Evidence for Extreme Climatic Warmth from Late Cretaceous Arctic Vertebrates

See allHide authors and affiliations

Science  18 Dec 1998:
Vol. 282, Issue 5397, pp. 2241-2243
DOI: 10.1126/science.282.5397.2241


A Late Cretaceous (92 to 86 million years ago) vertebrate assemblage from the high Canadian Arctic (Axel Heiberg Island) implies that polar climates were warm (mean annual temperature exceeding 14°C) rather than near freezing. The assemblage includes large (2.4 meters long) champsosaurs, which are extinct crocodilelike reptiles. Magmatism at six large igneous provinces at this time suggests that volcanic carbon dioxide emissions helped cause the global warmth.

The Cretaceous is commonly considered to have been ice-free with high atmospheric CO2 levels (1–3), but some isotopic and paleofloral evidence has implied that polar temperatures were near freezing (4–6). Here we describe a fossil vertebrate assemblage from the high Canadian Arctic that supports a Late Cretaceous [92 to 86 million years ago (Ma)] thermal maximum.

The Cretaceous of the Canadian Arctic is represented by sedimentary and volcanic rocks of the Sverdrup Basin (7), which are exceptionally well exposed on western Axel Heiberg Island (Fig. 1). The youngest rocks in the area are Late Cretaceous to Eocene sedimentary rocks of the Eureka Sound Group. Shallow marine to continental shale, siltstone, and sandstone are underlain by Late Cretaceous marine shale of the Kanguk Formation. On much of western Axel Heiberg Island, the Kanguk Formation unconformably overlies subaerially erupted flood basalts of the Cretaceous Strand Fiord Formation. These lavas are part of a large magmatic pulse, or large igneous province, that may include large parts of Ellesmere Island and the Arctic Ocean basin (8).

Figure 1

(A) Location of the Late Cretaceous vertebrate locality in the Canadian Arctic archipelago (triangle). Sea ice is shown in a stippled pattern. (B) Paleolatitude lines (dashed) using 90-Ma pole for North America (25). (C) Stratigraphic column showing vertebrate-bearing sedimentary layers from Expedition Fiord. Numerical ages for the basal Kanguk Formation are based on published time scales (10). The age for Cretaceous Arctic magmatism is based on new 40Ar/39Ar radiometric age data from the Strand Fiord Formation (this work) and Pb/U radiometric age data from gabbroic intrusions on Ellesmere Island (14).

Near Expedition Fiord (79°23.5′N, 92° 10.9′W), sedimentary rocks record the transition between the Strand Fiord lavas and Kanguk shale. The uppermost flow of the Strand Fiord Formation is overlain by 0.6 m of weathered basalt and soil (Fig. 1). The soil is overlain by approximately 3.0 m of shale and siltstone, which may represent a bay or estuary. We found well-preserved vertebrate fossils in several of the thin siltstone horizons in this sequence. Although disarticulated, related bones were in close proximity, suggesting limited transport.

Recent magnetostratigraphic study (9) at Strand Fiord (Fig. 1) suggests that the base of the Kanguk Formation is older than 83.5 Ma (geomagnetic chron 33R) (10). Ammonites of theScaphites depressus Zone indicate that the Kanguk Formation 194 m above its base at Glacier Fiord (11) is of late Coniacian age (∼86.0 to 87.0 Ma). Ammonites suggest that the basal Kanguk Formation is late early Turonian in age (∼92.0 to 91.0 Ma) on Amund Ringes Island (7).40Ar/39Ar incremental heating (12) of a whole-rock sample from the upper lava flows at Strand Fiord (9) has yielded an 11-step plateau (Fig. 2). These data indicate an age of 95.3 ± 0.2 Ma near the Cenomanian-Turonian boundary (13). U/Pb zircon analyses of gabbroic intrusions on northwestern Ellesmere Island, thought to be part of the same magmatic event as that represented by the Strand Fiord Formation, yield an age of 92.0 ± 1.0 Ma (early Turonian) (14). Together these data indicate that the vertebrate fossil assemblage is Turonian to Coniacian (∼92 to 86 Ma) in age.

Figure 2

40Ar/39Ar apparent age spectrum from whole-rock basalt sample from the Strand Fiord Formation.39Ar recoil is suggested by the discordance pattern at low temperature steps. The higher temperature data define a high-resolution plateau age, which is shown with the analytical error (1σ). The total systematic error is 0.9 Ma, including analytical errors, uncertainties in the age of the standard (28.03 Ma, Fish Canyon sanidine), and uncertainties in decay constants (12).

The fossils represent a diverse assemblage of nonmarine aquatic and semiaquatic vertebrates (Fig. 3), including fish, turtles, and champsosaurs. At least two types of fish are represented by scales similar to those described as Holostean A and Holostean B from Upper Cretaceous nonmarine sediments (15). Turtles and champsosaurs offer several advantages over other fossils used as climatic indicators, such as the latest Cretaceous dinosaurs of the North Slope, Alaska, because they are free from ambiguities posed by possible migration and warm-bloodedness (16,17). Turtles are represented by costals and peripherals that are comparable to shell elements of generalized aquatic cryptodires. Extant aquatic nonmarine turtles are ectothermic reptiles and have a climatically limited distribution. The length and warmth of summers limit turtle distributions, primarily by affecting the survival of eggs and hatchlings. The cold-adapted turtles Chelydra serpentina and Chrysemys picta provide a conservative estimate of the growing season required (18). Viable populations of these taxa are restricted to areas where the growing season has at least 100 frost-free days per year (19).

Figure 3

Late Cretaceous vertebrate fossils from Axel Heiberg Island. (A) through (F) are from champsosaurs. (A) Tibia; (B and C) femurs; (D) ischium; (E) same as (D), side view; (F) rib; (G) turtle peripheral bones; (H) turtle peripheral showing sulci; (I) champsosaur dorsal centrum; (J) same as (I), side view; (K) champsosaur second cervical centrum; (L) same as (K), side view.

Maximum temperatures during the warmest month of the year also provide a measure of the climatic requirements of these cold-adapted turtles. Naturally occurring viable populations of Chelydra serpentina and Chrysemys picta do not occur in areas with a warm-month average maximum temperature of less than 25°C (18, 19). This measure corresponds to a warm-month mean temperature of 17.5°C and a mean annual temperature of 2.5°C. Thus, by analogy, the turtles in the Late Cretaceous Axel Heiberg locality indicate that the mean annual temperature was at least 2°C, the warm-month average maximum temperature was at least 25°C, and the climate was frost-free for more than 100 days per year.

The Axel Heiberg vertebrate assemblage differs from others known from the Upper Cretaceous of Arctic North America in the abundance of semiaquatic reptiles (20) and in the presence of champsosaurs. Champsosaurs, which are thought to have been active semiaquatic predators (21), are represented by a tibia, a mandible, an ulna, femurs, ribs, gastralia, ischia, and centra (Fig. 3). The mandible fragment indicates that the snout was long and slender, comparable to that of the genus Champsosaurus from Upper Cretaceous and Paleocene rocks at lower latitudes elsewhere in North America. A substantial size range was present. A complete tibia allows us to estimate the length of one of the larger individuals by comparison with published data (22) and a similarly sized specimen in the collection of the Royal Tyrrell Museum of Paleontology (specimen RTMP 86.12.11). On the basis of this element, the length of champsosaurs from the Arctic locality reached at least 2.4 m.

As in extant ectothermic reptiles, temperature would have been a primary control on the distribution of champsosaurs. Tolerances can be hypothesized from the temperature limits of the extant reptiles that phylogenetically bracket champsosaurs. Recent analyses place champsosaurs in a primitive position in the Archosauromorpha (23), so the group is bracketed by crocodilians and lepidosauromorphs among living reptiles; the closest living relatives are crocodiles. Crocodiles are also the closest modern analogs in terms of body proportions, size, and mode of life. The thermal limit for viable populations of crocodiles is marked by a coldest-month mean temperature (24) of ∼5.5°C. The preferred operating temperature of crocodiles is 25° to 35°C, and this temperature is maintained for sufficient duration in areas with a minimum mean annual temperature greater than 14°C.

In lepidosauromorphs, the ability to tolerate climate extremes is size-related. Large lepidosaurs are unable to escape subcritical temperatures by behavioral or physiological means (24). Extant lepidosauromorphs with a body size comparable to the Arctic champsosaurs reported here, such as varanids and large iguanids, are more restricted in their temperature tolerances than are crocodiles. Thus, based on tolerances in the extant reptiles that phylogenetically bracket champsosaurs and are comparable in size, the most conservative estimate of the temperature tolerance of champsosaurs is provided by crocodilians, and these suggest that the mean annual temperature in this region was greater than 14°C. An inherent uncertainty is associated with this temperature estimate because champsosaurs are extinct. The discrepancy between the climate implied by the overall fossil assemblage and one where freezing conditions would be common is nevertheless large.

On the basis of global paleomagnetic data (25), our new fossil locality was at a paleolatitude of 72° ± 4°N (Fig. 1). Potential tectonic motions within the Canadian Arctic allow for slightly lower or higher values (26) beyond the 95% confidence interval quoted. However, a paleolatitude above the Late Cretaceous Arctic circle appears most likely for our site.

Turonian fossil flora from Kamchatka (3) (paleolatitude ∼70°) (25) suggest that the mean annual temperature was at least 7°C there and that the cold-month mean temperature was −4°C. Turonian flora from Novaya Sibir (3) (Fig. 1) in the Russian Arctic (paleolatitude ∼78°) (25) yield a mean annual temperature of 9°C and a cold-month mean temperature of 0°C. Because of the lowered metabolic and reproductive rates, it is doubtful that viable populations of large-bodied, active, ectothermic reptiles could be maintained under the seasonal freezing conditions implied by these monthly average temperature estimates. The fossil floral sites are adjacent to oceans, so the discrepancy between these cooler estimates and those implied by the Axel Heiberg fossil reptiles cannot be due to a continental climate gradient. Determining whether the difference reflects distance from a source of warm water currents, such as the Western Interior Seaway (3), must await results from additional sites. A similar discrepancy between temperature estimates based on flora and vertebrates has been noted for the early Cenozoic (27). However, the differences might reflect age differences [1 to 2 million years (My)] between the new Arctic vertebrate data and the fossil flora sites.

An increased flux of volcanic CO2 has often been offered as a mechanism for driving Cretaceous greenhouse warming, but only recently has a detailed temporal picture of volcanism become available. In addition to volcanism in the Arctic, basaltic volcanism occurred at five large igneous provinces during Turonian-Coniacian times, including (Fig. 4) emplacement of the Caribbean Oceanic Plateau (28), eruption of the Madagascar flood basalts (29), volcanism at Broken Ridge (30), emplacement of large parts of the Rio Grande Rise (31), and renewed volcanism on the Ontong Java Plateau (32). The Late Cretaceous also saw the emplacement of kimberlites in South Africa (33) and of alkalic rocks in the southern United States (34). Together these events define a restricted interval (<7 My) of extraordinary global magmatism. If the short-term (1000- to 100,000-year) effusion rates at these large igneous provinces were many times those averaged over several million years (35), CO2 input to the atmosphere could have stimulated greenhouse conditions and the warmth implied at our Arctic site.

Figure 4

Late Cretaceous large igneous provinces, kimberlites, and alkalic intrusions. (A) High Arctic large igneous province (8) (this paper). (B) North American alkalic intrusions (34). (C) Caribbean Oceanic Plateau (28). (D) Rio Grande Rise (31). (E) South African Group II kimberlites (33). (F) Madagascar flood basalts and possibly coeval oceanic flood basalts (29). (G) Broken Ridge (30). (H) Late Cretaceous Ontong Java Plateau volcanism (32).

The presence of reptiles at Arctic latitudes offers challenges for efforts to model Cretaceous climates. The high polar temperatures implied here exacerbate the problems of simulating warm polar conditions without also raising equatorial temperatures to unreasonably high values (36). The warm equable climate that is often associated with the Cretaceous probably did not characterize the entire period (17). Some data suggest relatively cool climates in the Early Cretaceous (37). Nevertheless, the Arctic vertebrates and coeval global volcanism suggest that the Greenhouse Earth analog (1) may be found in the Turonian-Coniacian time interval.

  • * To whom correspondence should be addressed. E-mail: john{at}


View Abstract

Navigate This Article