Module 7: Paleobiogeography & Extinctions

John J. Sepkoski Jr. (Sepkoski 1981 Paleobiology 7, 36–53; Sepkoski 1993 Paleobiology 19, 43–51; Sepkoski 2002 Bulletin of American Paleontology 363) spent his career compiling the first-appearance and last-appearance dates of every marine animal genus known from the fossil record. The resulting diversity curve — plotted as number of genera versus geologic age over the 541 Myr of the Phanerozoic — is the single most influential graph in paleobiology.

The curve shows a tripartite structure dominated by three evolutionary faunas:

• Cambrian fauna: dominated by trilobites, inarticulate brachiopods, hyoliths, and problematic phyla; peaks at ~510 Ma and declines through the Ordovician.
• Paleozoic fauna: dominated by articulate brachiopods, rugose and tabulate corals, crinoids, cephalopods, and ostracods; peaks at ~440 Ma and declines through the Devonian and especially the Permo-Triassic.
• Modern fauna: dominated by bivalves, gastropods, scleractinian corals, echinoids, and teleost fish; rises from ~250 Ma, the dominant fauna of the Mesozoic and Cenozoic.

Alroy et al. (2008 Science) used subsampling methods on the Paleobiology Database to argue that much of the apparent Cenozoic diversification in the Sepkoski curve reflects the “pull of the Recent” — better preservation and sampling in younger strata. Corrected curves show a flatter post-Paleozoic plateau, though the overall tripartite fauna structure and the Big Five extinction troughs are robust.

Raup and Sepkoski (1982 Science 215, 1501–1503) identified five extinction events in the Phanerozoic whose intensity falls well outside the background distribution of per-capita extinction rates. Raup (1991) formalised this as the “kill curve”: the relationship between extinction intensity and recurrence time, with the Big Five occurring at intensities above the 95th-percentile background threshold.

• Ordovician–Silurian (443 Ma): ~57% of marine genera, ~85% of species. Caused by late-Ordovician glaciation (Gondwana over South Pole), sea-level drop, and subsequent warming / anoxia during deglaciation. Hit brachiopods, trilobites, conodonts severely.
• Late Devonian (375–360 Ma): ~35% of marine genera across a prolonged Frasnian-Famennian and Hangenberg pulse. Reef ecosystems collapsed; tropical coral-stromatoporoid reefs never fully recovered. Linked to marine anoxia, cooling, and the expansion of land plants (McGhee 1996).
• Permian–Triassic (252 Ma), The Great Dying: ~83% of marine genera, ~96% of marine species, ~70% of terrestrial vertebrate families. The most severe biotic crisis in Earth history. Caused by Siberian Traps flood-basalt volcanism (Payne & Clapham 2012 Annual Review of Earth and Planetary Sciences): massive CO2 and SO2 release, ocean acidification, warming, and widespread anoxia. Complete faunal turnover from Paleozoic to Modern fauna.
• Triassic–Jurassic (201 Ma): ~47% of marine genera. Associated with the Central Atlantic Magmatic Province (CAMP) eruptions accompanying the Pangean breakup. Conodonts went extinct entirely; dinosaurs rose to dominance.
• Cretaceous–Paleogene (66 Ma): ~40% of marine genera, 75% of species, all non-avian dinosaurs, ammonites, and mosasaurs. Caused by the Chicxulub impact (Alvarez et al. 1980 Science 208, 1095–1108 — the famous iridium anomaly; Hildebrand et al. 1991 located the 180 km crater in the Yucatan). Deccan Traps volcanism may have primed the biosphere (Schoene et al. 2019 Science).

The Cambrian Radiation or “explosion” is the geologically abrupt appearance of nearly all modern animal phyla in the fossil record over a window of ~25 Myr, beginning at the base of the Cambrian (~538.8 Ma; International Commission on Stratigraphy 2022). The most dramatic documentation comes from three Burgess-Shale-type Lagerstätten: the Chengjiang Biota of Yunnan (~518 Ma; Hou et al. 2017), the Sirius Passet Lagerstätte of Greenland (~518 Ma), and the Burgess Shale of British Columbia (~508 Ma; Walcott 1911, re-interpreted by Whittington, Briggs, Conway Morris 1970s).

Marshall (2006 Annual Review of Earth and Planetary Sciences) reviewed competing explanations: (1) the oxygen threshold (Canfield et al. 2007 Science) — atmospheric/ocean O2 rose above ~10% PAL at the Ediacaran–Cambrian boundary, enabling large mobile predators and metabolically expensive bauplans; (2) the ecological revolution hypothesis — the origin of predation and filter feeding triggered evolutionary arms races and niche specialisation; (3) the genetic hypothesis — Hox-gene regulatory networks reached a threshold of complexity; (4) the geochemical “nutrient flood” — weathering of Ediacaran supercontinent breakup released phosphorus; (5) the “Snowball Earth release” — the end of the 717–635 Ma Cryogenian glaciations freed tropical oceans for rapid radiation.

Gould (Wonderful Life, 1989) emphasised the morphological disparity of Cambrian organisms — the “weird wonders” of the Burgess Shale — arguing that the Cambrian produced many more disparate bauplans than survive today, with “historical contingency” determining which lineages persisted to seed modern phyla. Conway Morris (The Crucible of Creation, 1998) argued the opposite: that the Cambrian disparity is more apparent than real and that much of it represents convergent evolution within standard bauplans.

The end-Permian extinction is the closest Earth life has come to being extinguished entirely. Over a geological interval of <100 kyr at the Permian–Triassic boundary (dated to 251.902 ± 0.024 Ma by CA-ID-TIMS zircon geochronology; Burgess et al. 2014), approximately 83% of marine animal genera and 96% of marine species disappeared. Reef ecosystems vanished for ~10 Myr (the “Early Triassic reef gap”). On land, gorgonopsian and dicynodont synapsids collapsed; tropical forests disappeared (the “coal gap” of the Early Triassic).

The proximate cause is now well established as the eruption of the Siberian Traps Large Igneous Province — a flood-basalt province covering over 7 million km² with lavas and pyroclastics up to 4 km thick. The Siberian Traps released enormous quantities of CO2, SO2, halogens and potentially mercury. Payne et al. (2010 Science) used stable-isotope and trace- metal proxies to show that ocean acidification was a principal kill mechanism. The “super-greenhouse” temperatures of the Early Triassic (SST ~36°C in the tropics; Sun et al. 2012 Science) created an “equatorial dead zone.”

Secondary mechanisms included widespread ocean anoxia and euxinia (Grice et al. 2005 Science, documenting green-sulphur-bacterial biomarkers indicating photic-zone euxinia); methane-clathrate dissociation; and ultraviolet radiation from stratospheric ozone depletion by volcanic halogens (Beerling et al. 2007). The Siberian Traps magmatism also intruded thick Tunguska-basin coal and evaporite deposits, metamorphically releasing additional CO2 and halogens — amplifying the crisis beyond what basalt eruption alone would produce (Svensen et al. 2009).

In 1980, Luis Alvarez, his son Walter Alvarez, Frank Asaro and Helen Michel reported that clay layers at the Cretaceous–Paleogene boundary in Gubbio (Italy), Stevns Klint (Denmark) and Woodside Creek (New Zealand) contain anomalously high iridium (Alvarez et al. 1980 Science 208, 1095–1108). Iridium is a platinum-group element enriched in chondritic meteorites ~10,000-fold above crustal abundance. The inferred impact hypothesis was controversial for a decade until Hildebrand et al. (1991 Geology) located the 180 km Chicxulub crater buried under the Yucatán peninsula, dated precisely to 66.04 ± 0.05 Ma (Renne et al. 2013 Science) coincident with the K–Pg boundary.

Kill mechanisms include: (1) a global ejecta layer of molten spherules that rained back into the atmosphere, heating it to hundreds of degrees Celsius within hours and igniting global wildfires (Robertson et al. 2013); (2) a “nuclear winter” period of > 2 years of blocked sunlight from sulphate aerosols released by vapourising gypsum beds at the impact site (Brugger et al. 2017 GRL); (3) ocean acidification from sulfuric acid deposition (Henehan et al. 2019 PNAS); (4) tsunamis with estimated 1.5 km run-up in the Gulf of Mexico; and (5) longer-term climate cooling lasting decades.

The Deccan Traps of India were erupting during the same geological interval (66.3–65.5 Ma; Schoene et al. 2019 Science, Sprain et al. 2019), and some authors have argued for a volcanic trigger. The current consensus is that the impact was the principal cause but that the biosphere was perturbed by Deccan volcanism, giving a “one-two punch” pattern. The exceptional preservation at the Tanis site, North Dakota (DePalma et al. 2019 PNAS) captured fish with impact spherules in their gills — victims within hours of the impact.

Mass extinctions are not random with respect to ecology. Jablonski (2005 PNAS) showed that different kinds of traits survive different kinds of crises. At the K–Pg boundary, broad geographic range was the single strongest predictor of genus survival — a rule that breaks down for species-level survival during background times. During the Permo-Triassic, physiological tolerance to low-oxygen/low-pH conditions selected survivors: heavily calcified bivalves and corals suffered disproportionately, while burrowing bivalves and conodonts that could tolerate hypoxia did better (Knoll et al. 1996 Science).

Taxa that rebuild biospheres are often ecological generalists (“disaster taxa”) — the bivalve Claraia and the fern Isoetesblanketed post-extinction landscapes within the Early Triassic. Ecospace is typically refilled by 5–10 Myr, but the composition of the recovery fauna often differs fundamentally from what preceded it. The Permo-Triassic permanently shifted marine dominance from sessile Paleozoic benthos to mobile Modern-fauna bivalves and gastropods — the “Mesozoic Marine Revolution” of Vermeij (1977).

Between ~50,000 and 10,000 years ago, most of Earth’s megafauna (animals > 44 kg) went extinct. Australia lost ~85% of its genera by ~46 ka, coincident with human arrival (Miller et al. 2005 Science, dating the extinction via Genyornis egg shells). North America lost ~72% of its megafauna by ~13 ka, coincident with the Clovis archaeological culture. Eurasia saw staggered losses over 30–10 ka. Madagascar lost its giant lemurs, elephant birds and pygmy hippos by ~1 ka, after human colonisation in the first millennium CE. New Zealand’s moa disappeared within 100 years of Polynesian arrival ~1280 CE (Holdaway et al. 2014 PNAS).

Paul Martin’s (1967, 1984) overkill hypothesis proposed that human hunting was the principal cause. Supporting evidence: the timing correlates more tightly with human arrival than with any climatic event on each continent; Africa and southern Asia, where hominins evolved, retained megafauna because they co-evolved with humans; islands consistently lose megafauna within centuries of first human arrival.

Barnosky et al. (2004 Science 306, 70–75) argued for a climate-change role, especially in Eurasia, where losses were staggered over a glacial-interglacial transition and many taxa went extinct before human arrival. The current consensus is a synergistic climate-plus-humans model (Koch & Barnosky 2006 Annual Review of Ecology, Evolution and Systematics): climate stressed populations and fragmented ranges; human hunting delivered the killing blow for reduced and fragmented populations. Simulation models (Alroy 2001 Science; Bradshaw et al. 2021 Nature Communications) reproduce the continental extinction chronologies only under synergy.

AMS (Accelerator Mass Spectrometry) radiocarbon dating is the backbone of late-Pleistocene extinction chronology. Sample sizes of 5–10 mg of collagen-bearing bone allow individual specimen dating. Stuart (2004, 2015 Quaternary Science Reviews) compiled > 1000 AMS dates for Eurasian megafauna, showing that Mammuthus primigenius (woolly mammoth) persisted on Wrangel Island until ~4000 BP, long after mainland extinction ~12 ka — the last known mammoths overlapped with the Old Kingdom of Egypt.

Ancient DNA adds a phylogenetic and demographic dimension. The woolly mammoth genome (Palkopoulou et al. 2015) shows declining heterozygosity through the late Pleistocene, accelerating into the terminal decades on Wrangel. Cave bears (Stiller et al. 2014 Molecular Biology and Evolution) show steady decline beginning well before the LGM, consistent with a primarily human-pressure scenario. Sedimentary ancient DNA (sedaDNA; Willerslev et al. 2014 Nature) now allows extinction chronology to be built from Arctic permafrost sediments without fossil bones, detecting mammoth DNA into the mid-Holocene.

Before the recognition of plate tectonics, paleobiogeographers already saw that past faunas were partitioned into distinctive provinces. The Lower-Paleozoic trilobite provinces of Cook & Taylor (1975) divide the early Paleozoic into Laurentian, Baltic, Avalonian, Siberian and Gondwanan realms. The Glossopteris flora of the Permian (southern Gondwana) and the Dicroidium flora of the Triassic provided key evidence for Wegener’s continental-drift hypothesis. The Mesozoic has distinct Boreal and Tethyan ammonite provinces whose boundaries shift with sea-level and tectonic changes.

Provinciality is itself a biodiversity metric: high provinciality (many distinct biotas) indicates geographic and tectonic fragmentation, low provinciality indicates a supercontinent or open-ocean connection. Valentine & Moores (1970 Nature) argued that Phanerozoic diversity is driven in large part by continental configuration: peaks of diversity correspond to times of continental fragmentation (like today), troughs to supercontinent assembly (like the Permian Pangea). This tectonic interpretation remains a core organising principle of paleobiogeography.

Recovery from mass extinctions is typically characterised by (1) a dead zone of 0.5–2 Myr of very low diversity; (2) a disaster-taxon phase dominated by opportunistic, wide-ranging generalists; (3) a ~5–10 Myr rebuild of diversity with new dominant clades; and (4) a multi-million-year radiation refilling ecospace. Erwin (1998, 2001) studied the Early Triassic recovery in detail, showing ecospace occupation lagged genus counts — many new genera occupied similar niches to pre-extinction genera before new ecological interactions emerged. Full ecosystem complexity (predator guilds, reef ecosystems, coral dominance) took 8–12 Myr to rebuild.

The K–Pg recovery was accelerated: mammals diversified explosively within 300 kyr of the boundary (Lyson et al. 2019 Science, from the Corral Bluffs locality, Colorado), and global forest canopies recovered within 1 Myr. The key factor was that the K–Pg crisis did not cause widespread oceanic anoxia, unlike the Permo-Triassic; the pelagic ecosystem recovered while the Chicxulub-crater regional biosphere was still hostile. Recovery is faster and more complete when the kill mechanism is short-lived.

Ceballos, Ehrlich & Dirzo (2015 Science Advances 1, e1400253) estimated modern vertebrate extinction rates at 100–1000× background, and argued that even the most conservative estimate places current losses in the range of the Big Five. Their metric used only documented species-level extinctions since 1500 CE (~322 vertebrate species) and compared to the estimated background extinction rate of ~2 extinctions per 10,000 species per 100 years.

Direct drivers of contemporary extinction follow IPBES (2019)’s ranking: habitat loss/conversion (40%), direct exploitation (24%), climate change (18%), pollution (12%), invasive species (6%). Unlike previous mass extinctions driven by geological or extraterrestrial forcing, the current crisis is entirely attributable to a single species’ land-use change and greenhouse-gas emissions.

Barnosky et al. (2011 Nature) formalised the test: at current rates of IUCN-listed species loss, a Sixth Mass Extinction of Big-Five severity would be reached within a few centuries to a few millennia, depending on trajectory. McCallum (2015 Biodiversity and Conservation) and more recent data suggest the upper end of these timelines is conservative. The key distinction from background losses is that the driver (human land use + climate change) is still accelerating, while previous extinction triggers ran to completion over bounded geological intervals.

Paleobiogeographic inference depends on precise geological dating. The canonical tools:

• U–Pb zircon (CA-ID-TIMS): the gold standard for ages 1 Ma to 4.4 Ga, with precision of ~0.01%. Volcanic ash beds in sedimentary sequences yield ages usable for biostratigraphic calibration (Bowring & Schmitz 2003).
• Ar–Ar: widely used for volcanic samples 0.1–4 Ga; Fish Canyon Tuff monitor standard.
• AMS 14C: the workhorse for samples up to ~50 ka BP; IntCal20 calibration curve (Reimer et al. 2020).
• Optically stimulated luminescence (OSL): burial dates for quartz/feldspar grains, 100 yr to ~200 ka.
• 87Sr/86Sr strontium stratigraphy: seawater Sr isotopes traced through carbonates give ages for the last 500 Myr.
• δ\(^{13}\)C and δ\(^{18}\)O chemostratigraphy: correlation of excursions across sections, useful for K–Pg and Permo-Triassic boundary recognition.

Modern paleobiogeographic studies integrate multiple dating methods: the Chicxulub impact age of 66.043 ± 0.011 Ma (Sprain et al. 2018) uses 40Ar/39Ar on spherules plus U–Pb on zircons in tuffs bracketing the boundary. The Permo-Triassic boundary age of 251.902 ± 0.024 Ma (Burgess et al. 2014) uses CA-ID-TIMS on multiple interbedded ash layers in Meishan, China.