Age of cichlids: new dates for ancient lake fish radiations

Timing divergence events allow us to infer the conditions under which biodiversity has evolved and gain important insights into the mechanisms driving evolution. Cichlid fishes are a model system for studying speciation and adaptive radiation, yet, we have lacked reliable timescales for their evolution. Phylogenetic reconstructions are consistent with cichlid origins prior to Gondwanan landmass fragmentation 121-165 MYA, considerably earlier than the first known fossil cichlids (Eocene). We examined the timing of cichlid evolution using a relaxed molecular clock calibrated with geological estimates for the ages of 1) Gondwanan fragmentation and 2) cichlid fossils. Timescales of cichlid evolution derived from fossil-dated phylogenies of other bony fishes most closely matched those suggested by Gondwanan breakup calibrations, suggesting the Eocene origins and marine dispersal implied by the cichlid fossil record may be due to its incompleteness. Using Gondwanan calibrations, we found accumulation of genetic diversity within the radiating lineages of the African Lakes Malawi, Victoria and Barombi Mbo, and Palaeolake Makgadikgadi began around or after the time of lake basin formation. These calibrations also suggest Lake Tanganyika was colonized independently by the major radiating cichlid tribes that then began to accumulate genetic diversity thereafter. These results contrast with the widely accepted theory that diversification into major lineages took place within the Tanganyika basin. Together, this evidence suggests that ancient lake habitats have played a key role in generating and maintaining diversity within radiating lineages and also that lakes may have captured preexisting cichlid diversity from multiple sources from which adaptive radiations have evolved.

Provenance discrimination of Lower Cretaceous synrift sandstones (eastern Iberian Chain Spain) Constraints from detrital modes heavy minerals and geochemistry

Petrography, geochemical whole-rock composition, and chemical analyses of tourmaline were performed in order to determine the source areas of Lower Cretaceous Mora, El Castellar, and uppermost Camarillas Formation sandstones from the Iberian Chain, Spain. Sandstones were deposited in intraplate subbasins, which are bound by plutonic and volcanic rocks of Permian, Triassic, and Jurassic age, Paleozoic metamorphic rocks, and Triassic sedimentary rocks. Modal analyses together with petrographic and cathodoluminescence observations allowed us to define three quartz-feldspathic petrofacies and recognize diagenetic processes that modified the original framework composition. Results from average restored petrofacies are: Mora petrofacies = P/F >1 and Q(r)70 F(r)22 R(r)9; El Castellar petrofacies = P/F >1 and Q(r)57 F(r)25 R(r)18; and Camarillas petrofacies = P/F ∼ zero and Q(r)64 F(r)28 R(r)7 (P—plagioclase; F—feldspar; Q—quartz; R—rock fragments; r—restored composition). Trace-element and rare earth element abundances of whole-rock analyses discriminate well between the three petrofacies based on: (1) the Rb concentration, which is indicative of the K content and reflects the amount of K-feldspar modal abundance, and (2) the relative modal abundance of heavy minerals (tourmaline, zircon, titanite, and apatite), which is reproduced by the elements hosted in the observed heavy mineral assemblage (i.e., B and Li for tourmaline; Zr, Hf, and Ta for zircon; Ti, Ta, Nb, and their rare earth elements for titanite; and P, Y, and their rare earth elements for apatite). Tourmaline chemical composition for the three petrofacies ranges from Fe-tourmaline of granitic to Mg-tourmaline of metamorphic origin. The three defined petrofacies suggest a mixed provenance from plutonic and metamorphic source rocks. However, a progressively major influence of granitic source rocks was detected from the lowermost Mora petrofacies toward the uppermost Camarillas petrofacies. This provenance trend is consistent with the uplift and erosion of the Iberian Massif, which coincided with the development of the latest Berriasian synrift regional unconformity and affected all of the Iberian intraplate basins. The uplifting stage of Iberian Massif pluton caused a significant dilution of Paleozoic metamorphic source areas, which were dominant during the sedimentation of the lowermost Mora and El Castellar petrofacies. The association of petrographic data with whole-rock geochemical compositions and tourmaline chemical analysis has proved to be useful for determining source area characteristics, their predominance, and the evolution of source rock types during the deposition of quartz-feldspathic sandstones in intraplate basins. This approach ensures that provenance interpretation is consistent with the geological context.

Sixty-four-detector CT angiography of infrarenal aortic neck length and angulation: prospective analysis of interobserver variability

PURPOSE: To quantify the interobserver variability of abdominal aortic aneurysm (AAA) neck length and angulation measurements. MATERIALS AND METHODS: A total of 25 consecutive patients scheduled for endovascular AAA repair underwent follow-up 64-row computed tomographic (CT) angiography in 0.625-mm collimation. AAA neck length and angulation were determined by four blinded, independent readers. AAA neck length was defined as the longitudinal distance between the first transverse CT slice directly distal to the lowermost renal artery and the first transverse CT slice that showed at least a 15% larger outer aortic wall diameter versus the diameter measured directly below the lowermost renal artery. Infrarenal AAA neck angulation was defined as the true angle between the longitudinal axis of the proximal AAA neck and the longitudinal axis of the AAA lumen as analyzed on three-dimensional CT reconstructions. RESULTS: Mean deviation in aortic neck length determination was 32.3% and that in aortic neck angulation was 32.1%. Interobserver variability of aortic neck length and angulation measurements was considerable: in any reader combination, at least one measurement difference was outside the predefined limits of agreement. CONCLUSIONS: Assessment of the longitudinal extension and angulation of the infrarenal aortic neck is associated with substantial observer variability, even if measurement is carried out according to a standardized protocol. Further studies are mandatory to assess dedicated technical approaches to minimize variance in the determination of the longitudinal extension and angulation of the infrarenal aortic neck.

Neotethys closure history of Anatolia: insights from 40Ar–39Ar geochronology and P–T estimation in high pressure metasedimentary rocks

The multiple high-pressure (HP), low-temperature (LT) metamorphic units of Western and Central Anatolia offer a great opportunity to investigate the subduction- and continental accretion-related evolution of the eastern limb of the long-lived Aegean subduction system. Recent reports of the HP–LT index mineral Fe-Mg-carpholite in three metasedimentary units of the Gondwana-derived Anatolide–Tauride continental block (namely the Afyon Zone, the Ören Unit and the southern Menderes Massif) suggest a more complicated scenario than the single-continental accretion model generally put forward in previous studies. This study presents the first isotopic dates (white mica 40Ar–39Ar geochronology), and where possible are combined with P–T estimates (chlorite thermometry, phengite barometry, multi-equilibrium thermobarometry), on carpholite-bearing rocks from these three HP–LT metasedimentary units. It is shown that, in the Afyon Zone, carpholite-bearing assemblages were retrogressed through greenschist-facies conditions at c. 67–62 Ma. Early retrograde stages in the Ören Unit are dated to 63–59 Ma. In the Kurudere–Nebiler Unit (HP Mesozoic cover of the southern Menderes Massif), HP retrograde stages are dated to c. 45 Ma, and post-collisional cooling to c. 26 Ma. These new results support that the Ören Unit represents the westernmost continuation of the Afyon Zone, whereas the Kurudere–Nebiler Unit correlates with the Cycladic Blueschist Unit of the Aegean Domain. In Western Anatolia, three successive HP–LT metamorphic belts thus formed: the northernmost Tavşanlı Zone (c. 88–82 Ma), the Ören–Afyon Zone (between 70 and 65 Ma), and the Kurudere–Nebiler Unit (c. 52–45 Ma). The southward younging trend of the HP–LT metamorphism from the upper and internal to the deeper and more external structural units, as in the Aegean Domain, points to the persistence of subduction in Western Anatolia between 93–90 and c. 35 Ma. After the accretion of the Menderes–Tauride terrane, in Eocene times, subduction stopped, leading to continental collision and associated Barrovian-type metamorphism. Because, by contrast, the Aegean subduction did remain active due to slab roll-back and trench migration, the eastern limb (below Southwestern Anatolia) of the Hellenic slab was dramatically curved and consequently teared. It therefore is suggested that the possibility for subduction to continue after the accretion of buoyant (e.g. continental) terranes probably depends much on palaeogeography.

Upper crustal deformation in continent-continent collision: A case study from the Bernard nappe complex (Valais, Switzerland)

The Penninic nappes in the Swiss Alps formed during continental collision between the Adriatic and European plates in Cenozoic times. Although intensely studied, the finite geometry of the basement-bearing Penninic nappes in western Switzerland has remained a matter of debate for decades (e.g., “Siviez-Mischabel dilemma”) and the paleogeographic origin of various nappes has been disputed. Here, we present new structural data for the central part of the Penninic Bernard nappe complex, which contains pre-Permian basement and Permo-Mesozoic metasedimentary units. Our lithological and structural observations indicate that the discrepancy between the different structural models proposed for the Bernard nappe complex can be explained by a lateral discontinuity. In the west, the presence of a Permian graben caused complex isoclinal folding, whereas in the east, the absence of such a graben resulted mainly in imbricate thrusting. The overall geometry of the Bernard nappe complex is the result of three main deformation phases: (1) detachment of Mesozoic cover sediments along Triassic evaporites (Evolène phase) during the early stages of collision, (2) Eocene top-to-the-N(NW) nappe stacking (Anniviers phase), and (3) subsequent backfolding and backshearing (Mischabel phase). The southward localized backshearing is key to understand the structural position and paleogeographic origin of units, such as the Frilihorn and Cimes Blanches “nappes” and the Antrona ophiolites. Based on these observations, we present a new tectonic model for the entire Penninic region of western Switzerland and discuss this model in terms of continental collision zone processes.

Temperature-Time History of Subducted Continental-Crust, Mount Olympos Region, Greece

In the Mt. Olympos region of northeastern Greece, continental margin strata and basement rocks were subducted and metamorphosed under blueschist facies conditions, and thrust over carbonate platform strata during Alpine orogenesis. Subsequent exposure of the subducted basement rocks by normal faulting has allowed an integrated study of the timing of metamorphism, its relationship to deformation, and the thermal history of the subducted terrane. Alpine low-grade metamorphic assemblages occur at four structural levels. Three thrust sheets composed of Paleozoic granitic basement and Mesozoic metasedimentary cover were thrust over Mesozoic carbonate rocks and Eocene flysch; thrusting and metamorphism occurred first in the highest thrust sheets and progressed downward as units were imbricated from NE to SW. 40Ar/39Ar spectra from hornblende, white mica, and biotite samples indicate that the upper two units preserve evidence of four distinct thermal events: (1) 293–302 Ma crystallization of granites, with cooling from >550°C to <325°C by 284 Ma; (2) 98–100 Ma greenschist to blueschist-greenschist transition facies metamorphism (T∼350–500°C) and imbrication of continental thrust sheets; (3) 53–61 Ma blueschist facies metamorphism and deformation of the basement and continental margin units at T<350–400°C; (4) 36–40 Ma thrusting of blueschists over the carbonate platform, and metamorphism at T∼200–350°C. Only the Eocene and younger events affected the lower two structural packages. A fifth event, indicated by diffusive loss profiles in microcline spectra, reflects the beginning of uplift and cooling to T<100–150°C at 16–23 Ma, associated with normal faulting which continued until Quaternary time. Incomplete resetting of mica ages in all units constrains the temperature of metamorphism during continental subduction to T≤350°C, the closure temperature for Ar in muscovite. The diffusive loss profiles in micas and K-feldspars enable us to “see through” the younger events to older events in the high-T parts of the release spectra. Micas grown during earlier metamorphic events lost relatively small amounts of Ar during subsequent high pressure-low temperature metamorphism. Release spectra from phengites grown during Eocene metamorphism and deformation record the ages of the Ar-loss events. Alpine deformation in northern Greece occurred over a long time span (∼90 Ma), and involved subduction and episodic imbrication of continental basement before, during, and after the collision of the Apulian and Eurasian plates. Syn-subduction uplift and cooling probably combined with intermittently higher cooling rates during extensional events to preserve the blueschist facies mineral assemblages as they were exhumed from depths of >20 km. Extension in the Olympos region was synchronous with extension in the Mesohellenic trough and the Aegean back-arc, and concurrent with westward-progressing shortening in the external Hellenides.

Simulation of the Impact of Dams and Fishing Weirs on Reproductive Potential of Silver-Phase American Eels in the Kennebec River Basin, Maine

I modeled the cumulative impact of hydroelectric projects with and without commercial fishing weirs and water-control dams on the production, survival to the sea, and potential fecundity of migrating female silver-phase American eels, Anguilla rostrata in the Kennebec River basin, Maine, This river basin has 22 hydroelectric projects, 73 water-control dams, and 15 commercial fishing weir sites. The modeled area included an 8,324 km(2) segment of the drainage area between Merrymeeting Bay and the upper limit of American eel distribution in the basin. One set of input,, (assumed or real values) concerned population structure (Le., population density and sex ratio changes throughout the basin, female length-class distribution, and drainage area between dams), Another set concerned factors influencing survival and potential fecundity of migrating American eels (i.e., pathway sequences through projects, survival rate per project by length-class. and length-fecundity relationship). Under baseline conditions about 402,400 simulated silver female American eels would be produced annually reductions in their numbers due to dams and weirs would reduce the realized fecundity (i.e., the number of eggs produced by all females that survived the migration). Without weirs or water-control dams, about 63% of the simulated silverphase American eels survived their freshwater spawning migration run to the sea when the survival rate at each hydroelectric dam was 9017, 40% survived at 80% survival per dam, and 18% survived at 60% survival per dam. Removing the lowermost hydroelectric dam on the Kennebec River increased survival by 6.0-7.6% for the basin. The efficient commercial weirs reduced survival to the sea to 69-76%( of what it would have been without weirs', regardless of survival rates at hydroelectric dams. Water-control dams had little impact on production in this basin because most were located in the upper reaches of tributaries. Sensitivity analysis led to the conclusion that small changes in population density and female length distribution had greater effects on survival and realized fecundity than similar changes in turbine survival rate. The latter became more important as turbine survival rate decreased. Therefore, it might be more fruitful to determine population distribution in basins of interest than to determine mortality rate at each hydroelectric project.

Dating polygenetic metamorphic assemblages along a transect across the Western Alps

Multichronometric analyses were performed on samples from a transect in the French-Italian Western Alps crossing nappes derived from the Briançonnais terrane and the Piemonte-Liguria Ocean, in an endeavour to constrain the high-pressure (HP) metamorphism and the retrogression history. 12 samples of white mica were analysed by 39Ar-40Ar stepwise heating, complemented by 2 samples from the Monte Rosa 100 km to the NE and also attributed to the Briançonnais terrane. One Sm-Nd and three Lu-Hf garnet ages from eclogites were also obtained. White mica ages decrease from ca. 300 Ma in the westernmost samples (Zone Houillère), reaching ca. 300 °C during Alpine metamorphism, to < 48 Ma in the internal units to the East, which reached ca. 500 °C during Alpine orogeny. The conventional “thermochronological” interpretation postulates Cretaceous Eo-Alpine HP metamorphism and younger “cooling ages” in the higher-temperature samples. However, Eocene Lu-Hf and Sm-Nd ages from the same samples cannot be interpreted as post-metamorphic cooling ages, which makes a Cretaceous eclogitization untenable. The age date from this transect require instead to replace conventional “thermochronology” by an approach combining age dating with detailed geochemical, petrological and microstructural investigations. Petrology reveals important mineralogical differences along the transect. Samples from the Zone Houillère mostly contain detrital mica. White mica with Si > 6.45 atoms per formula unit becomes more abundant eastward. Across the whole traverse, HP phengitic mica forms the D1 foliation. Syn-D2 mica is Si-poorer and associated with nappe stacking, exhumation, and hydrous retrogression under greenschist facies conditions. D1 phengite is very often corroded, overgrown or intergrown by syn-D2 muscovite. Most importantly, syn-D2 recrystallization is not limited to S2 schistosity domains; microchemical fingerprinting shows that it also can form pseudomorphs after crystals that could be mistaken to have formed during D1 based on microstructural arguments alone. Thereby the Cl concentration in white mica is a useful discriminator, since D2 retrogression was associated with a less saline fluid than eclogitization. Once the petrological stage is set, geochronology is straightforward. All samples contain mixtures of detrital, syn-D1 and syn-D2 mica, and retrogression phases (D3) in greatly varying proportions according to local pressure-temperature-fluid activity-deformation conditions. The correlation of age vs. Cl/K clearly identifies 47 ± 1 Ma as the age of formation of syn-D1 mica along the entire transect, including the Monte Rosa nappe samples. The inferred age of the greenschist-facies low-Si syn-D2 mica generation ranges within 39-43 Ma, with local variations. Coexistence of D1 and D2 ages, and the constancy of non-reset D1 ages along the entire transect, are strong evidence that the D1 white mica ages are very close to formation ages. Volume diffusion of Ar in white mica (activation energy E = 250 kJ/mol; pressure-adjusted diffusion coefficient D’0 < 0.03 cm2 s-1) has a subordinate effect on mineral ages compared to both prograde and retrograde recrystallization in most samples. Eocene Lu-Hf and Sm-Nd garnet ages are prograde and predate the HP peak.

Retrieving the paleoclimatic signal from the deeper part of the EPICA Dome C ice core

An important share of paleoclimatic information is buried within the lowermost layers of deep ice cores. Because improving our records further back in time is one of the main challenges in the near future, it is essential to judge how deep these records remain unaltered, since the proximity of the bedrock is likely to interfere both with the recorded temporal sequence and the ice properties. In this paper, we present a multiparametric study (δD-δ18Oice, δ18Oatm, total air content, CO2, CH4, N2O, dust, high-resolution chemistry, ice texture) of the bottom 60 m of the EPICA (European Project for Ice Coring in Antarctica) Dome C ice core from central Antarctica. These bottom layers were subdivided into two distinct facies: the lower 12 m showing visible solid inclusions (basal dispersed ice facies) and the upper 48 m, which we will refer to as the "basal clean ice facies". Some of the data are consistent with a pristine paleoclimatic signal, others show clear anomalies. It is demonstrated that neither large-scale bottom refreezing of subglacial water, nor mixing (be it internal or with a local basal end term from a previous/initial ice sheet configuration) can explain the observed bottom-ice properties. We focus on the high-resolution chemical profiles and on the available remote sensing data on the subglacial topography of the site to propose a mechanism by which relative stretching of the bottom-ice sheet layers is made possible, due to the progressively confining effect of subglacial valley sides. This stress field change, combined with bottom-ice temperature close to the pressure melting point, induces accelerated migration recrystallization, which results in spatial chemical sorting of the impurities, depending on their state (dissolved vs. solid) and if they are involved or not in salt formation. This chemical sorting effect is responsible for the progressive build-up of the visible solid aggregates that therefore mainly originate "from within", and not from incorporation processes of debris from the ice sheet's substrate. We further discuss how the proposed mechanism is compatible with the other ice properties described. We conclude that the paleoclimatic signal is only marginally affected in terms of global ice properties at the bottom of EPICA Dome C, but that the timescale was considerably distorted by mechanical stretching of MIS20 due to the increasing influence of the subglacial topography, a process that might have started well above the bottom ice. A clear paleoclimatic signal can therefore not be inferred from the deeper part of the EPICA Dome C ice core. Our work suggests that the existence of a flat monotonic ice–bedrock interface, extending for several times the ice thickness, would be a crucial factor in choosing a future "oldest ice" drilling location in Antarctica.

A systematic evaluation of the Zr-in-rutile thermometer in ultra-high temperature (UHT) rocks

The Zr-in-rutile geothermometer is potentially a widely applicable tool to estimate peak metamorphic temperatures in rocks from diverse geological settings. In order to evaluate its usefulness and reliability to record and preserve high temperatures in granulite facies rocks, rutile from UHT rocks was investigated to assess different mechanisms of Zr (re-)distribution following cooling from high temperature. Granulite facies paragneisses from the lowermost part of the Ivrea Zone, Italy, incorporated as thin sheets into the extensive basaltic body of the Mafic Complex were selected for this study. The results show that Zr-in-rutile thermometry, if properly applied, is well suited to identify and study UHT terranes as it preserves a record of temperatures up to 1190 °C, although the thermometer is susceptible to partial post-peak metamorphic resetting by Zr diffusion. Texturally homogeneous rutile grains preserve Zr concentrations corresponding to temperatures of prograde rutile growth. Diverse rutile textures and relationships between some rutile host grains and included or adjacent Zr-bearing phases bear testimony to varying mechanisms of partial redistribution and resetting of Zr in rutile during cooling and link Zr-in-rutile temperatures to different steps of the metamorphic evolution. Rutile grains that equilibrated their Zr concentrations at temperatures above 1070 °C (i.e. 1.1 wt% Zr) could not retain all Zr in the rutile structure during cooling and exsolved baddeleyite (ZrO2). By subsequent reaction of baddeleyite exsolution lamellae with SiO2, zircon needles formed before the system finally closed at 650–700 °C without significant net loss of Zr from the whole host rutile grain. By reintegration of zircon exsolution needles, peak metamorphic temperatures of up to 1190 °C are derived for the studied rocks, which demonstrates the suitability of this solution thermometer to record UHT conditions and also confirms the extraordinary geological setting of the lowermost part of the Ivrea Zone.

Development and Light Response of Leaves of Metasequoia and Close Relatives

Metasequoia glyptostroboides is a useful nearest living relative (NLR) of the Eocene fossil Metasequoia. Research on modern Metasequoia might give us some clues about its fossil counterpart. During this study the leaf anatomy of Metasequoia, Glyptostrobus, Sequoia and Taxodium was investigated with light microscopy and transmission electron microscopy. Metasequoia exhibits several characteristics of typical sciaphilic plants, such as slightly arched outer cell walls in the adaxial epidermal cells, strongly arched outer cell walls in the abaxial epidermal cells, mesophyll composed of spongy cells, chloroplasts with well-developed grana not only in mesophyll cells but in both the adaxial and abaxial epidermis. Based on comparison of leaf morphology and anatomy, we conclude that Metasequoia is best adapted to low light intensities, Sequoia and Taxodium are intermediate, and Glyptostrobus is adapted to higher light intensities. The effects of light intensity on mesophyll plastids of Metasequoia leaves were studied with trees grown under different light intensities. Metasequoia had the ability to synthesize chlorophyll under complete darkness and was stressed under high light. These characteristics would provide adaptive advantages for Metasequoia to adapt to low intensity, low angle, polar light at their Eocene high latitude paleoenvironments, particularly during the polar spring when light levels are exceedingly low. It provides evidence to explain why Metasequoia was the dominant tree species in Eocene high latitudes. The thesis is written as an article to be submitted to the American Journal of Botany.

Geochemical and palynological results of IODP Hole 302-M0004A from Lomonosov Ridge, Arctic Ocean

Several episodes of abrupt and transient warming, each lasting between 50,000 and 200,000 years, punctuated the long-term warming during the Late Palaeocene and Early Eocene (58 to 51 Myr ago) epochs**1,2. These hyperthermal events, such as the Eocene Thermal Maximum 2 (ETM2) that took place about 53.5 Myr ago**2, are associated with rapid increases in atmospheric CO2 content. However, the impacts of most events are documented only locally**3,4. Here we show, on the basis of estimates from the TEX86' proxy, that sea surface temperatures rose by 3-5 °C in the Arctic Ocean during the ETM2. Dinoflagellate fossils demonstrate a concomitant freshening and eutrophication of surface waters, which resulted in euxinia in the photic zone. The presence of palm pollen implies**5 that coldest month mean temperatures over the Arctic land masses were no less than 8 °C, in contradiction of model simulations that suggest hyperthermal winter temperatures were below freezing**6. In light of our reconstructed temperature and hydrologic trends, we conclude that the temperature and hydrographic responses to abruptly increased atmospheric CO2 concentrations were similar for the ETM2 and the better-described Palaeocene-Eocene Thermal Maximum**7,8, 55.5 Myr ago.

(Table 1) Summary of general lithology in IODP Exp302 cores

The Cenozoic ice-rafted debris (IRD) history of the central Arctic is reconstructed utilizing the terrigenous coarse sand fraction in IODP 302 cores from 0 to 273 meters composite depth. This Holocene - middle Eocene quantitative record of terrigenous sand accumulation on the Lomonosov Ridge, along with qualitative information on grain texture and composition, confirms the interpretation that ice initiation (sea ice and glacial ice) occurred ~46 Ma in the Arctic, and provides a long-term pattern of Arctic ice expansion and decay since the middle Eocene. IRD mass accumulation rates range from 0 to 0.13 g/cm2/ka in the middle Eocene and from 0 to 0.36 g/cm2/ka in the Neogene. IRD mass accumulation rate (MAR) maxima in the Miocene and Pliocene cooccur with either glacial initiation or intensification in the sub-Arctic. The 46.25 Ma IRD onset in the central Arctic slightly precedes the earliest evidence of ice in the Antarctic, and compares in timing with a >1000 ppm decrease in atmospheric concentrations of CO2. The decline of pCO2 in the middle Eocene may have driven both poles across the temperature threshold that enabled the nucleation of glaciers on land and partial freezing of the surface Arctic Ocean, especially during times of low insolation.