Northward dispersal of dinosaurs from Gondwana to Greenland at the mid-Norian (215 – 212 Ma, Late Triassic) dip in atmospheric p CO 2

The earliest dinosaurs (theropods and sauropodomorphs) are found in fossiliferous early Late Triassic strata dated to about 230 million years ago (Ma), mainly in northwestern Argentina and southern Brazil in the Southern Hemisphere temperate belt of what was Gondwana in Pangea. Sauropodomorphs, which are not known for the entire Triassic in then tropical North America, eventually appear 15 million years later in the Northern Hemi- sphere temperate belt of Laurasia. The Pangea supercontinent was traversable in principle by terrestrial vertebrates, so the main barrier to be surmounted for dispersal between hemispheres was likely to be climatic; in particular, the intense aridity of tropical desert belts and unstable climate in the equatorial humid belt accompanying high atmospheric p CO 2 that characterized the Late Triassic. We revisited the chronostratigraphy of the dinosaur-bearing Fleming Fjord Group of central East Greenland and, with additional data, produced a correlation of a detailed magnetostratigraphy from more than 325 m of composite section from two field areas to the age-calibrated astrochronostratigraphic polarity time scale. This age model places the earliest occurrence of sauro- podomorphs ( Plateosaurus ) in their northernmost range to ∼ 214 Ma. The timing is within the 215 to 212 Ma (mid-Norian) window of a major, robust dip in atmospheric p CO 2 of uncertain origin but which may have resulted in sufficiently lowered climate barriers that facilitated the initial major dispersal of the herbivorous sauropodomorphs to the temperate belt of the Northern Hemisphere. Indications are that carnivorous

The earliest dinosaurs (theropods and sauropodomorphs) are found in fossiliferous early Late Triassic strata dated to about 230 million years ago (Ma), mainly in northwestern Argentina and southern Brazil in the Southern Hemisphere temperate belt of what was Gondwana in Pangea. Sauropodomorphs, which are not known for the entire Triassic in then tropical North America, eventually appear 15 million years later in the Northern Hemisphere temperate belt of Laurasia. The Pangea supercontinent was traversable in principle by terrestrial vertebrates, so the main barrier to be surmounted for dispersal between hemispheres was likely to be climatic; in particular, the intense aridity of tropical desert belts and unstable climate in the equatorial humid belt accompanying high atmospheric pCO 2 that characterized the Late Triassic. We revisited the chronostratigraphy of the dinosaurbearing Fleming Fjord Group of central East Greenland and, with additional data, produced a correlation of a detailed magnetostratigraphy from more than 325 m of composite section from two field areas to the age-calibrated astrochronostratigraphic polarity time scale. This age model places the earliest occurrence of sauropodomorphs (Plateosaurus) in their northernmost range to ∼214 Ma. The timing is within the 215 to 212 Ma (mid-Norian) window of a major, robust dip in atmospheric pCO 2 of uncertain origin but which may have resulted in sufficiently lowered climate barriers that facilitated the initial major dispersal of the herbivorous sauropodomorphs to the temperate belt of the Northern Hemisphere. Indications are that carnivorous theropods may have had dispersals that were less subject to the same climate constraints. magnetostratigraphy | Triassic | dinosaurs | Pangea | paleoclimate S ome of the earliest documented occurrences of true dinosaurs are in the Ischigualasto Formation of northwestern Argentina (1,2) and the Santa Maria Formation of southern Brazil (3) that were part of Gondwana and are of late Carnian age as indicated by high precision U-Pb zircon dates of 229 to 233 Ma (4,5). These early Late Triassic dinosaurs occur at paleolatitudes of about 50°S (6). Gondwana and the northern continent assembly of Laurasia constituted most of the world landmasses in the supercontinent of Pangea, whose contiguous extent should have posed no obvious physical barriers to latitudinal dispersal of land vertebrates. Nonetheless, the earliest documented dinosaur occurrences in Laurasia (7) are typically attributed to the succeeding Norian Stage, as in the Chinle Formation of the North American Southwest, the Keuper Group in Germany, and the Fleming Fjord Group of central East Greenland, which is the northernmost occurrence of dinosaurs in the Late Triassic at ∼43°N (8) (Fig. 1).
The chronostratigraphy of the Chinle Formation has been significantly improved recently with integrated results from the Colorado Plateau Coring Project (9)(10)(11). These data strongly validate the Newark-Hartford astrochronostratigraphic polarity time scale [APTS; (12)] for synchronizing Late Triassic stratigraphic sequences and their fossil assemblages around the globe.
Here, we present an updated chronology for the Fleming Fjord Group of Greenland based on a magnetostratigraphy (13) expanded by data from a second sampling area, which necessitates a revised older-age assignment for the documented fossil occurrences of plateosaurid Plateosaurus (14)(15)(16). Together, with an assessment linked to the APTS of the age of the Löwenstein Formation in the Germanic basin that has the earliest occurrences of Plateosaurus in Europe (17), we are able to chart a more precise temporal pattern of the dispersal of early sauropodomorph dinosaurs that can be compared to well-dated records of atmospheric partial pressure of carbon dioxide (pCO 2 ) concentrations to explore climate change as a contributing factor.

Results
Magnetostratigraphy of Fleming Fjord Group Sections. The Fleming Fjord Group [elevated in rank from Formation (18)] is composed of fluvial and lacustrine sediments about 350 m thick, subdivided in upward succession into the Edderfugledal, Malmros Klint, and Ørsted Dal formations [elevated in rank from members (18)]. The Fleming Fjord Group overlies, in apparent conformity, the Gipsdalen Group, but the overlying Kap Stewart Group, which records the end-Triassic event (ETE) (19), is likely to be disconformable with an intervening break in sedimentation (20).
Paleomagnetic sampling of outcrop sections at Tait Bjerg ( Fig. 2) was included in the summer 1992 field season to help constrain the age of the rich vertebrate fauna (14). A magnetostratigraphy with seven magnetozones from F1n at the base up to F4n were delineated Significance Sharply contrasting climate zonations under high atmospheric pCO 2 conditions can exert significant obstacles to the dispersal of land vertebrates across a supercontinent. This is argued to be the case in the Triassic for herbivorous sauropodomorph dinosaurs, which were confined to their initial venue in the Southern Hemisphere temperate belt of Pangea for about their first 15 million years. Sauropodomorphs only appear in the fossil record of the Northern Hemisphere temperate belt about 214 million years ago based on a composite magnetostratigraphy of the Fleming Fjord Group in East Greenland. The coincidence in timing within a major dip in atmospheric pCO 2 from published paleosol records suggests the dispersal was related to a concomitant attenuation of climate barriers in a greenhouse world.
in a ∼210-m outcrop section of the Malmros Klint and the lower part of the Ørsted Dal formations (13) (Fig. 3). Paleomagnetic samples were also collected during the summer 1995 field season of fossil hunting at "Track Mountain" near MacKnight Bjerg (Fig. 2) from a ∼100-m-thick section of the Carlsberg Fjord and Tait Bjerg members of the Ørsted Dal Formation and about a 40-m section of the upper Edderfugledal and lowermost Malmros Klint formations. The samples were processed in the same manner as described for the summer 1992 field season samples, featuring comprehensive progressive thermal demagnetization to isolate a characteristic magnetization from a pervasive Cenozoic overprint (SI Appendix). The "Track Mountain" data were thus far only used (in conjunction with the Tait Bjerg results) for a statistical analysis of Late Triassic paleolatitudes (8). Based on the overall lithostratigraphy, the newly realized polarity magnetozones in the "Track Mountain" section extend from the top of F2r in the lower Carlsberg Fjord Member to F5n in the upper part of the Tait Bjerg Member, whereas magnetozones F0r and F1n could be  Table S3). The match of magnetozones between the overlapping portions of the sections is unequivocal, giving us confidence that the polarity sequence is correctly delineated and allowing us to extend the magnetostratigraphy upward from the upper Carlsberg Fjord Member into the Tait Bjerg Member of the Ørsted Dal Formation as well as downward from the lowermost Malmros Klint into the Edderfulgedal formations (Fig. 3). A total of 12 magnetozones are identified, although 3 (F0r, F4r, and F5n) more tentatively by only single sample sites because of less favorable magnetic recording properties in the more mauve-colored Edderfulgedal Formation and Tait Bjerg Member.
Correlation of Magnetostratigraphy to APTS. A critical age constraint in any viable attempt to correlate the magnetostratigraphic pattern to the APTS is the ETE, which has long been associated with the Kap Stewart Group (formerly Formation) on the basis of palynology and fossil megaflora (19,21,22). There is a marked facies change between the lacustrine gray mudstones and dolomitic marlstones and limestones of the Tait Bjerg Member of the Ørsted Dal Formation and the overlying black mudstones and coarse-grained and pebbly channel and sheet sandstones of the Kap Stewart Group, which has been interpreted to reflect an erosional unconformity at some localities (18,20,23). The abrupt megafloral turnover within the Kap Stewart Group also tends to be ascribed to a hiatus in sedimentation (21), although this view has been strongly disputed by others who maintain that strata containing the last occurrence of Lepidopteris designate the ETE in Greenland (19). We assume the ETE level is within the Kap Stewart Group whose base could nevertheless be associated with an unconformity. The age of the ETE is well calibrated by U-Pb dating at 201.6 Ma (24) and was used to anchor the APTS (12).
The age of the underlying Fleming Fjord Group, especially the Malmros Klint and Ørsted Dal formations, has been loosely attributed to the late Norian-Rhaetian based on vertebrate fossil assemblages (18). For numerical reference, the Norian/Rhaetian boundary has been placed at 205.5 Ma based on U-Pb dating of a marine bivalve biostratigraphy in Peru (25) or at 209.5 Ma based on magnetostratigraphic correlation of a proposed boundary stratotype section with a conodont zonation in Austria to the APTS (26) via the Pizzo Mondello marine section in Sicily (27). There seems to be better agreement on the definition and age of the Carnian/Norian boundary, which is placed at about 227 Ma based on correlation of the ammonoid and conodont-bearing Pizzo Mondello section to the APTS (27). The available data thus suggests the Rhaetian is ∼4 or 8 My long [the so-called "short-" and "long-Rhaetian" options (28)], and the Norian is, in complement, 21.5 to 17.5 My long.
For purposes of correlation, the APTS for the Triassic is assumed to be complete. It was constructed from data of cores with virtually complete recovery of the kilometers-thick Newark basin  Table S1). The rock units are according to ref. 18; stratigraphic occurrences of fossil dinosaurs are indicated by (S, sauropodomorph Plateosaurus) and (T, theropod) (14)(15)(16)(56)(57)(58). The rVGP latitude is the rotated latitude of the calculated virtual geomagnetic pole for the sample site C component direction compared to the mean overall paleomagnetic pole; the rVGP latitudes approaching +90°and −90°signify normal and reverse polarity, respectively. The ticks on the 0°axis are sampling levels that did not provide acceptable paleomagnetic data. The open and closed shading for polarity magnetozones are for reverse (suffix r) and normal (suffix n) polarity intervals that are labeled upward from F0r to F5n. The horizontal axis is APTS (12) showing polarity chrons, maxima in 405-ky eccentricity cycles (Ecc405) numbered from most recent maxima at 0.216 Ma and geologic ages in Ma. The correlation options labeled A, B, and C (SI Appendix, Table S2) are alternative links of the polarity magnetozone sequence to the APTS, all initially keyed to possible counterparts to long magnetozone F2r and within the younger age constraint of 201. 6  section with sediment accumulation rates of ∼150 m/My and higher; moreover, significant portions of the polarity sequence have been successfully replicated in various sections regionally [e.g., Dan River Basin (29)] and farther afield, especially the Colorado Plateau where the temporal astronomical pacing assumed in the APTS has been verified by high-precision U-Pb dating (9,10). The polarity interval lengths conform to a Poisson distribution and average around 0.5 My [∼2 reversals per My; (30)]. The presence of 12 polarity magnetozones in the Fleming Fjord Group would thus imply an overall duration of roughly 6 My. With these general constraints in mind, cross-plots of a range of correlation options (from older to younger: A, B, and C) between the Fleming Fjord Group composite magnetostratigraphy and the APTS are shown in Fig. 3 and listed in SI Appendix, Table  S4. The three correlation options are keyed to thick magnetozone F2r, which at 86.35 m represents about one-quarter of the sampled 325-m thick composite magnetostratigraphy. In Option A, magnetozone F2r is correlated to 1.48-My-long Chron E14r, which would imply a sediment accumulation rate of 58.3 m/My for this interval. If the adjoining magnetozones are correlated to the APTS, magnetozones F1n-F1r-F2n stratigraphically below F2r correspond to chrons E13n-E13r-E14n, with only short Subchron E13n.1r apparently missing in F1n, and magnetozones F3n-F3r-F4n-F4r-F5n immediately above F2r correspond to chrons E15n-E15r-E16n-E16r-E17n with high fidelity, such that even short magnetozone F3n.1r has a plausible counterpart in Subchron E15r.1r. However, this fit requires long-term variations in sediment accumulation rates of about a factor of two: 24.9 m/My stratigraphically below, 58.3 m/My within, and 27.1 m/My above magnetozone F2r.
In Option B, which had been the favored original correlation (13), magnetozone F2r is correlated to 1.39-My-long Chron E17r, which would imply a sediment accumulation rate of 62.1 m/My for this interval, not that different from Option A. However, difficulties emerge with Option B in that relatively thin (25-m-thick) magnetozone F1n is correlated to a 1.8-My-long Chron E16n, implying a low sediment accumulation rate of only 13.9 m/My, whereas a 47.5-m-thick magnetozone F4n is correlated to only a 0.15-My-long Chron E19n, implying a contrastingly high sediment accumulation rate of 316.7 m/My that is more than a factor of 5 higher than for magnetozone F2r. Also bothersome with this correlation is that there is no obvious counterpart in the APTS to thin magnetozone F3n.1r.
In Option C, the youngest of the correlation options considered, magnetozone F2r is correlated to 1.59-My-long Chron E20r, which would imply a sediment accumulation rate of 54.3 m/My for this interval. A distinguishing feature of this correlation scheme is that it could be accommodated by much the same rate of sediment accumulation over the entire section: 47. However, the appeal of a fairly uniform sediment accumulation rate for the entire section with Option C comes at the expense of requiring a low fidelity polarity record in the lower part. This weakness in the magnetostratigraphic correlation makes us disinclined to accept Option C as providing a valid age model. In contrast, Option B has high magnetostratigraphic fidelity with essentially all polarity chrons accounted for but requires huge swings in sediment accumulation rate to correlate the magnetozones in the lower and upper parts of the section. Option B was initially preferred (13) but has become less tenable with the downward and upward extension of the original magnetostratigraphy. We therefore explore the implications of now-favored Option A, which has a good fidelity magnetostratigraphic record with longterm changes in sediment accumulation rate that are large but seem compatible with changes in lithology (e.g., higher rate in the most sand-rich facies in the Malmros Klint Formation).
For The upward extrapolation to only 209 Ma also implies that the Fleming Fjord Group is of mostly Norian age, irrespective of the "short" or "long" Rhaetian time scale proposals. Extrapolating the sediment accumulation rate of 24.9 m/My for magnetozones F2n to F1n downward to the contact 47 m below with the underlying Kap Seaforth Formation of the Gipsdalen Group would add 1.89 My and give an age of ∼220 Ma, or early Norian, for the base of the Fleming Fjord Group. This is somewhat younger but not grossly inconsistent with an age range of 226 to 235 Ma (Carnian-early Norian) assigned to the Gipsdalen Group on the basis of rather uncertain biostratigraphic constraints (32). Stratigraphically below the Gipsdalen Group, the continental Pingo Dal Group is also notably lacking in age-diagnostic fossils (32). Only the underlying Wordie Creek Group with a rich ammonoid fauna has had a firm age assignment for the Triassic of East Greenland [Induan, Early Triassic (33)]; its marine deposition was followed by syn-rift alluvial progradation and postrift continental deposition, which dominated the rest of the Triassic in the region (18,34).  (5), and the Ischigualasto Formation of northwestern Argentina [e.g., early theropod Herrerasaurus (1, 37, 38)] dated to ∼230 Ma (4), which is overlain by the Los Colorados Formation, a ∼600-m-thick succession of continental deposits spanning 227 to 212.5 Ma according to magnetochronology (6) and containing a vertebrate assemblage in its upper part (La Esquina fauna, correlated to magnetozones LC7r and LC8n = chrons E14r and E15n of the APTS, 215 to 212.5 Ma) that is rich in dinosaurs (2) such as, sauropodomorph Coloradisaurus (39). Recently published U-Pb dates from the lower Elliot Formation in the Karoo basin of southern Africa support temporal correlations of some of its fauna, including the sauropodomorph Plateosauravus, to that of the Los Colorados Formation (40).
The best-documented, earliest dinosaur occurrences in North America come from the Chinle Formation. Theropods of arguable affinities ["Camposaurus"; (41)] have been reported from the Placerias Quarry in Arizona (42,43) where a U-Pb detrial zircon date of 219.39 ± 0.16 Ma was obtained on the main Bone Bed (44). However, recycled zircons are prevalent in the lower Chinle Formation (45) and may be responsible for apparent age disagreements of several million years; for example, a sample (SBJ) from an outcrop in the Sonsela Member with a highprecision U-Pb detrital zircon date of 219.32 ± 0.26 Ma (46) seems to correspond to age estimates that are 2 to 4 My younger for the Sonsela Member in the PFNP-1A drill core (10,47). Such uncertainties directly affect the dating of the major biotic turnover at the Adamanian-Revueltian transition that is usually placed somewhere within the Sonsela Member (48). Magnetostratigraphic correlations show that the Sonsela Member in the PFNP-1A core encompasses chrons E14n-E15n, spanning 216 to 213 Ma (10), but which would extend back to as much as 220 Ma if U-Pb detrital zircon dates in the lower Chinle Formation (44,47) were to be taken at face value. In any case, a more securely constrained dinosaur fauna of the Chinle Formation is found at Hayden Quarry at Ghost Ranch in New Mexico, where taxa including the early theropod Tawa and Chindesaurus, a possible rare example of a herrerasaurid theropod dinosaur from outside South America [(49); but see (50)] have been described. A U-Pb zircon date in Hayden Quarry of 211.9 ± 0.7 Ma (51) is consistent with the occurrence of a similar fossil assemblage in the Petrified Forest Member of the Chinle Formation in PFNP in Arizona (52), where its age is well constrained between ∼209 and 212.5 Ma by congruent U-Pb zircon geochronology and magnetochronology (9,46).
At higher latitudes in the Northern Hemisphere, the earliest well-dated occurrence of dinosaurs in Europe, in this case the plateosaurid Plateosaurus, is in the Löwenstein Formation of the Keuper Group in the Germanic basin (53,54). This formation also contains the conchostracan Shipingia, whose first appearance is in the oldest part of the Passaic Formation of the Newark sequence (17), which corresponds to Chron E13n with an estimated age of around 217 Ma in the APTS (12). Shipingia is also associated with the Alaunian substage which, on the basis of conodont biostratigraphy and magnetostratigraphy at Silicka Brezova (Slovakia), extends up through Chron E15n (55) and hence provides an age range of ∼217 to 212.5 Ma according to the APTS for the dinosaur-bearing Löwenstein Formation.
At the northernmost known paleolatitude fossil sites, in East Greenland, Plateosaurus bone fossils have been found in a halfdozen sites from at least four different stratigraphic levels in the Malmros Klint Formation (from as low as 25 m below its contact with the Ørsted Dal Formation) and the Carlsberg Fjord Member, where a complete skeleton was found at Lepidopteriselv (Fig. 2) 5 m below its contact with the Tait Bjerg Member (14)(15)(16). There are also preliminary reports of theropod bone fossils in the middle ("Theropod Mound") and lowermost Carlsberg Fjord Member (56,57) and in the middle of the Malmros Klint Formation as far down as 65 m below its top (58). The earliest documented occurrence of Plateosaurus is in the middle part of magnetozone F2r, which, according to the age model of Option A, is within Chron E14r and corresponds to an interpolated age of 214.2 Ma (Fig. 4). The oldest reported theropod site would be near the base of magnetozone F2r, corresponding to a somewhat older age of about 214.8 Ma. These datum levels for oldest dinosaur fossils are about 5 My older than estimated from our superceded age model (Option B, Fig. 3) and would place them in a very different phase of suggested effects of 10-My-scale monsoon dynamics in the Late Triassic (59).

Discussion
From the foregoing, a review of global chronostratigraphic evidence suggests that, after appearing by around 230 Ma in the temperate belt of Gondwana in the Southern Hemisphere (as recorded by fossils in well-dated units like the Ischigualasto Formation of northwestern Argentina and the Santa Maria Formation in southern Brazil), bone fossils of sauropodomorph dinosaurs are not found in strata of the Northern Hemisphere until 217 to 212.5 Ma in the Germanic Basin of Europe (Löwenstein Formation) and more precisely at close to 214 Ma in Jameson Land of central East Greenland (Malmros Klint Formation). The great circle distance across Pangea between Jameson Land and northwestern Argentina was vast, about 95°o f arc or over 10,000 km but with no intervening major seaways or high mountain ranges, a route should have been traversable by land-bound terrestrial vertebrates (Fig. 1). Yet, something evidently held sauropodomorphs back for about 15 My from dispersing widely until around 214 Ma when they show up in the temperate belt (northern Europe, Greenland) of the Northern Hemisphere. Interestingly, there is scant evidence of sauropodomorphs in the entire Triassic of North America (e.g., Chinle and Newark Supergroup) and northern Africa (Argana basin, Morocco), which evidently were not areas to linger being within or close to the tropical arid belt.
Given a tectonically stable supercontinent assembly of Pangea, barriers to dispersal were most likely climatological. The Late Triassic was a time of generally very high atmospheric pCO 2 values according to empirical estimates from pedogenic carbonate barometry (60) that are broadly consistent with carbon cycle models (61) (Fig. 4). Climate modeling studies show that higher atmospheric pCO 2 values are associated with more accentuated contrasts between climate belts, especially in precipitation to evaporation (P-E) (62). The more arid climate belts that ecologically separated the southern temperate belt of Gondwana, where dinosaurs apparently originated, from eventual venues in the hospitable counterpart northern temperate belt of Laurasia (Europe and Greenland) may have presented formidable barriers to dispersal. Moreover, the low latitudes of North America in the interior of Pangea in a high-pCO 2 world would have experienced extreme climate fluctuations that affected plant communities and help to explain the rarity to virtual absence of herbivorous sauropodomorphs (63) and related vertebrates (64) in this near-equatorial setting.
After more or less steadily high pCO 2 values of around 4,000 parts per million (ppm) for the initial half of the Late Triassic (∼233 to 215 Ma), the pedogenic carbonate pCO 2 data show that atmospheric pCO 2 values plunged to around 2,000 ppm at 215 to 212 Ma before increasing again to high values (60) (Fig. 4). This interval of reduced pCO 2 levels has been documented in contemporaneous paleosols from widely separated sites in the Newark, Hartford, and Chama basins (60), including Hayden Quarry at Ghost Ranch in New Mexico with the theropods Tawa and Chindesaurus (63). By synchronizing δ 18 O records of seasurface temperature change to the Newark pedogenic pCO 2 record embedded within the APTS, Knobbe and Schaller (65) demonstrate that the response for a halving of pCO 2 is in good 6 (17) whose age assignment is based on biostratigraphic correlations for Alaunian to marine magnetostratigraphic sections (27,55). Paleolatitude of 37°N at 215 Ma. Column D: The composite magnetostratigraphic section for Fleming Fjord Group of East Greenland using correlation Option A (this study). Stratigraphic levels with sauropodomorph (Plateosaurus) fossil sites are indicated by "S" and theropod fossil sites by "T" (see Fig. 3). Paleolatitude of 43°N at 215 Ma. The color bar on the paleolatitude scale along the bottom shows the relative precipitation minus evaporation (P − E) values as a function of latitude, with a green shade indicating more positive values (wet) and a yellow shade indicating more negative values (arid), based on a generalized general circulation model of the coupled ocean-atmosphere climate system (62 have arrived earlier there (68) as well as in East Greenland (57,58).
The cause of the mid-Norian dip in atmospheric pCO 2 is unclear. Long term monsoonal effects may have played a role, although the suggested timing of the mid-Norian dinosaur dispersal is now a few million years earlier than the ∼212 Ma that had been considered (59). According to correlation networks, the dip in pCO 2 seems to shortly follow a large excursion in δ 13 C recorded in marine carbonates at Pizzo Mondello centered on Chron E14n [∼216 to 215 Ma (69)], again pointing to a major perturbation in the global carbon cycle in the Alaunian interval that might be coupled to the dip in pCO 2 . The 85-km-diameter Manicouagan impact crater in Canada with a preliminary U-Pb zircon date of 215.5 Ma (70) and that had far-field effects (71) provides a possible trigger of cascading events eventually affecting climate, for example, widespread deforestation, erosion, and increased weathering consumption of pCO 2 . The intriguing near-coincidence in timing among the Manicouagan impact, δ 13 C excursion, and dip in atmospheric pCO 2 with our postulated mid-Norian dinosaur dispersal event (and perhaps the Adamanian-Revueltian biotic turnover) requires more precise age registry to determine how these phenomena might be arrayed in a plausible temporal sequence to explore possible modes of causality.

Conclusions
An updated age model for the fossiliferous Fleming Fjord Group of central East Greenland places the first occurrences of sauropodomorph dinosaurs at their northernmost range in Laurasia at about 214 Ma, long after their initial appearance at about 230 Ma at their southernmost range in Gondwana. The Late Triassic tended to have very high atmospheric pCO 2 values, which climate models suggest would result in intensified aridity in the desert belts and extreme swings in humidity and aridity at equatorial latitudes. These prevailing conditions are suggested to have acted as barriers for dispersal that kept herbivorous sauropodomorph dinosaurs corralled in the southern temperate belt of Pangea.
Their apparent breakout by ∼214 Ma occurs within a large reported drop in atmospheric pCO 2 concentration at 215 to 212 Ma, which may have lowered climate barriers and allowed rapid dispersal of sauropodomorphs to the far reaches of Europe and Greenland in the Northern Hemisphere temperate belt. Direct evidence is still lacking of sauropodomorphs in their presumed passage across the low paleolatitudes of North America where theropod dinosaurs are, however, present as a rare component of some Triassic vertebrate assemblages and may even be older than 214 Ma, representing more sporadic dispersals not obviously related to climate thresholds. On the other hand, decreasing atmospheric pCO 2 values over the last ∼8 million years of the Triassic, essentially the Rhaetian, might have resulted in further climateinduced dispersal opportunities for the sauropodomorphs.

Materials and Methods
The same paleomagnetic sampling and analytical procedures described for the 1992 field season at Tait Bjerg (13) were followed for the 1995 sampling campaign at "Track Mountain" (MacKnight Bjerg) as described in the SI Appendix. Site-level data sets (SI Appendix, Tables S1 and S2) were used for the composite magnetostratigraphy of the Fleming Fjord Group shown in Fig. 3.
Data Availability. All study data are included in the article and/or SI Appendix.
ACKNOWLEDGMENTS. We thank our host institutions for patient and generous support of this research, including the Paleomagnetic Research Fund at Lamont-Doherty and for various NSF grants, most recently EAR-0958859 (D.V.K.), to the Carlsberg Foundation (L.B.C.) for logistical support in Greenland, and the late Farish A. Jenkins of the Museum of Comparative Zoology at Harvard University for organizing and providing support for the 1992 and 1995 field seasons and inviting D.V.K. to join. We are grateful to colleagues Paul Olsen, Giovanni Muttoni, and Morgan Schaller for productive, ongoing discussions on wide-ranging Triassic matters and Malte Mau, Octávio Mateus, and Jesper Milàn on the Greenland record, in particular. We greatly appreciate the thoughtful and constructive comments by the journal reviewers, which allowed us to improve the manuscript. This is Lamont-Doherty Earth Observatory Contribution 8469.