al. (1980). The later integration of crater, ejecta, boundary and biological evidence follows Schulte et al. (2010). The account compresses a long and contested research history. It does not imply that impact ideas began in 1980 or that every objection disappeared when Chicxulub was identified.

End-Ordovician

Finnegan et al. (2011) constrain the magnitude and duration of Late Ordovician-Early Silurian glaciation. Finnegan et al. (2012) link climatic change to extinction selectivity. Kozik et al. (2022) provide evidence for rapid marine oxygen variability and support the emphasis on changing oxygen conditions. These studies support a two-pulse model involving cooling, glaciation and sea-level fall followed by deglaciation and renewed oxygen stress.

The exact sequence, regional expression and contribution of volcanism, nutrient change and anoxia remain active questions. The book therefore avoids a single-cause verdict. Its comparison between high taxonomic loss and a more limited long-term ecological reset follows the broad synthesis in Marshall (2023).

Late Devonian

Fan et al. (2020) and Marshall (2023) support treating the Late Devonian as an extended interval of elevated loss with several pulses, especially the Kellwasser and Hangenberg crises. Smart et al. (2023) modelled phosphorus delivery from expanding land plants and volcanism and found that the combination could help produce Kellwasser marine anoxia. This is retained as a leading mechanistic model, not proof that land plants alone caused the crisis.

Marshall et al. (2020) document malformed plant spores across the Devonian-Carboniferous boundary and argue for ultraviolet-B damage associated with ozone disruption during the Hangenberg event. That mechanism applies to a particular pulse and terrestrial record. It is not silently extended to the whole Late Devonian. Claims about reef collapse, placoderm loss and prolonged ecological consequences are kept separate from any single causal hypothesis.

End-Permian

Burgess, Bowring and Shen (2014) supply the high-precision extinction chronology. Burgess, Muirhead and Bowring (2017) link the extinction interval to the onset of widespread Siberian Traps sill emplacement in volatile-rich sediments. Penn et al. (2018) model temperature-dependent hypoxia and show that warming and oxygen limitation can explain much of the marine geographical and physiological selectivity. Clarkson et al. (2015) provide boron-isotope evidence for ocean acidification during part of the crisis. Li et al. (2026) model an erosion feedback in which vegetation loss increases weathering and phosphorus delivery, delaying peak warming while preconditioning marine anoxia. Stanley (2016) supports the revised marine magnitude estimate.

The body presents these as a connected carbon, heat, oxygen and chemistry crisis. It does not claim that one mechanism explains every marine or terrestrial victim. The duration of the sharpest extinction pulse, the distribution of euxinia, regional acidification and terrestrial magnitude remain less certain than the event's rank and association with Siberian magmatism.

End-Triassic

Blackburn et al. (2013) provide zircon U-Pb dates linking the extinction to the Central Atlantic Magmatic Province. Percival et al. (2017) report mercury enrichments consistent with pulsed volcanism. Capriolo et al. (2020) provide evidence for deep carbon dioxide in CAMP magmas. Trudgill et al. (2025) use boron isotopes in fossil oysters from Lavernock Point, Wales, to reconstruct a pH fall of at least 0.29 units during the main carbon-isotope excursion after the extinction horizon. Their oyster record does not span the initial excursion at the extinction. A carbon-cycle model in the same study suggests that the initial excursion could have produced a larger pH decline if carbon source and release rate were comparable.

The 2025 study is the newest Triassic-Jurassic research used materially in the book. Publication date, sampled interval and modelled interval are kept separate. The local oyster result is not universalised into one global pH curve, and the conditional initial-pulse estimate is labelled as inference. The extinction-to-dinosaur relationship is stated as ecological opportunity for already existing dinosaur lineages, not as the creation or guaranteed triumph of dinosaurs.

End-Cretaceous

Alvarez et al. (1980) is the original iridium-impact paper. Schulte et al. (2010) synthesise the global evidence linking Chicxulub to the K-Pg extinction. Henehan et al. (2019) reconstruct rapid ocean acidification and a prolonged recovery of marine carbon cycling. Hull et al. (2020) compare impact and volcanism across the boundary, while Chiarenza et al. (2020) use climate and habitat models to support impact, rather than Deccan volcanism, as the main driver of non-avian dinosaur extinction.

Field et al. (2018) support the discussion of bird survival in relation to global forest collapse. Vajda, Raine and Hollis (2001) is the source for a post-impact fern spike in New Zealand. The paper by Sheehan and Hansen (1986) is the original source used here for detritus feeding as a buffer to extinction. Freshwater and detrital buffering is retained as a plausible contribution to selective survival, not a settled universal rule. The exact duration and severity of darkness, cooling, fire and acidification vary among models and regions.

The possible sixth

Barnosky et al. (2011), Pimm et al. (2014) and Ceballos et al. (2015) support the claim that recent extinction rates in well-studied groups exceed estimated background rates while realised magnitude remains below a completed Big Five-scale event. Ceballos, Ehrlich and Dirzo (2017) support the emphasis on population and range loss before final species extinction. IPBES (2019) supports the five broad direct drivers: land and sea use change, direct exploitation, climate change, pollution and invasive alien species. Wiens and Saban (2025) and Saban and Wiens (2025) analyse recent species and genus losses across many groups and show why no single modern rate or trend should be universalised: documented extinctions are uneven across taxa and habitats, concentrated strongly on islands, and not uniformly accelerating through the most recent century.

There is genuine disagreement about terminology and extrapolation. Cowie, Bouchet and Fontaine (2022) argue that recorded extinctions are substantially underestimated, especially among invertebrates, and defend the sixth-extinction framing. Wiens and Saban (2025) argue that the 75 per cent criterion has not been quantitatively demonstrated and criticise projections that treat threatened species as though they were already extinct. Their disagreement continued in paired 2025 letters: Cowie, Bouchet and Fontaine argued that undercounting and the direction of change justify speaking of the possible start of a sixth mass extinction, while Wiens and Saban maintained that this does not demonstrate the palaeontological classification. Marshall (2023) proposes an "incipient Anthropocene mass extinction" formulation. The book's phrase "unfinished trajectory" is an editorial synthesis designed to preserve three facts at once: documented rates and pressures are exceptional, the conventional final magnitude has not been reached, and future magnitude is not predetermined.

Misconceptions, practical lenses and further reading

The seven corrections draw on the same sources rather than a separate evidence set. The practical lenses are analytical transfers from extinction research, not empirical claims that geological events predict business, medical or personal outcomes. The limits section explicitly blocks that misuse.

The four Go Deeper editions were checked against publisher and catalogue records: Benton (2023), Erwin (2006), Raup (1991) and Kolbert (2014). Benton is the accessible overview; Erwin is the deep event study; Raup is the major conceptual interpretation; Kolbert is the reported modern account and contrasting present-day emphasis.

Bibliography

Original research and assessments

Alroy, John, et al. "Phanerozoic Trends in the Global Diversity of Marine Invertebrates." Science 321, no. 5885 (2008): 97-100. DOI: 10.1126/science.1156963.

Alvarez, Luis W., Walter Alvarez, Frank Asaro and Helen V. Michel. "Extraterrestrial Cause for the Cretaceous-Tertiary Extinction." Science 208, no. 4448 (1980): 1095-1108. DOI: 10.1126/science.208.4448.1095.

Barnosky, Anthony D., et al. "Has the Earth's Sixth Mass Extinction Already Arrived?" Nature 471, no. 7336 (2011): 51-57. DOI: 10.1038/nature09678.

Blackburn, Terrence J., et al. "Zircon U-Pb Geochronology Links the End-Triassic Extinction with the Central Atlantic Magmatic Province." Science 340, no. 6135 (2013): 941-945. DOI: 10.1126/science.1234204.

Burgess, Seth D., Samuel Bowring and Shu-zhong Shen. "High-Precision Timeline for Earth's Most Severe Extinction." Proceedings of the National Academy of Sciences 111, no. 9 (2014): 3316-3321. DOI: 10.1073/pnas.1317692111.

Burgess, Seth D., James D. Muirhead and Samuel A. Bowring. "Initial Pulse of Siberian Traps Sills as the Trigger of the End-Permian Mass Extinction." Nature Communications 8 (2017): 164. DOI: 10.1038/s41467-017-00083-9.

Capriolo, Manfredo, et al. "Deep CO2 in the End-Triassic Central Atlantic Magmatic Province." Nature Communications 11 (2020): 1670. DOI: 10.1038/s41467-020-15325-6.

Ceballos, Gerardo, et al. "Accelerated Modern Human-Induced Species Losses: Entering the Sixth Mass Extinction." Science Advances 1, no. 5 (2015): e1400253. DOI: 10.1126/sciadv.1400253.

Ceballos, Gerardo, Paul R. Ehrlich and Rodolfo Dirzo. "Biological Annihilation via the Ongoing Sixth Mass Extinction Signaled by Vertebrate Population Losses and Declines." Proceedings of the National Academy of Sciences 114, no. 30 (2017): E6089-E6096. DOI: 10.1073/pnas.1704949114.

Chiarenza, Alfio Alessandro, et al. "Asteroid Impact, Not Volcanism, Caused the End-Cretaceous Dinosaur Extinction." Proceedings of the National Academy of Sciences 117, no. 29 (2020): 17084-17093. DOI: 10.1073/pnas.2006087117.

Clarkson, Matthew O., et al. "Ocean Acidification and the Permo-Triassic Mass Extinction." Science 348, no. 6231 (2015): 229-232. DOI: 10.1126/science.aaa0193.

Fan, Jun-xuan, et al. "A High-Resolution Summary of Cambrian to Early Triassic Marine Invertebrate Biodiversity." Science 367, no. 6475 (2020): 272-277. DOI: 10.1126/science.aax4953.

Field, Daniel J., et al. "Early Evolution of Modern Birds Structured by Global Forest Collapse at the End-Cretaceous Mass Extinction." Current Biology 28, no. 11 (2018): 1825-1831.e2. DOI: 10.1016/j.cub.2018.04.062.

Finnegan, Seth, et al. "The Magnitude and Duration of Late Ordovician-Early Silurian Glaciation." Science 331, no. 6019 (2011): 903-906. DOI: 10.1126/science.1200803.

Finnegan, Seth, et al. "Climate Change and the Selective Signature of the Late Ordovician Mass Extinction." Proceedings of the National Academy of Sciences 109, no. 18 (2012): 6829-6834. DOI: 10.1073/pnas.1117039109.

Henehan, Michael J., et al. "Rapid Ocean Acidification and Protracted Earth System Recovery Followed the End-Cretaceous Chicxulub Impact." Proceedings of the National Academy of Sciences 116, no. 45 (2019): 22500-22504. DOI: 10.1073/pnas.1905989116.

Hull, Pincelli M., et al. "On Impact and Volcanism across the Cretaceous-Paleogene Boundary." Science 367, no. 6475 (2020): 266-272. DOI: 10.1126/science.aay5055.

Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services. Summary for Policymakers of the Global Assessment Report on Biodiversity and Ecosystem Services. Bonn: IPBES Secretariat, 2019. DOI: 10.5281/zenodo.3553458.

Kozik, Nevin P., et al. "Rapid Marine Oxygen Variability: Driver of the Late Ordovician Mass Extinction." Science Advances 8, no. 46 (2022): eabn8345. DOI: 10.1126/sciadv.abn8345.

Li, Shihan, et al. "Erosion-Driven Delayed Warming and Marine Stress Prior to the End-Permian Mass Extinction." Nature Communications 17 (2026): 5456. DOI: 10.1038/s41467-026-74636-2.

Marshall, John E. A., et al. "UV-B Radiation Was the Devonian-Carboniferous Boundary Terrestrial Extinction Kill Mechanism." Science Advances 6, no. 22 (2020): eaba0768. DOI: 10.1126/sciadv.aba0768.

Penn, Justin L., Curtis Deutsch, Jonathan L. Payne and Erik A. Sperling. "Temperature-Dependent Hypoxia Explains Biogeography and Severity of End-Permian Marine Mass Extinction." Science 362, no. 6419 (2018): eaat1327. DOI: 10.1126/science.aat1327.

Percival, Lawrence M. E., et al. "Mercury Evidence for Pulsed Volcanism during the End-Triassic Mass Extinction." Proceedings of the National Academy of Sciences 114, no. 30 (2017): 7929-7934. DOI: 10.1073/pnas.1705378114.

Pimm, Stuart L., et al. "The Biodiversity of Species and Their Rates of Extinction, Distribution, and Protection." Science 344, no. 6187 (2014): 1246752. DOI: 10.1126/science.1246752.

Saban, Kristen E., and John J. Wiens. "Unpacking the Extinction Crisis: Rates, Patterns and Causes of Recent Extinctions in Plants and Animals." Proceedings of the Royal Society B: Biological Sciences 292, no. 2057 (2025): 20251717. DOI: 10.1098/rspb.2025.1717.

Raup, David M., and J. John Sepkoski Jr. "Mass Extinctions in the Marine Fossil Record." Science 215, no. 4539 (1982): 1501-1503. DOI: 10.1126/science.215.4539.1501.

Schulte, Peter, et al. "The Chicxulub Asteroid Impact and Mass Extinction at the Cretaceous-Paleogene Boundary." Science 327, no. 5970 (2010): 1214-1218. DOI: 10.1126/science.1177265.

Sheehan, Peter M., and Thor A. Hansen. "Detritus Feeding as a Buffer to Extinction at the End of the Cretaceous." Geology 14, no. 10 (1986): 868-870. DOI: 10.1130/0091-7613(1986)14%3C868:DFAABT%3E2.0.CO;2.

Signor, Philip W., and Jere H. Lipps. "Sampling Bias, Gradual Extinction Patterns and Catastrophes in the Fossil Record." In Geological Implications of Impacts of Large Asteroids and Comets on the Earth, edited by Leon T. Silver and Peter H. Schultz, 291-296. Geological Society of America Special Paper 190, 1982. DOI: 10.1130/SPE190-p291.

Smart, Matthew S., et al. "The Expansion of Land Plants during the Late Devonian Contributed to the Marine Mass Extinction." Communications Earth & Environment 4 (2023): 449. DOI: 10.1038/s43247-023-01087-8.

Stanley, Steven M. "Estimates of the Magnitudes of Major Marine Mass Extinctions in Earth History." Proceedings of the National Academy of Sciences 113, no. 42 (2016): E6325-E6334. DOI: 10.1073/pnas.1613094113.

Trudgill, Molly, et al. "Pulses of Ocean Acidification at the Triassic-Jurassic Boundary." Nature Communications 16 (2025): 6471. DOI: 10.1038/s41467-025-61344-6.

Vajda, Vivi, J. Ian Raine and Christopher J. Hollis. "Indication of Global Deforestation at the Cretaceous-Tertiary Boundary by New Zealand Fern Spike." Science 294, no. 5547 (2001): 1700-1702. DOI: 10.1126/science.1064706.

Modern syntheses, interpretations and books

Benton, Michael J. Extinctions: How Life Survives, Adapts and Evolves. London: Thames & Hudson, 2023.

Cowie, Robert H., Philippe Bouchet and Benoît Fontaine. "The Sixth Mass Extinction: Fact, Fiction or Speculation?" Biological Reviews 97, no. 2 (2022): 640-663. DOI: 10.1111/brv.12816.

Cowie, Robert H., Philippe Bouchet and Benoît Fontaine. "Denying That We May Be Experiencing the Start of the Sixth Mass Extinction Paves the Way for It to Happen." Trends in Ecology & Evolution 40, no. 8 (2025): 722-723. DOI: 10.1016/j.tree.2025.06.001.

Erwin, Douglas H. Extinction: How Life on Earth Nearly Ended 250 Million Years Ago. Princeton: Princeton University Press, 2006.

Erwin, Douglas H. "Lessons from the Past: Biotic Recoveries from Mass Extinctions." Proceedings of the National Academy of Sciences 98, no. 10 (2001): 5399-5403. DOI: 10.1073/pnas.091092698.

Hull, Pincelli M. "Life in the Aftermath of Mass Extinctions." Current Biology 25, no. 19 (2015): R941-R952. DOI: 10.1016/j.cub.2015.08.053.

International Commission on Stratigraphy. International Chronostratigraphic Chart. Version 2026/06. 2026.

Jablonski, David. "Lessons from the Past: Evolutionary Impacts of Mass Extinctions." Proceedings of the National Academy of Sciences 98, no. 10 (2001): 5393-5398. DOI: 10.1073/pnas.101092598.

Kolbert, Elizabeth. The Sixth Extinction: An Unnatural History. New York: Henry Holt and Company, 2014.

Marshall, Charles R. "Forty Years Later: The Status of the Big Five Mass Extinctions." Cambridge Prisms: Extinction 1 (2023): e5. DOI: 10.1017/ext.2022.4.

Raup, David M. Extinction: Bad Genes or Bad Luck? New York: W. W. Norton, 1991.

Rudwick, Martin J. S. Georges Cuvier, Fossil Bones, and Geological Catastrophes: New Translations and Interpretations of the Primary Texts. Chicago: University of Chicago Press, 1997.

Wiens, John J., and Kristen E. Saban. "Recent Extinctions of Plant and Animal Genera Are Rare, Localized, and Decelerated." PLOS Biology 23, no. 9 (2025): e3003356. DOI: 10.1371/journal.pbio.3003356.

Wiens, John J., and Kristen E. Saban. "Questioning the Sixth Mass Extinction." Trends in Ecology & Evolution 40, no. 4 (2025): 375-384. DOI: 10.1016/j.tree.2025.01.002.

Wiens, John J., and Kristen E. Saban. "Proponents of the Sixth Mass Extinction Admit It Is Unsupported." Trends in Ecology & Evolution 40, no. 8 (2025): 724-725. DOI: 10.1016/j.tree.2025.06.012.