19.08.02 · eco-evo-bio / macroevolution

Mass extinctions: the Big Five, recovery dynamics, and the kill curve

stub3 tiersLean: nonepending prereqs

Anchor (Master): Jablonski, D. — Annu. Rev. Earth Planet. Sci. 23 (1995) 75-102

Intuition Beginner

Earth has experienced five events so catastrophic that they are called mass extinctions -- episodes in which a huge fraction of the planet's species disappeared in a geological instant. These are not the ordinary, constant background rate of species loss; they are sharp spikes in which entire lineages vanish together.

The most devastating was the Permian extinction (~~252 million years ago), which wiped out roughly 90% of all species on Earth. It is the closest life has ever come to total extinction. The most famous is the Cretaceous extinction (~~66 million years ago), which killed off all non-avian dinosaurs when an asteroid slammed into what is now the Yucatan Peninsula. Birds are the surviving descendants of those dinosaurs.

After each mass extinction, the survivors diversified to fill the ecological roles left vacant by the lost species. This recovery is not quick: it takes millions of years for ecosystems to rebuild their complexity and for new evolutionary radiations to produce replacement diversity. The mammalian radiation after the dinosaurs died out is a prime example -- small, shrew-like survivors gave rise to whales, bats, horses, primates, and thousands of other forms.

Some scientists argue that we are currently causing a sixth mass extinction, driven by habitat destruction, climate change, and overexploitation rather than by asteroids or volcanic eruptions. Understanding past mass extinctions helps us evaluate whether that claim is justified and what its consequences might be.

Visual Beginner

The Big Five mass extinctions, showing approximate time, severity, and likely cause:

Event	Time (Ma)	Estimated species loss	Likely cause
End-Ordovician	~444	~86%	Climate change (glaciation and sea-level drop)
Late Devonian	~360	~75%	Climate change, ocean anoxia
End-Permian	~252	~90-96%	Massive volcanism (Siberian Traps)
End-Triassic	~201	~80%	Volcanism (Central Atlantic Magmatic Province)
End-Cretaceous	~66	~76%	Asteroid impact (Chicxulub)

Worked example Beginner

The Cretaceous-Paleogene (K-Pg) extinction (~66 Ma) is the best-understood mass extinction and the clearest case of an impact-driven event. Here is what happened, in sequence:

The impact. An asteroid approximately 10 km in diameter struck Earth at Chicxulub, on the Yucatan coast of Mexico. The impact released energy equivalent to roughly $1 0^{8}$ megatons of TNT -- about a billion times the energy of the Hiroshima bomb.
Immediate effects. The impact generated a fireball, mega-tsunamis, and global firestorms. Ejecta thrown into the atmosphere rained back as tektites (glass spherules) over much of the planet. Airborne dust and soot blocked sunlight for months to years, halting photosynthesis.
Medium-term effects. With photosynthesis shut down, food chains collapsed from the bottom up. Herbivorous dinosaurs starved, followed by the carnivores that ate them. The impact winter produced years of cold and darkness. Organisms dependent on living plant matter (folivorous insects, herbivorous vertebrates) suffered extreme losses.
Who survived and why. Small-bodied generalist mammals survived by sheltering underground and feeding on detritus (dead organic matter), which remained available even when living plants were scarce. Freshwater ecosystems suffered less than terrestrial and marine ones, because their food webs were partly detritus-based. Organisms with low food requirements and broad environmental tolerance made it through.
Who was lost. All non-avian dinosaurs, pterosaurs (flying reptiles), ammonites (coiled cephalopods), mosasaurs and plesiosaurs (large marine reptiles), and approximately 80% of marine plankton species. The loss of ammonites is particularly significant: they had been diverse and abundant for over 300 million years.
Recovery. Within 1-2 million years, surviving mammal lineages began diversifying rapidly into the ecological roles vacated by dinosaurs. By 10 million years post-extinction, mammals occupied the full range of body sizes and trophic positions from mouse-sized insectivores to rhino-sized herbivores.

Check your understanding Beginner

Exercise (easy, multiple choice).

Which mass extinction was the most severe, eliminating approximately 90-96% of all species?

A. End-Cretaceous (K-Pg) B. End-Ordovician C. End-Permian D. End-Triassic

Hint

This event, sometimes called "The Great Dying," occurred approximately 252 million years ago and is associated with the eruption of the Siberian Traps.

Answer

Option C. The end-Permian mass extinction (~252 Ma) eliminated approximately 90-96% of all species -- marine and terrestrial -- making it the most catastrophic event in the history of life. It is associated with massive volcanic activity (the Siberian Traps), which triggered extreme global warming, ocean acidification, and widespread ocean anoxia. Recovery took 5-10 million years, far longer than after any other mass extinction.

Exercise (easy, multiple choice).

After a mass extinction, what typically happens to the surviving species?

A. They remain small and inconspicuous indefinitely B. They diversify to fill ecological roles left vacant by extinct species C. They immediately return to their pre-extinction diversity D. They go extinct shortly afterward due to ecosystem collapse

Hint

When ecological niches are emptied by extinction, what does natural selection favour among the survivors?

Answer

Option B. After a mass extinction, surviving lineages diversify through adaptive radiation to fill the ecological niches vacated by extinct species. This process takes millions of years. The classic example is the mammalian radiation after the K-Pg extinction: small generalist mammals evolved into herbivores, carnivores, marine mammals, flying mammals (bats), and many other forms -- occupying roles previously held by dinosaurs, pterosaurs, and marine reptiles.

Formal definition Intermediate+

Background extinction vs mass extinction

Background extinction is the normal rate at which species go extinct during periods of environmental stability. For marine invertebrates, the background extinction rate is approximately 1 species per million species per year, meaning that in a pool of one million species, roughly one goes extinct each year. Species are constantly turning over: new species arise by speciation and old ones disappear by extinction, at roughly equal rates during normal times.

A mass extinction is defined as a statistically significant elevation of the extinction rate above background, affecting a broad range of taxonomic groups across multiple ecosystems, within a geologically brief interval (typically less than a few million years). Operationally, mass extinctions are identified by sharp drops in the diversity curve (number of genera or families) in the marine fossil record. Raup and Sepkoski (1982) showed that the Big Five mass extinctions stand out as outliers above the continuous distribution of extinction intensities.

The Big Five in detail

The five recognized mass extinctions, ranked by severity:

End-Ordovician (~444 Ma). Approximately 86% of species lost. Two pulses of extinction, both linked to glaciation on Gondwana. Sea level dropped dramatically as water was locked in ice sheets, eliminating shallow marine habitats. The glaciation was followed by rapid warming and sea-level rise. Brachiopods, trilobites, graptolites, and conodonts suffered heavy losses.
Late Devonian (~360 Ma). Approximately 75% of species lost. A prolonged extinction spanning several million years with multiple pulses. Causes include global cooling, ocean anoxia (expansion of oxygen-depleted waters), and possibly volcanism. Tabulate corals and stromatoporoid sponges -- the architects of Devonian reef systems -- were devastated. Placoderm fish (including the apex predator Dunkleosteus) disappeared.
End-Permian (~252 Ma). Approximately 90-96% of species lost. The most severe extinction in Earth's history. Caused by the eruption of the Siberian Traps -- the largest known volcanic event -- which released massive volumes of CO $_{2}$ , methane, and toxic gases over approximately 600,000 years. Consequences included extreme global warming (+8-10 degrees C), ocean acidification, widespread ocean anoxia (the "superanoxic event"), and possible hydrogen sulfide release. Trilobites, which had been in decline, went entirely extinct. Rugose corals, most brachiopod lineages, and approximately 95% of marine species were lost. On land, approximately 70% of tetrapod families disappeared.
End-Triassic (~201 Ma). Approximately 80% of species lost. Associated with the eruption of the Central Atlantic Magmatic Province (CAMP) as the supercontinent Pangaea began to rift apart. Ocean acidification and anoxia eliminated large numbers of marine species, including most ammonites and the conodonts (a group of jawless vertebrates that had survived three previous mass extinctions). On land, many archosaur and amphibian lineages disappeared, clearing ecological space for the dinosaurs that would dominate the Jurassic and Cretaceous.
End-Cretaceous (~66 Ma). Approximately 76% of species lost. Caused by the Chicxulub asteroid impact. Non-avian dinosaurs, pterosaurs, ammonites, mosasaurs, and plesiosaurs went extinct. As discussed in the worked example, the impact produced an impact winter that collapsed photosynthesis-based food webs. Mammals, birds, crocodilians, turtles, and most freshwater organisms survived.

Selectivity of extinction

Mass extinctions are not random with respect to species traits. The selectivity differs among events but some general patterns emerge:

Geographic range. Species with broad geographic ranges survive at higher rates than endemic species with restricted distributions. This is the most consistently documented selectivity filter across multiple mass extinctions (Jablonski, 1986, 1995).
Body size. Large-bodied species are disproportionately lost in sudden catastrophes (e.g., K-Pg) because they require more food and have slower reproductive rates. In more gradual events (e.g., end-Permian), body size is a weaker predictor.
Trophic level. Top predators and specialized consumers are more vulnerable than primary producers and detritivores, because perturbations propagate up food chains.
Physiological tolerance. Eurytopic generalists (species tolerant of a wide range of temperatures, salinities, or oxygen levels) survive at higher rates than stenotopic specialists.

The kill curve (Raup, 1991)

David Raup (1991) introduced the kill curve to describe the relationship between the intensity of a disturbance and the resulting fraction of species lost. The curve plots extinction fraction (proportion of species eliminated) against some measure of causal intensity (energy of impact, volume of volcanic eruption, magnitude of climate change). Key properties:

For low-intensity disturbances, extinction is near background levels.
As intensity increases, extinction rises slowly at first, then steeply.
At very high intensities, the curve approaches 100% extinction.

The kill curve formalizes the intuition that there is a threshold of disturbance intensity above which extinction accelerates catastrophically, and that the relationship is nonlinear: doubling the causal magnitude more than doubles the extinction fraction.

Key theorem with proof Intermediate+

Theorem (Raup's kill curve -- extinction intensity and causal magnitude). Let $E$ be the fraction of species going extinct in a disturbance event, and let $I$ be a measure of the causal intensity of that event (proportional to energy released, area affected, or duration). Under the assumption that each species has an independent survival probability that decreases with $I$ , the kill curve is a sigmoid function of $I$ , with a region of accelerating extinction (inflection point) at intermediate intensities.

Proof sketch. Model each species $i$ as having a survival probability $p_{i} (I)$ that depends on the causal intensity $I$ . Assume $p_{i} (I)$ is a decreasing function of $I$ for all species, and that species vary in their susceptibility. The expected extinction fraction is:

E (I) = 1 - \frac{1}{S} i = 1 \sum S p_{i} (I),

where $S$ is the total number of species. If the $p_{i} (I)$ are modeled as sigmoid survival functions -- for example, $p_{i} (I) = 1/ (1 + e^{a_{i} (I - I_{i}^{*})})$ where $I_{i}^{*}$ is the threshold intensity that kills species $i$ and $a_{i}$ controls how sharp the threshold is -- then the aggregate kill curve is the average of many sigmoids with different thresholds.

When the distribution of thresholds $I_{i}^{*}$ across species is unimodal (most species have moderate vulnerability, few are extremely resistant or extremely susceptible), the average of these sigmoids is itself approximately sigmoid. The extinction fraction $E (I)$ is near zero for $I$ well below the median threshold, rises steeply through the region where most species' thresholds are being crossed, and asymptotes toward 1 for $I$ well above the maximum threshold.

The inflection point occurs near the median of the threshold distribution -- the intensity at which half of all species have crossed their lethal threshold. The slope at the inflection point is determined by the variance in susceptibility: more uniform susceptibility produces a steeper kill curve (sharp threshold), while highly variable susceptibility produces a more gradual rise. This matches the empirical observation that some mass extinctions (K-Pg, end-Permian) show abrupt extinction pulses, while others (Late Devonian) show more prolonged, stepped extinction patterns. $□$

The kill curve has a direct empirical implication: if extinction intensity is plotted against estimated causal magnitude for known extinction events, the data should trace out a sigmoid. Raup (1991) used the marine fossil record to argue that the data are consistent with this prediction, though the incompleteness of the fossil record makes precise quantitative fitting difficult.

Exercise 1

Exercise 1 (medium, short answer).

Explain the difference between background extinction and mass extinction. How did Raup and Sepkoski (1982) demonstrate that the Big Five are statistically distinct from background extinction rates?

Hint

Background extinction is a continuous, low-level process. Mass extinctions are spikes. How would you test whether the spikes are statistically significant outliers rather than the tail of a single distribution?

Answer

Background extinction is the normal rate of species turnover during periods of environmental stability -- approximately 1 species per million species per year for marine invertebrates. Mass extinction is a statistically significant elevation of the extinction rate above background, affecting multiple taxonomic groups across multiple ecosystems within a geologically brief interval.

Raup and Sepkoski (1982) compiled extinction rates for marine families across all geological stages of the Phanerozoic. They found that the distribution of extinction intensities has a long right tail, and that the Big Five stand out as distinct peaks that are unlikely to be drawn from the same distribution as the remaining data points. They used statistical tests to show that the five largest extinction events are outliers above the expected tail of the distribution, supporting the hypothesis that mass extinctions are caused by qualitatively different mechanisms (catastrophic impacts, extreme volcanism) rather than being merely the extreme end of normal extinction processes.

Key results Intermediate+

Result 1 (Recovery timescale scales with severity). The time required for diversity to recover to pre-extinction levels is proportional to the severity of the extinction. After the end-Permian extinction (~~96% species loss), recovery took 5-10 million years. After the end-Cretaceous (~~76% species loss), recovery took approximately 2-5 million years. After the end-Ordovician (~86% species loss, but taxonomically concentrated), recovery was faster: approximately 1-3 million years. The Permian recovery was exceptionally slow because the extinction eliminated entire ecosystem architectures, including reef systems and the ecological networks they supported, requiring evolutionary reconstruction from a more depauperate starting point.

Result 2 (Lazarus taxa and the Signor-Lipps effect). Lazarus taxa are species that disappear from the fossil record during a mass extinction but reappear later, having survived in refugia (small, isolated populations) that were not fossilized. Their apparent "resurrection" is an artifact of the incompleteness of the fossil record. The Signor-Lipps effect (Signor and Lipps, 1982) describes the reverse problem: because fossil preservation is incomplete, the last appearance of a species in the record is always earlier than its actual time of extinction. This causes mass extinctions to appear gradual in the fossil record even when they were geologically instantaneous -- species that went extinct simultaneously appear to have disappeared over a range of time because some last occurrences are missing. Correcting for the Signor-Lipps effect is essential for estimating the true duration of extinction events.

Result 3 (Disaster taxa and early recovery communities). In the immediate aftermath of mass extinctions, ecosystems are dominated by disaster taxa -- opportunistic, generalist species that thrive in disturbed, low-competition environments. After the end-Permian extinction, the bivalve Claraia and the brachiopod Lingula formed near-monospecific communities across vast areas of the seafloor. These early recovery communities are low-diversity, cosmopolitan, and ecologically simple. Over time, they are replaced by more diverse communities as surviving clades diversify and new ecological interactions develop.

Result 4 (Evolutionary radiations of survivors). The pattern of post-extinction recovery is not a simple restoration of what was lost. Instead, new clades radiate to fill vacated niches, often producing ecological structures that differ markedly from pre-extinction ecosystems. After the end-Cretaceous extinction, mammals radiated into niches previously occupied by dinosaurs. After the end-Triassic extinction, dinosaurs radiated into niches previously occupied by crurotarsan archosaurs. After the end-Permian extinction, archosaurs (including the lineage leading to dinosaurs and crocodilians) radiated into terrestrial predator niches previously occupied by therapsid carnivores. Each mass extinction reshuffled the evolutionary deck, and the new communities that emerged reflected the contingencies of which lineages survived and what their evolutionary potential permitted.

Exercise 2

Exercise 2 (medium, short answer).

What are Lazarus taxa? Explain why they appear to "return from the dead" in the fossil record and what they reveal about the completeness of the fossil record.

Hint

The fossil record is incomplete. If a species survives in low numbers in a small area that is not being fossilized, what would its fossil record look like before, during, and after the mass extinction?

Answer

Lazarus taxa are species that disappear from the fossil record during or before a mass extinction and then reappear millions of years later, as if they had risen from the dead (named after the biblical Lazarus). They do not actually go extinct and return; rather, they survive in low-abundance populations in small geographic areas (refugia) where fossil preservation is unlikely. During the extinction interval, the species' geographic range and population size are so reduced that the probability of fossilization drops to near zero. When the environment recovers and the species expands its range again, it reappears in the fossil record.

Lazarus taxa reveal that the fossil record substantially underestimates survivorship during mass extinctions: more species survive than the fossil record shows. The abundance of Lazarus taxa varies among mass extinctions -- the end-Permian has many Lazarus taxa, indicating that while many lineages survived, they were reduced to such low abundances that they were invisible in the fossil record for millions of years.

Exercise 3

Exercise 3 (medium, short answer).

Explain the Signor-Lipps effect. How does it create the illusion that a sudden mass extinction was gradual?

Hint

Fossilization is rare. The last fossil of a species found is not necessarily the last living individual of that species. What happens to the "last appearance datum" when preservation is incomplete?

Answer

The Signor-Lipps effect (Signor and Lipps, 1982) is a sampling artifact that causes geologically instantaneous mass extinctions to appear gradual in the fossil record. Because fossil preservation is incomplete (only a tiny fraction of individuals that ever lived are preserved), the last fossil occurrence of a species is almost always earlier than its actual time of extinction. The last appearance is pulled backward in time by a random amount that depends on the species' abundance, geographic range, and preservation potential.

When many species go extinct simultaneously (as in a sudden asteroid impact), their last fossil appearances are scattered backward in time by varying amounts, creating a pattern of stepwise or gradual extinction that is entirely an artifact of incomplete sampling. Rare species with low preservation potential have their last appearances pulled further back than common species, making them appear to go extinct earlier. Correcting for this effect requires statistical methods (such as confidence intervals on last appearances) to estimate the true extinction horizon.

Advanced treatment Master

Boundary events and geochemical signatures

Mass extinction boundaries are identified by distinctive geochemical and geological markers that record the catastrophic conditions:

Iridium anomaly. The K-Pg boundary clay worldwide contains iridium concentrations 10-100 times above background levels. Iridium is rare in Earth's crust but abundant in asteroids, providing the original evidence for the Alvarez impact hypothesis (Alvarez et al., 1980). The anomaly is present at over 100 K-Pg boundary sites globally, confirming a worldwide distribution of impact ejecta.
Shocked quartz. Quartz grains with planar deformation features (microscopic lamellae produced by extreme pressure) are found at the K-Pg boundary. These form only under the high-pressure conditions of meteorite impacts or nuclear explosions, not during normal geological processes.
Spherules and tektites. Glassy spherules (molten rock ejected from the impact site that solidified in flight) form a distinct layer at the K-Pg boundary. Their composition matches the target rock at Chicxulub, confirming the impact location.
Carbon isotope excursions. Sharp negative shifts in $δ^{13}$ C at extinction boundaries indicate massive disruption of the carbon cycle. The end-Permian boundary shows the largest carbon isotope excursion in the Phanerozoic, consistent with the release of enormous volumes of isotopically light carbon from volcanic CO $_{2}$ and/or methane hydrate destabilization.

End-Permian mechanisms in depth

The end-Permian extinction involved multiple, cascading kill mechanisms triggered by the eruption of the Siberian Traps:

Direct volcanic effects. Massive CO $_{2}$ and SO $_{2}$ emissions caused immediate acid rain, ozone depletion, and respiratory irritation from halogen compounds.
Runaway greenhouse warming. Atmospheric CO $_{2}$ from the Siberian Traps and from thermal decomposition of organic-rich sediments (the Tunguska Basin coal beds) raised global temperatures by 8-10 degrees C. This warming triggered further carbon release from methane hydrate deposits on continental shelves -- a positive feedback that amplified the initial warming.
Ocean anoxia and the superanoxic event. Warm water holds less dissolved oxygen than cold water. Combined with reduced ocean circulation (warm polar waters weaken the thermohaline circulation that ventilates the deep ocean), this produced widespread oxygen depletion. The superanoxic event eliminated most marine habitats below the photic zone.
Hydrogen sulfide poisoning. Under anoxic conditions, sulfate-reducing bacteria produce hydrogen sulfide (H $_{2}$ S). Modeling by Kump, Pavlov, and Arthur (2005) suggests that H $_{2}$ S accumulated to toxic levels in the ocean and atmosphere, directly poisoning marine and terrestrial organisms and further damaging the ozone layer.
Ocean acidification. Elevated atmospheric CO $_{2}$ dissolved in seawater, lowering pH and reducing the availability of carbonate ions needed by calcifying organisms. Organisms with calcium carbonate shells and skeletons (corals, brachiopods, echinoderms, foraminifera) suffered disproportionate losses.
The Smithian-Spathian crisis. Within the Early Triassic recovery (~2.5 Myr after the main extinction), a second extinction pulse during the Smithian-Spathian boundary eliminated many groups that had survived the initial Permian-Triassic event. This secondary crisis may have been caused by extreme temperature fluctuations during the recovery interval, suggesting that the Permian extinction was not a single event but a prolonged interval of environmental instability.

Jablonski's selectivity rules

Jablonski (1986, 1995) analyzed the selectivity of the end-Cretaceous extinction using well-preserved marine invertebrate fossil records from the Gulf and Atlantic Coastal Plains. His key findings established rules for survival during mass extinctions:

Geographic range at the species level is the strongest predictor of survival. Species with broad geographic ranges survived at significantly higher rates than endemic species, regardless of their ecological characteristics. This selectivity operated at the species level, not the clade level: within the same clade, widespread species survived and restricted species did not.
Clade-level traits matter during mass extinctions but not during background times. During background extinction, species-level traits (body size, trophic level, local abundance) are the best predictors of extinction risk. During mass extinctions, clade-level traits (geographic range of the entire clade, larval developmental mode) override species-level predictors. This switch in the hierarchy of selectivity is one of the strongest arguments that mass extinctions are governed by different rules than background extinction.
Larval dispersal mode. Marine invertebrate clades with planktotrophic larvae (long-lived, feeding larvae that disperse widely) survived at higher rates than clades with nonplanktotrophic larvae (short-lived, non-feeding larvae with limited dispersal), because planktotrophy is correlated with broad geographic range.

Turnover-pulse and coordinated stasis

The turnover-pulse hypothesis (Vrba, 1985) proposes that climate change causes synchronized extinction and speciation across multiple lineages. Periods of climatic stability produce stasis; abrupt climate shifts produce coordinated turnover -- a "pulse" of extinction and speciation that affects many clades simultaneously. The hypothesis predicts that origination and extinction rates should be clustered in time rather than randomly distributed.

Coordinated stasis (Brett and Baird, 1995) describes the observation that in some fossil sequences (particularly the Devonian of the Appalachian Basin), species assemblages remain stable in composition for millions of years and then change abruptly during brief extinction/recovery intervals. This pattern suggests that ecological associations are maintained over long periods and disrupted simultaneously during environmental perturbations, rather than gradually tracking environmental change. Both hypotheses remain debated; critics argue that the apparent synchrony may be an artifact of stratigraphic gaps and incomplete sampling.

Phylogenetic vs stratigraphic diversity estimates

Two complementary approaches estimate past biodiversity: stratigraphic (counting taxa present in the fossil record at each time interval) and phylogenetic (using molecular phylogenies of living species, calibrated with fossils, to infer past diversification rates). Discrepancies between these approaches are informative:

Molecular phylogenies of living birds and mammals suggest that many modern lineages originated before the K-Pg extinction, implying that these groups survived the extinction (perhaps at low diversity) and radiated afterward. The fossil record for these groups in the Cretaceous is sparse, possibly because early members were small, rare, and geographically restricted -- poor candidates for fossilization.
Stratigraphic range data provide direct evidence of when taxa existed but are subject to sampling bias (the Signor-Lipps effect, Lazarus taxa, variable preservation potential). Phylogenetic estimates are immune to these biases but are sensitive to model assumptions about diversification rates and the molecular clock.

The sixth extinction

The hypothesis that Earth is experiencing a sixth mass extinction driven by human activity is supported by several lines of evidence:

Barnosky et al. (2011) compared current extinction rates to background rates using IUCN Red List data. They found that current extinction rates for mammals, birds, and amphibians are 3-80 times higher than the background rate of approximately 1 extinction per million species per year, depending on the taxonomic group and the time window used. If all species currently classified as "threatened" (vulnerable, endangered, or critically endangered) go extinct within the next century, the extinction rate would reach levels comparable to the Big Five.
Ceballos et al. (2015) documented that the vertebrate species lost in the last century would have taken approximately 800-10,000 years to go extinct at the background rate, representing an acceleration of approximately two orders of magnitude.
Anthropogenic drivers include: habitat loss and fragmentation (the single largest driver, affecting approximately 85% of threatened species); climate change (shifting temperature and precipitation patterns faster than species can adapt or migrate); overexploitation (hunting, fishing, harvesting); invasive species and disease (introduced predators, competitors, and pathogens); and pollution (chemical contamination, plastic, nutrient loading).

The sixth extinction differs from the Big Five in that it is caused by a single species (Homo sapiens) rather than by geological or extraterrestrial events, it is orders of magnitude faster in terms of rate, and it is theoretically preventable. Conservation biology 19.14.01 addresses the strategies and trade-offs involved in mitigating this extinction event.

Exercise 4

Exercise 4 (hard, analysis).

Compare the end-Permian and end-Cretaceous mass extinctions in terms of their causes, selectivity patterns, and recovery dynamics. What do the differences reveal about how the nature of the killing mechanism shapes the evolutionary aftermath?

Hint

End-Permian: prolonged volcanism, gradual environmental degradation, ocean anoxia. End-Cretaceous: sudden impact, impact winter, food web collapse. Different causes should select for different survivors and produce different recovery trajectories.

Answer

Causes. The end-Permian was driven by prolonged (hundreds of thousands of years) massive volcanism (Siberian Traps) producing cascading environmental effects: extreme warming, ocean acidification, anoxia, and possible hydrogen sulfide poisoning. The end-Cretaceous was driven by a sudden (hours to years) asteroid impact producing impact winter, firestorms, and food web collapse. The key contrast is duration: the Permian event was a sustained, multi-millennial crisis; the Cretaceous event was a geologically instantaneous catastrophe.

Selectivity. The end-Permian selectively eliminated calcifiers (vulnerable to ocean acidification), organisms with narrow thermal tolerances (vulnerable to extreme warming), and organisms dependent on well-oxygenated marine habitats. Body size was a weaker predictor than physiological tolerance and habitat preference. The end-Cretaceous selectively eliminated large-bodied organisms at high trophic levels (vulnerable to food web collapse during impact winter), while small-bodied detritivores and freshwater organisms survived at higher rates. The Permian event filtered by physiology and habitat; the Cretaceous event filtered by body size and trophic position.

Recovery. The end-Permian recovery took 5-10 million years -- an order of magnitude longer than the K-Pg recovery (~2-5 million years). The Permian's prolonged recovery reflects the near-total destruction of marine ecosystems (reefs, the benthic invertebrate fauna, the plankton) and the need for evolutionary reconstruction from a highly depauperate starting point. Early Triassic ecosystems were dominated by disaster taxa and showed recurrent instability (the Smithian-Spathian crisis). The K-Pg recovery was faster because many marine clades survived intact and terrestrial ecosystems retained more structural complexity.

The differences reveal that the nature of the killing mechanism determines: (a) which lineages survive (physiological filtering vs trophic filtering), (b) how much ecological structure is destroyed (near-total vs partial), and (c) how long recovery takes (prolonged volcanism allows the killing mechanism to persist, delaying recovery; a sudden impact removes the stressor quickly, allowing recovery to begin immediately). The evolutionary aftermath is shaped not just by the severity of the extinction but by its mechanism.

Exercise 5

Exercise 5 (hard, analysis).

Critically evaluate the claim that Earth is experiencing a "sixth mass extinction." What evidence supports this claim? What are the main sources of uncertainty? How does the current extinction event differ from the Big Five?

Hint

Compare current rates to background rates (Barnosky 2011, Ceballos 2015). Consider the taxonomic groups for which we have the best data. What makes the current event different in mechanism, speed, and potential reversibility?

Answer

Evidence supporting the claim. Current extinction rates for well-studied vertebrate groups (mammals, birds, amphibians) are 3-80 times higher than the background rate, depending on the time window and assumptions. Approximately 1-2% of all evaluated species have gone extinct in the last 500 years -- a rate that, if extrapolated, would reach Big Five levels within centuries rather than millions of years. Population-level data show even more dramatic decline: Ceballos et al. (2017) documented that approximately 32% of vertebrate species are declining in population size and geographic range, indicating that the extinction rate is accelerating.

Sources of uncertainty. (a) We have assessed the conservation status of only a small fraction of all species (roughly 5% of estimated total species). Many invertebrates, fungi, and microorganisms have never been evaluated, and their extinction rates are unknown. (b) The background extinction rate itself is uncertain, estimated from the fossil record with wide confidence intervals. Different estimates of background rate yield different multiples of acceleration. (c) The definition of "mass extinction" is partly arbitrary: is a 3-80x increase above background enough, or must it reach the 75%+ species loss of the Big Five? (d) Short time windows are noisy: a cluster of extinctions in one decade may not represent a sustained trend.

How the sixth differs from the Big Five. (a) Cause: a single species (Homo sapiens) rather than geological or extraterrestrial events. (b) Speed: current rates are estimated at 100-1,000 times background, potentially faster than any of the Big Five. (c) Selectivity: anthropogenic extinction preferentially eliminates large-bodied, slow-reproducing, habitat-specialist species -- a pattern that overlaps with but differs from the selectivity of past mass extinctions. (d) Reversibility: in principle, human behavior can be changed to slow or halt the current extinction, unlike volcanic eruptions or asteroid impacts. (e) Scope: the current event is affecting terrestrial ecosystems more severely than marine ones (the reverse of the Big Five), though marine ecosystems are increasingly impacted by ocean acidification, overfishing, and warming.

The most defensible position is that the sixth extinction is underway but has not yet reached the species-loss threshold of the Big Five. If current trajectories continue, that threshold could be crossed within the next few centuries.

Connections Master

Macroevolution 19.08.01. Mass extinctions are a central topic within macroevolution, representing the most extreme perturbations to the tree of life. Unit 19.08.01 introduced the Big Five in overview; this unit develops the detailed mechanisms, selectivity rules, and recovery dynamics. The macroevolutionary concepts of species selection and differential diversification are directly applicable to understanding why certain clades survive mass extinctions and others do not.
Conservation biology 19.14.01. The sixth mass extinction debate directly connects to conservation biology. Understanding extinction selectivity from past mass extinctions informs predictions about which taxa are most vulnerable to current anthropogenic threats. The kill curve framework provides a quantitative basis for evaluating how different magnitudes of habitat loss, climate change, and other stressors translate into extinction risk.
Phylogenetics 19.07.01. Molecular phylogenetic estimates of diversification timing complement the fossil record in understanding mass extinction boundaries and recovery dynamics. Phylogenetic diversity metrics (the total branch length of the tree of life) quantify how much evolutionary history would be lost under different extinction scenarios, providing a measure of the evolutionary cost of the sixth extinction.
Population ecology 19.09.01. The population dynamics of extinction (minimum viable population, demographic stochasticity, Allee effects) connect the macroevolutionary patterns of mass extinction to the population-level processes that determine whether a species survives a perturbation. Small population size during mass extinction bottlenecks amplifies the role of genetic drift and inbreeding 19.04.01, potentially constraining recovery.
Biogeography 19.12.01. Geographic range -- the strongest predictor of mass extinction survival -- is a biogeographic property. Mass extinctions reorganize biogeographic patterns: after the end-Permian, cosmopolitan (widely distributed) taxa replaced endemic ones, and after the K-Pg, mammalian clades redistributed across continents as land connections shifted.
Ecosystem ecology 19.11.01. Mass extinctions fundamentally restructure ecosystem-level processes (energy flow, nutrient cycling, trophic structure). The transition from dinosaur-dominated to mammal-dominated terrestrial ecosystems after the K-Pg altered grazing patterns, seed dispersal, predator-prey dynamics, and nutrient cycling in ways that persisted for millions of years.

Historical & philosophical context Master

The recognition of mass extinction as a phenomenon distinct from background extinction has a complex intellectual history, intertwined with debates about uniformitarianism, catastrophism, and the nature of the geological record.

From uniformitarianism to catastrophism and back. Charles Lyell's uniformitarianism (1830s) held that geological processes operate at constant, gradual rates, and that the geological record should be interpreted as the product of forces observable in the present. Darwin extended this reasoning to the fossil record, interpreting apparent extinctions as artifacts of incomplete preservation. The discovery that mass extinctions are real, discrete events -- not merely the tails of continuous distributions -- required a departure from strict uniformitarianism. The Alvarez hypothesis (1980) represented a particularly dramatic departure: the proposal that an asteroid impact caused the K-Pg extinction reintroduced catastrophism (the idea that rare, high-magnitude events shape Earth history) as a legitimate scientific framework, but with the crucial difference that catastrophes were now to be evaluated empirically rather than assumed a priori.

The Alvarez hypothesis and its reception. Luis and Walter Alvarez, with Frank Asaro and Helen Michel, discovered the iridium anomaly at the K-Pg boundary in Gubbio, Italy, and proposed that it resulted from the impact of a large asteroid. The hypothesis was initially met with skepticism from geologists committed to gradualist explanations. The subsequent discovery of shocked quartz (Bohor et al., 1984), the identification of the Chicxulub crater (Hildebrand et al., 1991), and the global distribution of the iridium anomaly eventually produced a scientific consensus that the K-Pg extinction was caused by an impact. This episode demonstrated that catastrophic events are part of Earth's history and that the fossil record can preserve their signature.

Raup and Sepkoski and the statistical identification of mass extinctions. David Raup and Jack Sepkoski's quantitative analysis of marine family extinction rates (1982) provided the statistical foundation for distinguishing mass extinctions from background. Their finding that the Big Five were statistical outliers -- and their controversial claim of a 26-million-year periodicity in extinction peaks -- catalyzed decades of research into the causes and consequences of mass extinctions. The periodicity claim (which inspired the now-rejected hypothesis of a distant stellar companion, "Nemesis," periodically perturbing cometary orbits) was an instructive example of how statistical pattern-seeking in the fossil record can generate both productive hypotheses and speculative excess.

Jablonski and the rules of survival. David Jablonski's work on K-Pg extinction selectivity (1986, 1995) transformed the study of mass extinctions from descriptive paleontology into a predictive, quantitative science. By demonstrating that geographic range is the strongest predictor of survival during mass extinctions but not during background times, Jablonski showed that mass extinctions are governed by different rules than normal evolutionary turnover. This finding has direct implications for conservation biology: the traits that make species vulnerable to anthropogenic extinction (small range, specialized ecology) are partly the same traits that made species vulnerable to past mass extinctions.

The kill curve and the quantification of extinction. Raup's kill curve (1991) represented an attempt to formalize the relationship between disturbance intensity and extinction fraction. While the kill curve is a conceptual model rather than a precisely parameterized function (the fossil record provides too few data points to fit the curve rigorously), it provides a framework for comparing extinction events of different magnitudes and for predicting the consequences of future perturbations. The kill curve's sigmoid shape implies that there are thresholds of disturbance intensity beyond which extinction accelerates nonlinearly -- a finding with obvious relevance to the current biodiversity crisis.

The sixth extinction debate. The proposal that human activity is causing a sixth mass extinction has been advanced by paleontologists (Barnosky et al., 2011), ecologists (Ceballos et al., 2015), and science communicators (Elizabeth Kolbert's The Sixth Extinction, 2014). The debate is partly empirical (are current rates high enough to qualify?) and partly normative (should we define mass extinction by rate, by total species lost, by ecological impact, or by some combination?). The philosophical question of whether an extinction caused by one species belongs in the same category as extinctions caused by geological or astronomical events is itself debated. What is not debated is that current extinction rates are substantially elevated above background and that the loss of biodiversity has consequences for ecosystem function, human welfare, and the long-term evolutionary trajectory of life on Earth.

Contingency and the evolutionary significance of mass extinctions. Mass extinctions are the strongest argument for contingency in the history of life. Stephen Jay Gould argued that if the Chicxulub asteroid had missed Earth, non-avian dinosaurs might still dominate terrestrial ecosystems, and mammals might never have had the opportunity to radiate. Mass extinctions are not predictable from the competitive dynamics of species; they are external perturbations that override the normal processes of natural selection and ecological interaction. The evolutionary history of life is thus shaped by the interaction of deterministic processes (adaptation, competition, coevolution) and stochastic events (impacts, eruptions, climate perturbations) that cannot be predicted from biological first principles.

Bibliography Master

Alvarez, L. W., Alvarez, W., Asaro, F. & Michel, H. V. "Extraterrestrial cause for the Cretaceous-Tertiary extinction." Science 208 (1980) 1095-1108.
Barnosky, A. D. et al. "Has the Earth's sixth mass extinction already arrived?" Nature 471 (2011) 51-57.
Brett, C. E. & Baird, G. C. "Coordinated stasis and evolutionary ecology of Silurian to Middle Devonian faunas in the Appalachian Basin." In Erwin, D. H. & Anstey, R. L. (eds.), New Approaches to Speciation in the Fossil Record (Columbia University Press, 1995) 285-315.
Ceballos, G. et al. "Accelerated modern human-induced species losses: entering the sixth mass extinction." Sci. Adv. 1 (2015) e1400253.
Erwin, D. H. Extinction: How Life on Earth Nearly Ended 250 Million Years Ago (Princeton University Press, 2006).
Futuyma, D. J. & Kirkpatrick, M. Evolution, 4th ed. (Sinauer, 2017).
Hartl, D. L. & Clark, A. G. Principles of Population Genetics, 4th ed. (Sinauer, 2007).
Jablonski, D. "Background and mass extinctions: the alternation of macroevolutionary regimes." Science 231 (1986) 129-133.
Jablonski, D. "Extinctions in the fossil record." Annu. Rev. Earth Planet. Sci. 23 (1995) 75-102.
Kump, L. R., Pavlov, A. & Arthur, M. A. "Massive release of hydrogen sulfide to the surface ocean and atmosphere during intervals of oceanic anoxia." Geology 33 (2005) 397-400.
Raup, D. M. "The kill curve." Science 251 (1991) 1537.
Raup, D. M. & Sepkoski, J. J. "Mass extinctions in the marine fossil record." Science 215 (1982) 1501-1503.
Signor, P. W. & Lipps, J. H. "Sampling bias, gradual extinction patterns and catastrophes in the fossil record." Geol. Soc. Am. Spec. Pap. 190 (1982) 291-296.
Vrba, E. S. "Environment and evolution: alternative causes of the temporal distribution of evolutionary events." S. Afr. J. Sci. 81 (1985) 229-236.