Extinction Cuvier’Sextinction “Ohio Animal” of Species Is a Recent Discovery

Extinction Cuvier’sExtinction “Ohio animal” of species is a recent discovery

http://en.wikipedia.org/wiki/Georges_Cuvier • Georges Cuvier, anatomist and naturalist at the Museum of Natural History, Paris, used comparative anatomy to prove that fossil bones belonged to a species (American Mammoth) that no longer used comparative anatomy, Georgescalls it a mammoth, Cuvier an elephant 29 existed (1769-1832) Cuvier proved that fossils of mammoths were distinct from all living elephant species Cuvier’s “Ohio animal”

“Ohio Animal”

Indian vs. African elephant 30 Mammoth vs. Indian elephant lower jaw vs. Indian Elephant used comparative anatomy, calls it a mammoth, an elephant 29

Cuvier proved that fossils of mammoths were distinct from all living elephant species

Indian vs. African elephant 30 Mammoth vs. Indian elephant Extinction is ‘normal’

• >99% of all species that have ever existed are now extinct

• Extinction rates have varied quite a bit through time

• Different groups have different characteristic species durations:

Mammals: ~ 2 million years

Foraminifera: ~ 20 million years Marine Animal Extinction rates over the past 540 million years

• Late Ordovician • Permian-Triassic • Pleistocene-Recent

Bambach,(2006,(Ann.$Rev.$Earth$&$Plan.$Sci.( Potential drivers of extinction

Background Mass

• Competition • Rapid climate change • Predation • Sea level change • Habitat loss • Wholesale habitat loss • Disease • Ocean acidiﬁcation • Climate change • Anoxia/hypoxia • Bad luck • Bolide Impacts • Disease? Selectivity of extinction:

What traits might inﬂuence extinction risk under different scenarios?

Individual Population • Physiology • Geographic range • Thermal tolerance • Latitudinal range • Diet • Environmental range • Home range size • Population density • Reproduction • Population growth rate • Gestation period • Dispersal • etc. • etc. Selective signature of mass extinctions: what is unusual about extinctions relative to extinctions at other times? Influenza A virus recycling revisited not associated with an influenza event comparable to An analogy: age distribution of mortality Fig. 5. Deaths from pneumonia and influenza in USA in three that of H3 in 1889±91 or H1 in 1918±20. influenza pandemicsduring a (adapted ‘normal’ from ﬂu Dauer pandemic & Serfling (vs.44); data Another means of testing the virus recycling for 1892 for Massachusetts only) hypotheses is through the protective effect of pre- during the 1918 pandemic existing antibody, as observed in 1968±70 for H3 (Fig. 3) and in 1977 for H1. The steep, uninterrupted increase in excess mortality with age in 1892 (44) (Fig. 5) suggests that the population at that time had no prior experience with H3, at least as far back as about 1800. The excess mortality rate for 1957±58 also provides no evidence of protection of any specific age group (Fig. 3). The well-known ``W'' excess mortality curve for 1918 is somewhat more difficult to interpret (Fig. 5). Excess mortality was elevated for all ages, with the rates being highest for ages below 2 years, young adults and those aged 580 years. These data are more consistent with an extraordinary vulnerability of young adults (4) than with protection of older adults through pre-existing H1 antibody. Thus, age-related excess mortality patterns Dowdle, 1999, Bulletin of the WHO alone suggest that H3 has circulated twice in the past a 200 years, from 1889±91 to some undetermined time Table 1. Historically recognized pandemics attributed to influenza prior to 1918, and from 1968 to the present; H2 has circulated once in the past 130 years, from 1957 to Years of occurrence HA subtype No. of years since 1967; and H1 has circulated twice in the past last pandemic 170 years, from 1918 to 1956, and from 1977 to 1729±33 Ð ? the present. However, evidence suggests that the 1781±82 Ð 49 1977 H1 virus may not have been a natural event but 1830±33 Ð 48 the reintroduction of the virus to humans from a 1889±91 H3b 55 frozen source (45, 46). 1918±20 H1 27 Years for which there is strong evidence of 1957±58 H2 37 influenza pandemics, with subtype etiology, are listed 1968±69 H3 10 c in Table 1. 1977 H1 9 a Modified from Potter (2). b Determined through retrospective serological studies (seroarchaeology). Conclusions c No excess mortality. This review of the evidence for influenza A virus recycling supports claims for recycling of the H3 proportions. Virtually everyone born before 1967 subtype, which first appeared in the pandemic of has had some immunological experience with H2 1889±91 and nearly 80 years later in the pandemic of antigens. 1968±69. No firm evidence was found to link the H2 Pandemic planning must be open to any subtype to a pandemic other than that of 1957±58. possibility. It is not known when the next influenza Thus, there is no firm basis for predicting the pandemic will occur, or which of the 15 (or more) sequential reintroduction of H2 on grounds of haemagglutinin subtypes will be involved. Effective recycling. However, the inclusion of H2 as one of global surveillance remains the key to influenza the potential candidates for pandemic planning preparedness. n would be prudent. H2 variants continue to circulate widely among aquatic birds (47). H2 is also one of Acknowledgements only three of the 15 influenza A virus subtypes I wish to thank K. Koporc for adapting the data from known to have caused a pandemic, and the possibility different laboratories and different times to the same remains that the number of subtypes transmissible to style and scale, K. Koporc and C. Anderson for their humans may be restricted (48). On the other hand, assistance in preparing the manuscript, and N. Cox, reintroduction of an H2 variant into the human K. Fukuda, G. Noble and S. Schoenbaum for their population within the next few years would not be constructive comments. expected to produce a pandemic of disastrous

Bulletin of the World Health Organization, 1999, 77 (10) 825 The Late Ordovician Mass Extinction

http://www.lifeonthinice.org/index.php#mi=2&pt=1&pi=10000&s=0&p=8&a=0&at=0 Climate change & mass extinctions

3000 class Archaeocyatha Bivalvia Cret.-Pg.

L. Ord Brachiopoda Cephalopoda Conodonta

Perm.-Tri. Crinoidea

L. Dev Demospongea Echinoidea 2000 Gastropoda Graptolithina

Tri.-Jur. Gymnolaemata Ostracoda Rugosa Scleractinia

Extant genera Stenolaemata Stromatoporoidea 1000 Tabulata Trilobita z.otherother

climate 1 Warm

Number of genera marine animals 0 -1 0 -2 -3 Cm O S D C P T J K Pg Ng Cool

500 400 300 200 100 0 Millions of years before present Millions of years before present REPORTS

18 cord reflects secular variations in global climate Ordovician with notable expansions throughout oxygen isotope compositions [d Ophos =~18to (2, 4)but,assumingpresent-dayvaluesforsea- the pelagic realm. The onset of the first major 23‰ Vienna standard mean ocean water (V- 18 water (d Oseawater ~ −1), the calculated temper- biodiversity surge occurred during the Mid Or- SMOW)] (19)givingplausiblepaleotemperatures atures are unrealistically high, reaching 70°C in dovician with extensive colonization of the ben- (~33° to 17°C) (16–18, 20, 21), encouraging fur- the Early Ordovician. This seeded further spec- thos including the establishment of hardgrounds ther exploration of the technique. 18 ulation that d Oseawater composition, and by im- and reefal systems (14), although diversifications Here we show the feasibility of in situ oxygen plication hydrothermal processes, evolved with of many groups peaked throughout the Late Or- isotope analysis of single conodont elements at a time (3, 5–7). This is due to the inherent limita- dovician (12, 14, 15). The causal mechanism(s) 30-mmscaleusingtheSHRIMPIIionmicroprobe tion of the oxygen isotope proxy being a com- that drove these radiations has been elusive and at The Australian National University (ANU), bined signature of temperature and seawater enigmatic given the long-standing belief that super- recently configured for high-precision stable- isotopic compositions, thereby comprising two greenhouse conditions prevailed. Clearly, local and isotope analysis (22). Durango apatite was used 18 unknowns. The proposition that d Oseawater regional environmental conditions (e.g., sedimen- as the primary isotope standard, its composition changed significantly over time is, however, tation, eustasy, temperatures, nutrients, ocean cir- independently determined by gas isotope ratio 18 inconsistent with altered seafloor basalt composi- culation) would have substantially shaped the mass spectrometry (d Oapatite =9.4‰). The tions and models of the seawater isotopic budget, character of evolving marine communities, as tooth enameloid of a modern great white shark 18 18 which imply that d Oseawater has remained would various biophysical mechanisms. However, (d Oapatite =22.3‰)wasusedasasecondary essentially constant since the Archean (8–11). overarching global conditions (e.g., climate, sea standard. The standard deviation of replicate analy- This suggests that these older biogenic carbonates level) would have also played key roles in this ses of these standards ranged from 0.42 to 0.14‰ do not reflect primary paleoseawater composi- major reconfiguration of the marine biosphere. within sessions, with higher precision reflecting tions, yet these issues remain contentious. To better characterize the Ordovician climate ongoing technical improvements. The conodonts The temperature record of Ordovician oceans regime, we have used conodont apatite as a po- (179 analyses, 102 specimens) were sampled from is central to understanding links between sea- tentially robust temperature archive. Although not 20 Ordovician to Early Silurian temporal horizons water chemistry, climate change, major bio-events, as abundant or easy to analyze as carbonate, the from 8 sites of similar tropical paleolatitudes and thus fundamental Earth processes. The marine phosphate mineralogy of conodont microfossils (23)acrossGondwana(Australia)andLaurentia biosphere underwent a profound transformation is more stable than that of biogenic marine car- (Canada), of primarily shallow subtidal (with two during the Great Ordovician Biodiversification bonates. Furthermore, conodonts are ubiquitous distal slope) facies (22). Consistent shallow-water on July 25, 2008 Event (GOBE), recognized as the longest period in Cambrian-Triassic marine sequences world- sampling avoided isotopic variability that would of sustained biodiversifications, increasing family wide and evolved rapidly, providing fine strati- likely be expressed by mixed biofacies. Mean pop- and genus numbers three- to fourfold (12)(Fig. graphic resolution. A major drawback, however, ulation compositions (Fig. 2) were determined with 1). This unparalleled event is characterized by the is that conodonts are small (~0.1 to 3 mm long), aprecisionoftypically0.5‰ (95% confidence replacement of the Cambrian Evolutionary Fauna so previous analyses (16, 17)havetypicallyre- level). At this precision, no pronounced isotopic with considerably more complex Paleozoic and quired “bulk” sampling, even those using infrared differences were found between tissue types (hy- Modern Evolutionary Faunas. The GOBE was laser isotope ratio monitoring gas chromatographic aline or albid) within individual conodont ele- terminated with sudden and catastrophic extinc- mass spectrometry (18). This is less than ideal ments, or between most elements within a single tions during the latest Ordovician (Hirnantian), because such samples may contain contaminants sample population. A larger range of composi- www.sciencemag.org probably associated with rapid ice sheet growth and remnant basal tissue that can compromise the tions was found in some Hirnantian conodont at the South Paleopole (13). analysis (18). Moreover, compositional hetero- populations, and in an older sample (Manitoba) Although there are different regional biodi- geneity between different taxa will not be discern- possibly affected by hypersaline conditions. 18 versification patterns, which are well recognized ible by bulk analyses, thereby requiring analysis The measured conodont d Oapatite composi- from biotal provincialism, at a higher level there of monospecific samples that are rarely available tions are internally consistent across geographi- are important global trends. Major increases in in sufficient volume. Bulk analyses of Silurian- cally disparate sites from two cratons, with no biocomplexity and biodiversity began in the Early Carboniferous conodonts have nonetheless yielded systematic patterns related to facies, and are

thereby interpreted as a global temporal trend Downloaded from Fig. 1. Biodiversity pat- (Fig. 2). Furthermore, the earliest Ordovician terns of marine fauna conodonts from Australia have compositions through geological time. Trotter et al., 2008, Science equivalent to those from Texas reported by Bassett Middle Cambrian to Si- et al.(24), and the Wenlock data from Cornwallis lurian (except Pridoli) tax- Island are consistent with coeval samples from 2000 18 onomic diversity trends Gotland (18). Throughout the Ordovician, d Oapatite at genus level [modified increased from ~15.3 to ~19.6‰ (V-SMOW), from Sepkoski (15)]. In- equivalent to temperatures from ~42° to ~23°C 1600 18 18 set shows Phanerozoic (d Oseawater = −1). Our d Oapatite record pro- taxonomic diversity of vides a first-order primary trend of Ordovician marine faunas at fam- The Late Ordovician 1200 climate variability that discriminates four main ily level [modified from climate regimes: (i) sustained cooling during the Sepkoski (12)]. Cambrian: Early Ordovician (~15.3 to ~18.3‰), reflecting a M, Middle; U, Upper. 800 global shift from greenhouse conditions (~42°C) Mass Extinction Ordovician T, Tremadoc; to modern equatorial temperatures (~28°C) over Ar, Arenig; Ln, Llanvirn; ~25 million years (Myr); (ii) climate stability with C, Caradoc; As, Ashgill. 400 modern equatorial temperatures for ~20 Myr from Silurian: Lly, Llandovery; the Mid through Late Ordovician; (iii) a rapid W, Wenlock; Lw, Ludlow.Marine invertebrate genera temperature drop (25)duringtheHirnantian,mark- ing the well-known latest Ordovician glaciation; and (iv) a return to modern equatorial temperatures by the early Wenlock (Fig. 3). Higher–

www.sciencemag.org SCIENCE VOL 321 25 JULY 2008 551 • ~60% of marine genera disappear during the Katian (457- 445.6 mya) and Hirnantian (445.6 - 443.7 mya) stages

• Two pulses: end-Katian pulse coinciding with cooling & expansion of Gondwanan ice sheets, end-Hirnantian pulse coinciding with warming & contraction of ice sheets

• No clear selective signature with respect to which taxonomic groups go extinct The Late Ordovician globe

Very limited life on land, High CO2 , Low O2

Siberia Laurentia

Baltica

Gondwana Gondwana

modiﬁed after R. Blakey: http://jan.ucc.nau.edu/~rcb7/450moll.jpg The Ordovician Period: little or no animal life on land, but abundant marine life Who was hit by the extinction?

Tabulate and Trilobites Brachiopods Rugose Corals

http://louisvillefossils.blogspot.com/2012/12/ http://weekstrilobites.com/Flexicalymene.htm http://drydredgers.org/brachplaty.htm naming-fossils.html Mollusks Graptolites Crinoids

http://louisvillefossils.blogspot.com/2010/12/ http://www.fossilsforsale.com/site_arc/index.cfm? http://www.geo-logic.org/Palaeontology/Graptolites.htm action=item&prod_id=191& agaricocrinus-americanus-crinoid.html Who was hit by the extinction?

http://www.palaeocast.com/episode-2-isotelus-rex/#.UXHCgaUgF20 Ultimate cause: glaciation and subsequent deglaciation of south polar Gondwana

? Late Ordovician glacial deposits in Morocco

Le Heron et al., 2010, Sedimentary Geology What caused cooling? Some ideas:

May have been caused by movement of Gondwana over the south pole...

Siberia Laurentia

Baltica

Gondwana Gondwana

modiﬁed after R. Blakey: http://jan.ucc.nau.edu/~rcb7/450moll.jpg Or, increased chemical weathering of silicate rock due to Taconic mountain-building

The silicate weathering feedback:

modiﬁed after R. Blakey: http://jan.ucc.nau.edu/~rcb7/450moll.jpg

http://images.summitpost.org/original/499686.jpg Or, increased chemical weathering of continental silicate rocks by early terrestrial ecosystems

Late Ordovician moss and fungal spores

The silicate weathering feedback:

http://www.shef.ac.uk/content/1/c6/03/41/74/ Science. 2000 Sep 15;289(5486):1884-5. wellman-research-pic-dec-07.jpg Late Ordovician North America (Laurentia)

Anticosti Island

modiﬁed after R. Blakey: http://jan.ucc.nau.edu/~rcb7/450moll.jpg English Head

N Baie des Homards ° 45 ' 9 4 Point Laframboise Natiscotec Creek LEGEND Salmon River N Chicotte Macaire Creek ° 30 '

9 Lousy Cove

4 Jupiter Gun River Fox Pt. Silurian Merrimack

N Becscie ° 15 ' 9

4 Ellis Bay Vauréal

Ordovician 0 5 10 20 measured section kilometers 64°30' 64°00' 63°30' 63°00' 62°30' 62°00'

Figure 1 Vauréal Canyon Gun River Fm.

1 cm.

Ellis Bay Fm.

1 cm.

Becscie Fm. Vauréal Fm. Stratigraphic ranges of orthide brachiopods on Anticosti

Laframboise Mbr., Ellis Bay Fm.

Jin and Zhan, 2008 Western Anticosti: Pt. Laframboise Shallow-water deposits

Deep-water deposits Western Anticosti: Pt. Laframboise

Deep water deposits (Silurian)

Reefs CaCO3 shells record the chemistry of the water in which they grew...

Korte et al., 2005 Stable isotope ratios in fossil shells

-Isotopes of an element vary in number of neutrons, but have a ﬁxed number of protons and electrons

-Isotopes have identical chemical interactions, but are often sorted by mass

δ13C : ratio of 13C to12C, provides information about changes in carbon cycle

δ18O : ratio of 18O to16O, provides information about changes in climate

Anticosti Island chemostratigraphy and sequence stratigraphy

TABLE 1. LOCATIONS OF MEASURED SECTIONS USED IN THE STUDY CHEMOSTRATIGRAPHIC RESULTS Anticosti Island Section Locations Section ID Latitude (°N) Longitude (°W) Location name Stable isotopic data are tabulated in GSA Western sector Data Repository Table DR1 (see footnote 1). D821 49°54.072′ 64°29.458′ English Head D801 49°49.567′ 64°26.574′ Junction Cliff D802 49°48.855′ 64°25.781′ Western Anticosti Island D803 49°48.646′ 64°25.520′ D804, 901 49°48.350′ 64°25.259′ Laframboise Point D805 49°47.763′ 64°22.735′ Cape Henry At Laframboise Point in the west (Fig. 7), D806 49°47.545′ 64°23.066′ Cape Henry Lousy Cove Member mudstones record a base- 13 East-central sector line Ccarb value of ~0.5^. An isotopic dis- 920 49°29.530′ 62°32.631′ Natiscotec tributary continuity separates the top of the Lousy Cove 911 49°23.853′ 62°23.583′ 9 Mile Pool, right bank D816, D817, 908 49°24.015′ 62°23.175′ 9 Mile Pool, left bank mudstones from the overlying sandy carbon- 13 D814, 909A/B 49°24.328′ 62°21.420′ 8 Mile Pool ate grainstones. In the coarser unit, Ccarb and 912, 912A 49°22.762′ 62°12.320′ Macaire Creek Quarry 13C rise by 2^. Although there is strong D813 49°22.838′ 62°12.230′ Macaire Creek stream A org D815 49°22.498′ 62°12.225′ Macaire Creek stream B  eld evidence for an erosional surface separat- Eastern sector ing the grainstones from the oncolite platform 13 914/A/B 49°19.198′ 61°51.399′ Lousy Cove bed, the continuity of Ccarb across this surface indicates that not much time is missing 13 at this boundary. Above the oncolites, Ccarb climbs to +4^ and fluctuates around that value through the biohermal unit. Similarly, 13 In the east sector at section 914 (Lousy Cove), 5 to 10 cm of relief on the underlying Lafram- Corg shows a parallel rise below the onco- the oncolite platform bed of the Laframboise boise Member (Fig. 6E). Dark-gray skeletal lite bed and  uctuations around the maximum Member is 50J60 cm thick and contains abun- grainstones at the bottom of the member are above it (Fig. 7). Up section, high-resolution dant corals and oncolites (Fig. 6A). The base replaced with hummocky cross-strati ed  ne sampling shows that there is a smooth decline 13 truncates aulecerid stromatoporoid fossils of the calcarenite within 2 m of the base. Normally in Ccarb from the +4^ maximum down to underlying Lousy Cove Member. Oncolites are graded bioclastic interbeds are common. +0.5^ from the top of the biohermal unit nucleated around clasts of the underlying bed (Fig. 6C). Bioherms are less abundant here than in the other sectors, and they are considerably Jones et al., 2011 smallerPt. (Fig. Laframboise: 6B). Fox Point Member13 (Becscie18 Formation) δ C, δ O 20 The base of the Becscie Formation at sections D804 and 901 (Laframboise Point) in the western sector is characterized by 1 m of inter- bedded grainstones and lime mudstones (Figs. 4B, 4C, and 4D) that appear in some places 15 to inter nger with the Laframboise Member bioherms (Fig. 4D). Some grainstones display Laframboise Point Fox hummocky cross-strati cation or wave ripples. A few meters up-section, the grainstone beds become thinner and less common, and mud- 10 stone dominates. Height in section (m) In the east-central sector, the Becscie For- mation begins with thin resistant sandstone that overlies the Laframboise oncolite platform Lousy Cove Lousy bed where bioherms are absent (Fig. 5D). The 5 sandstone frequently displays low-angle cross- strati cation. Above the sandstone, the Fox Point Member typically becomes recessive and poorly exposed for 2 to 4 m. At section 911, on 0 the right bank of the Salmon River, across from –5–4 –3 –2 –1 024135–30 –28 –26 –24 –22 26 28 30 32 34 18 13 13 13 the upstream section (9 Mile Pool), bioherms of δ Ocarb, δ Ccarb (‰, VPDB) δ Corg (‰, VPDB) ∆ C (‰, VPDB) the Laframboise Member are directly overlain with skeletal grainstones with a diverse fauna, Figure 7. Geochemistry and lithostratigraphy of sections D804 and 901 from Laframboise including ramose bryozoans and nautiloids. Point, western Anticosti Island. Oncolite platform bed data are shaded in gray. Filled data The base of the Fox Point Member at section points are from Ellis Bay Formation; open data points are from Becscie Formation. See 914 (Lousy Cove) in the eastern sector displays Figure 8 for legend. VPDBUVienna Peedee belemnite.

Geological Society of America Bulletin, July/August 2011 1651 Anticosti Island chemostratigraphy and sequence stratigraphy

TABLE 1. LOCATIONS OF MEASURED SECTIONS USED IN THE STUDY CHEMOSTRATIGRAPHIC RESULTS Anticosti Island Section Locations Section ID Latitude (°N) Longitude (°W) Location name Stable isotopic data are tabulated in GSA Western sector Data Repository Table DR1 (see footnote 1). D821 49°54.072′ 64°29.458′ English Head D801 49°49.567′ 64°26.574′ Junction Cliff D802 49°48.855′ 64°25.781′ Western Anticosti Island D803 49°48.646′ 64°25.520′ D804, 901 49°48.350′ 64°25.259′ Laframboise Point D805 49°47.763′ 64°22.735′ Cape Henry At Laframboise Point in the west (Fig. 7), D806 49°47.545′ 64°23.066′ Cape Henry Lousy Cove Member mudstones record a base- 13 East-central sector line Ccarb value of ~0.5^. An isotopic dis- 920 49°29.530′ 62°32.631′ Natiscotec tributary continuity separates the top of the Lousy Cove 911 49°23.853′ 62°23.583′ 9 Mile Pool, right bank D816, D817, 908 49°24.015′ 62°23.175′ 9 Mile Pool, left bank mudstones from the overlying sandy carbon- 13 D814, 909A/B 49°24.328′ 62°21.420′ 8 Mile Pool ate grainstones. In the coarser unit, Ccarb and 912, 912A 49°22.762′ 62°12.320′ Macaire Creek Quarry 13C rise by 2^. Although there is strong D813 49°22.838′ 62°12.230′ Macaire Creek stream A org D815 49°22.498′ 62°12.225′ Macaire Creek stream B  eld evidence for an erosional surface separat- Eastern sector ing the grainstones from the oncolite platform 13 914/A/B 49°19.198′ 61°51.399′ Lousy Cove bed, the continuity of Ccarb across this surface indicates that not much time is missing 13 at this boundary. Above the oncolites, Ccarb climbs to +4^ and fluctuates around that value through the biohermal unit. Similarly, 13 In the east sector at section 914 (Lousy Cove), 5 to 10 cm of relief on the underlying Lafram- Corg shows a parallel rise below the onco- the oncolite platform bed of the Laframboise boise Member (Fig. 6E). Dark-gray skeletal lite bed and  uctuations around the maximum Member is 50J60 cm thick and contains abun- grainstones at the bottom of the member are above it (Fig. 7). Up section, high-resolution dant corals and oncolites (Fig. 6A). The base replaced with hummocky cross-strati ed  ne sampling shows that there is a smooth decline 13 truncates aulecerid stromatoporoid fossils of the calcarenite within 2 m of the base. Normally in Ccarb from the +4^ maximum down to underlying Lousy Cove Member. Oncolites are graded bioclastic interbeds are common. +0.5^ from the top of the biohermal unit nucleated around clasts of the underlying bed (Fig. 6C). Bioherms are less abundant here than in the other sectors, and they are considerably Jones et al., 2011 smallerδ 13(Fig.C 6B). is controlled Fox Point Member (Becscie Formation) by global organic 20 The base of the Becscie Formation at sections D804carbon and 901 (Laframboise burial Point) in the western sector is characterized by 1 m of inter- bedded grainstones and lime mudstones (Figs. 4B, 4C, and 4D) that appear in some places 15 to inter nger with the Laframboise Member bioherms (Fig. 4D). Some grainstones display Laframboise Point Fox hummocky cross-strati cation or wave ripples. A few• metersEnhanced up-section, upwelling, the productivity grainstone beds become and thinner organic and carbon less common, burial? and mud- 10 stone dominates. Height in section (m) In• theDecreased east-central organic sector, carbon the Becscie For- mation oxidation begins with rate? thin resistant sandstone that overlies the Laframboise oncolite platform Lousy Cove Lousy bed• whereWeathering bioherms ofare exposed absent (Fig. 5D). The 5 sandstonecarbonate frequently rocks? displays low-angle cross- strati cation. Above the sandstone, the Fox Point Member typically becomes recessive and poorly exposed for 2 to 4 m. At section 911, on 0 the right bank of the Salmon River, across from –5–4 –3 –2 –1 024135–30 –28 –26 –24 –22 26 28 30 32 34 18 13 13 13 the upstream section (9 Mile Pool), bioherms of δ Ocarb, δ Ccarb (‰, VPDB) δ Corg (‰, VPDB) ∆ C (‰, VPDB) the Laframboise Member are directly overlain with skeletal grainstones with a diverse fauna, Figure 7. Geochemistry and lithostratigraphy of sections D804 and 901 from Laframboise including ramose bryozoans and nautiloids. Point, western Anticosti Island. Oncolite platform bed data are shaded in gray. Filled data The base of the Fox Point Member at section points are from Ellis Bay Formation; open data points are from Becscie Formation. See 914 (Lousy Cove) in the eastern sector displays Figure 8 for legend. VPDBUVienna Peedee belemnite.

Geological Society of America Bulletin, July/August 2011 1651 Anticosti Island chemostratigraphy and sequence stratigraphy

TABLE 1. LOCATIONS OF MEASURED SECTIONS USED IN THE STUDY CHEMOSTRATIGRAPHIC RESULTS Anticosti Island Section Locations Section ID Latitude (°N) Longitude (°W) Location name Stable isotopic data are tabulated in GSA Western sector Data Repository Table DR1 (see footnote 1). D821 49°54.072′ 64°29.458′ English Head D801 49°49.567′ 64°26.574′ Junction Cliff D802 49°48.855′ 64°25.781′ Western Anticosti Island D803 49°48.646′ 64°25.520′ D804, 901 49°48.350′ 64°25.259′ Laframboise Point D805 49°47.763′ 64°22.735′ Cape Henry At Laframboise Point in the west (Fig. 7), D806 49°47.545′ 64°23.066′ Cape Henry Lousy Cove Member mudstones record a base- 13 East-central sector line Ccarb value of ~0.5^. An isotopic dis- 920 49°29.530′ 62°32.631′ Natiscotec tributary continuity separates the top of the Lousy Cove 911 49°23.853′ 62°23.583′ 9 Mile Pool, right bank D816, D817, 908 49°24.015′ 62°23.175′ 9 Mile Pool, left bank mudstones from the overlying sandy carbon- 13 D814, 909A/B 49°24.328′ 62°21.420′ 8 Mile Pool ate grainstones. In the coarser unit, Ccarb and 912, 912A 49°22.762′ 62°12.320′ Macaire Creek Quarry 13C rise by 2^. Although there is strong D813 49°22.838′ 62°12.230′ Macaire Creek stream A org D815 49°22.498′ 62°12.225′ Macaire Creek stream B  eld evidence for an erosional surface separat- Eastern sector ing the grainstones from the oncolite platform 13 914/A/B 49°19.198′ 61°51.399′ Lousy Cove bed, the continuity of Ccarb across this surface indicates that not much time is missing 13 at this boundary. Above the oncolites, Ccarb climbs to +4^ and fluctuates around that value through the biohermal unit. Similarly, 13 In the east sector at section 914 (Lousy Cove), 5 to 10 cm of relief on the underlying Lafram- Corg shows a parallel rise below the onco- the oncolite platform bed of the Laframboise boise Member (Fig. 6E). Dark-gray skeletal lite bed and  uctuations around the maximum Member is 50J60 cm thick and contains abun- grainstones at the bottom of the member are above it (Fig. 7). Up section, high-resolution dant corals and oncolites (Fig. 6A). The base replaced with hummocky cross-strati ed  ne sampling shows that there is a smooth decline 13 truncates aulecerid stromatoporoid fossils of the calcarenite within 2 m of the base. Normally in Ccarb from the +4^ maximum down to underlying Lousy Cove Member. Oncolites are graded bioclastic interbeds are common. +0.5^ from the top of the biohermal unit nucleated around clasts of the underlying bed (Fig. 6C). Bioherms are less abundant here than in the other sectors, and they are considerably Jones et al., 2011 smaller (Fig. 6B). δ18O is controlled by Fox Point Member (Becscie Formation) both local temperature 20 The base of the Becscie Formation at sec- tionsand D804 andglobal 901 (Laframboise ice volume Point) in the western sector is characterized by 1 m of inter- bedded grainstones and lime mudstones (Figs. 4B, 4C, and 4D) that appear in some places 15 to inter nger with the Laframboise Member bioherms (Fig. 4D). Some grainstones display Laframboise Point Fox hummocky cross-strati cation or wave ripples. A few• metersIncreased up-section, glaciation the grainstone of the beds become poles? thinner and less common, and mud- 10 stone dominates. Height in section (m) In the east-central sector, the Becscie For- mation• Cooling begins with of the thin tropics? resistant sandstone that overlies the Laframboise oncolite platform Lousy Cove Lousy bed• whereBoth? bioherms are absent (Fig. 5D). The 5 sandstone frequently displays low-angle cross- strati cation. Above the sandstone, the Fox Point Member typically becomes recessive and poorly exposed for 2 to 4 m. At section 911, on 0 the right bank of the Salmon River, across from –5–4 –3 –2 –1 024135–30 –28 –26 –24 –22 26 28 30 32 34 18 13 13 13 the upstream section (9 Mile Pool), bioherms of δ Ocarb, δ Ccarb (‰, VPDB) δ Corg (‰, VPDB) ∆ C (‰, VPDB) the Laframboise Member are directly overlain with skeletal grainstones with a diverse fauna, Figure 7. Geochemistry and lithostratigraphy of sections D804 and 901 from Laframboise including ramose bryozoans and nautiloids. Point, western Anticosti Island. Oncolite platform bed data are shaded in gray. Filled data The base of the Fox Point Member at section points are from Ellis Bay Formation; open data points are from Becscie Formation. See 914 (Lousy Cove) in the eastern sector displays Figure 8 for legend. VPDBUVienna Peedee belemnite.

Geological Society of America Bulletin, July/August 2011 1651 ORDOVICIAN 18 SILURIAN Brachiopod Temperature and Katianseawater Hirn. Rhuddanianδ O Aeroniantrends Trilobite from Rugose coral “clumped”-1 isotope paleothermometryBryozoan -2 Anticosti

c a l i t e -3 Cinci. Arch

O -4 U. Miss. V. 18

ORDOVICIAN SILURIAN Brachiopod ! -5 ORDOVICIAN SILURIAN Brachiopod Katian Hirn. Rhuddanian Aeronian Trilobite Katian Hirn. Rhuddanian Aeronian Trilobite 455 450 445 440 Rugose coral 18 Rugose coral -1 Bryozoan 38-1 Results:Age, δmyaOwater Bryozoan -2 Anticosti 36-2 Anticosti 34 c a l i t e -3 Cinci. Arch c a l i t e -3 Cinci. Arch 32 O -4 U. Miss. V. O -4 U. Miss. V.

18 18 30 ! -5 ! 28-5 T 455 450 445 440 Temperature, ° C 455 450 445 440440 38 Age, mya 383 Age,Age, myamya 36 362 34 34 LGM 1 32 32 s e a w t r 0 30 O 30

28 18 28-1 ! Ice-free

Temperature, ° C 455 450 445 440 Temperature, ° C 455455 450 445 440 Age, mya Age,Age, myamya

Finnegan et al., 2011, Science Results: Inferred Ice Volumes*

18 *Assuming mean δ Oice equivalent to Last Glacial Maximum ORDOVICIAN SILURIAN Katian Hirn. Rhuddanian Aeronian

3 150 km 6 100 Last Glacial Maximum

50 Present

Ice volume, 10 0 455 450 445 440 Age, mya

Present LGM Ordovician-Silurian boundary sections Atypical: Anticosti Island Typical: Kentucky

Rhuddanian (E. Sil.)

Rhuddanian (E. Sil.) Hirnantian (L. Ord.)

Katian (L. Ord.) Katian (L. Ord.) Build-up of glaciers on land drains shallow marine habitats

San Francisco Bay 18,000 years ago

Image courtesy Lynn Ingram Sedimentary rocks record changes in continental ﬂooding through time Late Ordovician-Early Silurian sedimentary rocks and fossil collections in Laurentia

• Gap-bound sedimentary packages from Macrostrat (Peters, 2005) • Fossil occurrences from PBDB (Alroy et al, 2008)

Whiterock Chazyan Blackriveran Kirkfield Shermanian Maysvillian Richmondian

Hirnantian Rhuddanian Aeronian Telychian Wenlock

Fig. 1. MapsSandstone of sedimentary rocks depositedMixed carbonate-clastic across Laurentia from Middle OrdovicianChert (Dapingian) through the Early Silurian (Wenlockian) time. Red points mark PaleoDB collections. Colored polygons indicate sedimentary rock distribution and lithotype. Blue carbonate, dark blue mixed carbonate-clastic, ¼ ¼ gray fine clastics, tan mixed clastics, yellow sand, orange coarse clastics, blue-green chert, pink evaporites, brown metamorphic indet., ¼ Mixed clastic¼ Carbonate ¼ ¼ ¼ ¼ ¼ dark green igneous indet. Only the uppermost unit in each column is plotted. ¼ Fine clastic Evaporite Finnegan et al., 2012, PNAS between meridional range and thermal tolerance range in mod- the potential importance of habitat/substrate preference. Em- ern marine species (23, 24) and observed during the onset of ploying alternative measures of the distribution of gap durations Carboniferous (25) and Cenozoic (26, 27) glaciations. (mean, maximum, and minimum) did not substantially change the We integrated paleontological (28) and macrostratigraphic results of analyses. To control for geographic range size, a major (29) databases for Late Ordovician-Early Silurian strata of Laur- correlate of extinction risk in many Phanerozoic intervals (34), entia and matched fossil occurrence records to spatiotemporally we measured Laurentian occupancy (percent of potential sites explicit, gap-bound stratigraphic packages following the proce- where we sampled the genus) and great-circle distance. dures of Heim and Peters (30–32), albeit at a higher temporal As an indirect measure of thermal tolerance, we determined resolution (Dataset S1). This data structure provided a framework the highest-paleolatitude occurrence (irrespective of hemisphere) within which relevant macrostratigraphic, macroecological, and of each genus in the PaleoDB during, or prior to, the interval in macroevolutionary parameters could be quantified (Fig. S2). The question. Genera previously sampled above 40° paleolatitude were paleoenvironmental information contained within the data struc- scored as one and those restricted to paleolatitudes <40° were ture, though only coarsely constrained by lithotype, allowed us to scored as zero. We chose this value to reflect the approximate untangle two factors convolved in the record bias hypothesis: boundary between tropical and temperate/polar waters indicated stratigraphic truncations and environmental truncations (“habitat by Late Ordovician general circulation models (35) and zooplank- bias” sensu ref. 15). We calculated the first and last appearance of ton biotopes (6). Because our analysis was limited to the low- each genus in Laurentia from the merged dataset with analyses latitude paleocontinent of Laurentia, all of the genera in the da- focused on Laurentian extinction (e.g., extirpation) rather than taset have demonstrated ability to maintain viable populations in global extinction. Laurentian extirpation does not always coincide relatively warm, low-latitude settings. Maximum paleolatitude with global extinction; however, global stratigraphic ranges in the provides a measure of their ability to also tolerate cooler, more PaleoDB may be too poorly resolved to differentiate these scenar- seasonably variable seawater temperatures. Genera with a record ios across the Katian-Hirnantian boundary, and we do not have of temperate or high-latitude occurrences should be less sensitive sufficient macrostratigraphic data for other paleocontinents. The to environmental cooling than exclusively tropical genera. These processes underlying continental extirpation and global extinction genera also tend to be older, wider ranging, more speciose, and are probably similar but the former case is complicated by the po- to have broader habitat ranges than genera that were limited to tential for reinvasion from other paleocontinents and terranes (33). low latitudes (32, 36), all of which may reduce their susceptibility For each genus sampled in a given time interval, we calculated to extinction (34, 37). Tocontrol for the covariance of these factors, four potential determinants of extinction risk relevant to evalu- we quantified genus age (time since first appearance), global geo- ating the eustatic common cause and record bias hypotheses: per- graphic range (great circle distance), Laurentian and global species cent truncation (i.e., the percentage of sites occupied by a genus richness, and substrate preference (proportion of occurrences in that experienced stratigraphic truncation in that interval), per- carbonate vs. clastic units) for each genus in each interval. Finally, cent environmental truncation (i.e., the percentage of occupied we used the PaleoDB to assign a number of static variables that we sites that experienced a major shift in depositional environment assumed were invariant throughout a genus’ duration, including as measured by sedimentary lithology), median stratigraphic gap taxonomic class, trophic group, motility, life habit, and Laurentian duration (i.e., median time apportioned to the local hiatus for all endemicity. occupied sites that experienced truncation), and median environ- We used random forest classification models (38) to evaluate mental gap duration (i.e., median time to recurrence of a given the relative importance of each variable for determining extinc- lithofacies for all occupied sites that experienced environmental tion risk in ten Late Ordovician and Early Silurian time slices. truncation). Percent truncation and median stratigraphic gap Although random forest models have attractive properties for duration test the proposition that preservation probability and evaluating overall variable importance and constructing predic- extinction risk depend only on the distribution in time and space tive models with a high degree of accuracy (38), they do not of preserved sedimentary rock, whereas percent environmental provide easily interpreted measures of effect sign, strength, or truncation and median environmental gap duration acknowledge statistical significance. To complement the random forest ap-

2of6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1117039109 Finnegan et al. Dapingian Darriwilian Sandbian early Katian late Katian

Whiterock

Hirnantian Rhuddanian Aeronian Telychian Wenlock

Hirnantian Chazyan

Blackriveran−Kirkfield

Rhuddanian

Shermanian−Maysvillian Dapingian Darriwilian Sandbian early Katian late Katian

Aeronian Whiterock

Draining of shallow tropical seaways Hirnantian Rhuddanian Aeronian Telychian Wenlock

Richmondian Telychian late Katian HirnantianHirnantian Chazyan

Shallow ocean

Exposed land

Salt Blackriveran−Kirkfield basin 0° 0° Rhuddanian Open Wenlock ocean

Shermanian−Maysvillian

Aeronian

Richmondian Telychian

Wenlock Dapingian Darriwilian Sandbian early Katian late Katian

Whiterock

Hirnantian Rhuddanian Aeronian Telychian Wenlock

Hirnantian Chazyan

Blackriveran−Kirkfield

Rhuddanian

Shermanian−Maysvillian Dapingian Darriwilian Sandbian early Katian late Katian

Aeronian Whiterock

Draining of shallow tropical seaways Hirnantian Rhuddanian Aeronian Telychian Wenlock

Richmondian Telychian late Katian HirnantianHirnantian Chazyan

Shallow ocean

Exposed land

Salt Blackriveran−Kirkfield basin 0° 0° Rhuddanian Open Wenlock ocean

Hypothesis: genera that had large areas of their Late Ordovician geographic ranges drained should have Shermanian−Maysvillian experienced exceptionally high extinction rates Aeronian

Richmondian Telychian

Wenlock Ranges reﬂect interaction of climate and geography

30 north

0 300 500 700 900 Mean sea surface temperature Bivalve speciesrank.order latitudinal ranges 394 ..Hrmn ta./Plegorpy aaolmtlg,Pleeooy20(04 385–401 (2004) 210 Palaeoecology Palaeoclimatology, Palaeogeography, / al. et Herrmann A.D.

Changing temperatures would have imposed additional stresses on genera with limited thermal tolerance

Modeled Late Ordovician sea surface temperatures

Before glaciation During glaciation 394 ..Hrmn ta./Plegorpy aaolmtlg,Pleeooy20(04 385–401 (2004) 210 Palaeoecology Palaeoclimatology, Palaeogeography, / al. et Herrmann A.D.

Modiﬁed from Herrmann et al., 2004, P3 Fig. 6. Annual mean sea surface temperature distribution for simulations with high sea level, Ashgill paleogeography, and pCO levels of (a) 8 , and (b) 15 PAL; high sea level, 2 Â Â Caradoc paleogeography, and pCO levels of (c) 8 , and (d) 15 PAL; low sea level, Ashgill paleogeography, and pCO levels of (e) 8 . (For color see online version). 2 Â Â 2 Â

Changing temperatures would have imposed additional stresses on genera with limited thermal tolerance

Modeled Late Ordovician sea surface temperatures

Before glaciation During glaciation 394 ..Hrmn ta./Plegorpy aaolmtlg,Pleeooy20(04 385–401 (2004) 210 Palaeoecology Palaeoclimatology, Palaeogeography, / al. et Herrmann A.D.

Fig. 6. Annual mean sea surface temperature distribution for simulations with high sea level, Ashgill paleogeography, and pCO levels of (a) 8 , and (b) 15 PAL; high sea level, 2 Â Â Caradoc paleogeography, and pCO levels of (c) 8 , and (d) 15 PAL; low sea level, Ashgill paleogeography, and pCO levels of (e) 8 . (For color see online version). 2 Â Â 2 Â proach, we used multiple logistic regression (39) to examine the paleocontinents and terranes (33, 40), lag between the creation subset of the variables most commonly implicated in determining of new habitat via cratonic flooding and evolutionary response in extinction risk throughout the time series. Finally, we used pre- genus origination (33, 41), and, possibly, to under-sampling of LOME (early Katian) extinction patterns to train a random forest Rhuddanian-Aeronian strata (42). The per-capita extinction rate model for predicting which late Katian genera would have been (43) of sampled genera and the package truncation rate peaked expected to go extinct if the LOME simply represented a conti- in the second half of the Katian stage; however, extinction rates proach,nuation we of used“background multiple” logisticextinction regression processes. (39) Differences to examine the be- paleocontinentsremained slightly and elevated terranes in the (33, Hirnantian 40), lag between stage despite the creation a sharp subsettween of the the predicted variables and most observed commonly selectivity implicated patterns in determining highlight ofdecline new habitat in the viarate cratonic of package flooding truncations and evolutionary (Fig. 2B). response in extinctionspecific changes risk throughout in extinction the regime time series. that accompanied Finally, we used the pre- first genusThe origination number of (33, through-rangers 41), and, possibly, (genera to under-sampling that are sampled of LOMEpulse of (early the LOME. Katian) extinction patterns to train a random forest Rhuddanian-Aeronianat some point before strata and after (42). the The analyzed per-capita interval extinction but rate not model for predicting which late Katian genera would have been (43)within of it) sampled increases genera sharply and in the the package Hirnantian truncation stage and rate remains peaked expectedResults and to go Discussion extinct if the LOME simply represented a conti- inrelatively the second high half until of the the late Katian Early stage; Silurian however, period extinction (Fig. 2A). rates This nuationLaurentian of sampled“background genus” diversityextinction and processes. total number Differences of sedimen- be- remainedpattern is slightly in accord elevated with in the previous Hirnantian studies stage that despite found a sharp that, tweentary packages the predicted (i.e., local and sections) observed display selectivity similar patterns trends (Fig. highlight 2A): declinealthough in preservation the rate of probability package truncations decreased (Fig. through 2B). this interval, specifica Middle changes Ordovician in extinction to Late regime Ordovician that accompanied rise followed the by first a theThe decrease number could of through-rangers not fully explain (genera the late that Katian are extinction sampled pulseHirnantian of the drop LOME. and Early Silurian recovery. Notably, the num- atpulse some (42, point 44, before45). We and excluded after the through-ranger analyzed interval genera but from not ber of Early Silurian packages rebounded much faster than withinselectivity it) increases analyses sharply for both in practical the Hirnantian and theoretical stage and reasons.remains Resultssampled and diversity, Discussion likely due to delayed immigration from other relativelyPractically, high it is until impossible the late to Early measure Silurian aspects period of the(Fig. geographic 2A). This Laurentian sampled genus diversity and total number of sedimen- pattern is in accord with previous studies that found that, tary packages (i.e., local sections) display similar trends (Fig. 2A): although preservation probability decreased through this interval, a Middle Ordovician to Late Ordovician riseA followed by a the decrease could not fully explain the late Katian extinction Hirnantian drop and Early Silurian recovery. Notably, the num- pulse (42, 44, 45). We excluded through-ranger genera from ber of Early Silurian packages rebounded much faster than selectivity analyses for both practical and theoretical reasons. sampled diversity, likely due to delayed immigration from other Practically, it is impossible to measure aspects of the geographic

A B EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES Determinants of marine invertebrate extinction risk

B EARTH, ATMOSPHERIC, ECOLOGY Polar Ice VolumeC AND PLANETARY SCIENCES

Tropical Temperature

% Habitat Loss C ECOLOGY

Finnegan et al., 2012, PNAS D

Fig. 2. (A) Time series of sampled genus diversity, number of through-ranger genera, and number of stratigraphic packages (i.e., local sections) in Laurentia from Middle Ordovician (Dapingian) through Early Silurian (Wenlockian) time. (B) Genus extinction rate (43) and package truncation rate (14) within each interval. Diversity and extinction rates are based only on the genera sampled in a given interval and exclude unobserved “through-ranger” genera. (C) Heat- map showing the importance (measured as OOB error) of each predictor in classifying genera as extinct or surviving from random forest models of each interval. Variables are ranked from top to bottom by their importance in the late Katian interval. (D) Area under receiver operating characteristic curve (AUC), a measure of model sensitivity, for each interval. Perfect classification would be indicated by an AUC of 1.0,amodelthatisnobetterthanrandom would have an AUC of approximately 0.5. The Rhuddanian model fails to predict any extinctions and hence AUC is undefined for this interval. Fig. 2. (A) Time series of sampled genus diversity, number of through-ranger genera, and number of stratigraphic packages (i.e., local sections) in Laurentia Dap Dapingian, Darr Dariwillian, San Sandbian, K early Katian, K late Katian, H Hirnantian, Rh Rhuddanian, Ae Aeronian, Tely from¼ Middle Ordovician¼ (Dapingian) through¼ Early Silurian1 (Wenlockian)¼ time.2 (¼B) Genus extinction¼ rate (43) and package¼ truncation rate¼ (14) within each¼ Telychian, Wen Wenlockian. interval. Diversity¼ and extinction rates are based only on the genera sampled in a given interval and exclude unobserved “through-ranger” genera. (C) Heat- map showing the importance (measured as OOB error) of each predictor in classifying genera as extinct or surviving from random forest models of each interval. Variables are ranked from top to bottom by their importance in the late Katian interval. (D) Area under receiver operating characteristic curve Finnegan et al. PNAS Early Edition ∣ 3of6 (AUC), a measure of model sensitivity, for each interval. Perfect classification would be indicated by an AUC of 1.0,amodelthatisnobetterthanrandom would have an AUC of approximately 0.5. The Rhuddanian model fails to predict any extinctions and hence AUC is undefined for this interval. Dap Dapingian, Darr Dariwillian, San Sandbian, K early Katian, K late Katian, H Hirnantian, Rh Rhuddanian, Ae Aeronian, Tely ¼ ¼ ¼ 1 ¼ 2 ¼ ¼ ¼ ¼ ¼ Telychian, Wen Wenlockian. ¼

Finnegan et al. PNAS Early Edition ∣ 3of6 Fig. S3. Diversity, extinction rate, and selectivity trends for the subset of genera that were sampled at every interval within their stratigraphic range. (A)Genus diversity and number of stratigraphic packages in Laurentia from Middle Ordovician (Dapingian) through Early Silurian (Wenlockian) time. (B)Genusextinction rate and package truncation rate within each interval. (C–E) Results of multiple logistic regressions of extinction risk on percent truncation (A), percent environmental truncation (B), and maximum paleolatitude (C) for each interval. Positive log-odds indicate that extinction risk increases as the variable in question increases, and vice versa. In addition to the three variables figured, regressions controlled for genus age, occupancy, substrate preference, and Laurentian and global geographic range. Dap Dapingian, Darr Dariwillian, San Sandbian, K early Katian, K late Katian, H Hirnantian, Rh Rhuddanian, ¼ ¼ ¼ 1 ¼ 2 ¼ ¼ ¼ Ae Aeronian, Tely Telychian, Wen Wenlockian. Hirnantian and Rhuddanian intervals contain too few genera for analyzing some predictors. ¼ ¼ ¼

Dapingian-earlyBefore mass extinction Katian Selective Signature: lateMass Katian extinction 80 Rhuddanian-WenlockianAfter mass extinction

Exclusively low- 60 Finnegan et al., latitude genera 2012, PNAS 40

much harder hit % Extinction % Extinction % Extinction than those with 20 broad latitudinal 0 distributions 20 30 40 50 60 70 80 90 Maximum Maximum Paleolatitude paleolatitude

Fig. S4. Proportional extinction of genera as a function of maximum paleolatitude of occurrence for the late Katian interval compared to the averages for the Dapingian and early Katian (Middle-Late Ordovician) and Rhuddanian to Wenlockian (Early Silurian). Error bars are 95% binomial confidence intervals and largely reflect variation in the number of genera in each bin.

Vandenbrouke et al., 2010

Finnegan et al. www.pnas.org/cgi/doi/10.1073/pnas.1117039109 2of3 Mean Temperature (°C) Mean Temperature Conclusions:

• The Late Ordovician glaciation was at least as large, in terms of ice volumes, as the Pleistocene glaciation • Tropical seawater temperatures fell by ~5º C during the Late Ordovician glacial maximum • Growth of glaciers caused sea levels to fall and drove a massive reduction in the area of shallow seaways

• Reduction of shallow seaways combined with cooling temperatures led to large-scale habitat loss and resulting extinction Why did Late Ordovician glaciation cause a major mass extinction, but not subsequent glaciations?

Late Ordovician Sea level, continental conﬁguration & biogeography

Mio-Pliocene

R. Blakey: http://jan.ucc.nau.edu