Fossil Plants and Global Warming at the Triassic-Jurassic Boundary

See allHide authors and affiliations

Science  27 Aug 1999:
Vol. 285, Issue 5432, pp. 1386-1390
DOI: 10.1126/science.285.5432.1386


The Triassic-Jurassic boundary marks a major faunal mass extinction, but records of accompanying environmental changes are limited. Paleobotanical evidence indicates a fourfold increase in atmospheric carbon dioxide concentration and suggests an associated 3° to 4°C “greenhouse” warming across the boundary. These environmental conditions are calculated to have raised leaf temperatures above a highly conserved lethal limit, perhaps contributing to the >95 percent species-level turnover of Triassic-Jurassic megaflora.

The end-Triassic mass extinction event [205.7 ± 4 million years ago (Ma)] was the third largest in the Phanerozoic, resulting in the loss of over 30% of marine genera (1) and 50% of tetrapod species (2), >95% turnover of megafloral species (3,4), and marked microfloral turnover in Europe (5) and North America (6). Many causal mechanisms have been suggested (7), but because of the paucity of Triassic-Jurassic (T-J) oceanic sediments suitable for geochemical analyses (7–9), the associated environmental conditions remain poorly characterized and the causal mechanism or mechanisms equivocal. Here we provide a temporally detailed investigation of the atmospheric CO2 and climatic conditions associated with the T-J mass extinctions, from analyses of the ecophysiological characteristics of fossil megafloras preserved in composite terrestrial T-J sections in Jameson Land, East Greenland (10), and Scania, southern Sweden (11).

Evidence from sedimentary facies (12), paleosols (13), sea-level change (14), and flood basalt volcanism (15) all indirectly suggest perturbation of the T-J global C cycle. Therefore, likely atmospheric CO2variations across the T-J boundary were determined from the stomatal density (SD, number of pores per unit area) and stomatal index (SI, proportion of pores expressed as a percentage of epidermal cells) of fossil leaf cuticles from 7 genera from 16 beds in Jameson Land and 11 genera from 13 beds in Scania. SD and SI are inversely related to the ambient CO2 concentration during growth (16, 17) and can be used to reconstruct geological time series of atmospheric CO2 (18,19). The standardized SD and SI records for both localities were obtained and corrected for changes in species composition between individual beds to remove taxonomic bias (20) (Fig. 1).

Figure 1

Mean detrended stomatal densities (SD) and indices (SI) calculated [(as in (20)] from the primary data (Table 1) for (A) Scania, southern Sweden, and (B) Jameson Land, East Greenland. “Mega,” “Micro,” and “Trans” denote megafloral and microflora zonation and the transition flora in Greenland, respectively. Arrows indicate direction of time. The T-J boundary is recorded in Greenland at 50 m (10), and these strata are correlated with the Scanian strata by the first occurrence of Thaumatopteris flora at the base of the transition zone in Greenland and theThaumatopteris-Lepidopteris boundary in Scania. The scales from Greenland (depth of plant-bearing beds of the Cape Stewart formation from the Jameson horizon in the Neills Cliff formation) and Sweden (position of plant-bearing beds or localities from composite sections of the Höganas and Höör Sandstone formations of Scania) are not directly comparable beyond stage level. The succession of the different beds or localities in (A) are taken from (3). 1, Bjuv α; 2, Bjuv 1; 3, Höganäs Lower; 4, Stabbarp Ton 10; 5, Skromberga; 6, Bjuv 2 and 3; 7, Hyllinge; 8, Bosarp; 9, Bjuv 4; 10, Skromberga Upper; 11, Stabbarp N.U.G.; 12, Stabbarp J. Mol; 13, Höganäs Upper; 14, Helsingborg; 15, Palsjö; 16, Munka Tågarp; 17, Rodalsberg; 18, Dompang; 19, Sofiero; and 20, Höör. Dating of the boundary is from (34).

High-resolution records of SD and SI from both sites showed reductions during the middle Rhaetian persisting into the Hettangian and then returning to preexcursion values (Fig. 1). The reductions are consistent with an increase in atmospheric CO2concentration across the T-J boundary, in agreement with inferences from geochemical analyses of fossil soils (13). SD, but not SI, is sensitive to changes in precipitation; however, the presence of coals at both localities throughout the Rhaetian and Hettangian indicates stability in the local precipitation regime (3), and so this explanation for the reduction in SD can be discounted. The very low SI of fossil leaves across the T-J boundary (∼3 to 4) has only previously been observed in Lower Devonian fossil axes (∼2 to 3) (21) that developed in CO2 concentrations of between 2400 and 3000 parts per million (ppm) (22).

Atmospheric CO2 concentrations were calculated from stomatal ratios [a ratio of the SI of the nearest living morphological or ecological (or both) equivalent of the fossil divided by the SI of the fossil taxa] of T-J Ginkgoales and Cycadales (Table 1) (21, 23). These calculations indicated that CO2 increased from 600 to 2100-2400 ppm across the T-J boundary (Table 1, Fig. 2)—a more conservative rise than calculated from two Early Jurassic paleosols (4800 ppm) (13). A fourfold increase in CO2 could have raised mean global temperatures by 3° to 4°C (Fig. 2) (24). The most probable source of this CO2 was extensive T-J basaltic volcanicity associated with the breakup of Pangea, which produced the Central Atlantic Magmatic Province (CAMP, 199 ± 2.4 Ma), covering an estimated 7 × 106 km2 (15). If volcanism was the sole source of the T-J CO2 rise (25), the maximum volume of basalt required to have produced 1500 to 1800 ppm CO2 would be 3.6 × 105 km3(26), which is well within that estimated by Marzoliet al. (15) (2 × 106km3).

Figure 2

Reconstructed atmospheric CO2concentration (ratio to present day) (solid line) calculated [as in (21)] from the stomatal ratios of a subset of species [Ginkgoalean and Cycadalean species only (Table 1)] from the total data set (as in Fig. 1) for (A) Greenland and (B) Sweden compared with model estimates (+) (22) and mean global temperature (Δt, in degrees Celsius) (dashed line) calculated from a CO2-greenhouse formulation (24). RCO2 is the ratio of reconstructed CO2 concentration to a preindustrial value of 300 ppm. Mean stable carbon isotope composition (δ13C, in per mil) of fossil leaves from (C) Greenland and (D) Sweden used to constrain the leaf energy-balance model (Fig. 3) (28). The duration of CO2 and temperature changes in (A) and (B) cannot be directly compared beyond stage level because of difference in scales. Arrows indicate direction of time.

Table 1

Mean primary stomatal density (SD), index (SI), and ratio (SR) data in Jameson Land (per depth, 90 to 15 m) and in Scania (per bed or locality number, 1 to 20).

View this table:

For terrestrial vegetation from Greenland and Sweden (paleolatitude ∼45°) growing in high CO2 (900 to 1200 ppm) and a warm end-Triassic climate (summer temperatures of >30°C) (27), the additional climatic warming may have impaired leaf photosynthetic function, severely reducing carbon uptake. To test this hypothesis, we calculated leaf temperatures from energy-balance considerations using measurements of fossil leaf width, stomatal density and size, and δ13C (28). Two sets of calculations were performed, one with constant typical end-Triassic tropical climate (27) and CO2 (22) and one with variable warming and variable CO2 as reconstructed (Fig. 2). The results indicated that large leaves (>3 to 4 cm) in full sunlight of the upper canopy (that is, Ginkgo) or plants of open habitats could have reached noon temperatures at least 10°C above air temperatures, exceeding the observed heat limit for CO2 uptake of present-day tropical taxa (29). Furthermore, transpirational cooling of leaves would have been much reduced at the T-J boundary because of very low stomatal densities (∼25 to 40 mm−2) reducing stomatal conductance, a feature exacerbated by the high leaf-to-air vapor pressure deficits occurring under high temperatures (30).

The extent of high-temperature injury to leaves across the T-J boundary would be size-dependent, because size influences boundary layer conductance and heat dissipation. Heat loss from leaves can be maximized by reductions in leaf size or increases in leaf dissection (or both) (29). Model simulations (28) indicate that the reconstructed rise in CO2 and temperature (Fig. 2) would have caused size-dependent thermal damage to leaves, influencing the survival and distribution or extinction of large-leaved species at the T-J boundary. This hypothesis was tested by measuring the maximum width of the smallest leaf unit (that is, leaf, leaflet, or leaf dissection widths) in the Greenland flora, for which the sedimentary sequence is known to be more continuous than in Sweden (4). The only species common to both Upper Triassic and Lower Jurassic floral zones (seven species), in a flora of over 200 species, are characterized by highly dissected (Clathopteris meniscoides and Todites geoppertiana) or very narrow leaves (Equisetites muensteri) (Fig. 3). Furthermore, species of families most probably dominant in the upper canopy (such as Ginkgoales), with large entire leaves and the greatest predicted temperature overburden (such as Ginkgoites obovatus), appear to have been replaced by species with more dissected leaves (such as G. fimbriatus), over the initial period of increasing temperature and CO2, and extremely dissected leaves (Czekanowskia and Baeira G. muensteriana), under maximum temperature and CO2conditions (Figs. 2 and 3). A similar pattern of species extinction or replacement occurs within the Anomozamites,Saginopteris, Todites, Dictyophyllym, and Pterophyllum, the percentage reductions in leaf size of the replacement species being greatest among those taxa having greater initial leaf size (for example, reductions of 99, 80, and 60 to 10% were observed across the T-J boundary in genera with leaves or leaflets initially >5, 2 to 3, and 1 to 0.5 cm wide, respectively) (Fig. 3). All new species establishing in the transition zone (10) had highly dissected or narrow leaves (0.1 to 1.5 cm wide) in all depositional environments (31).

Figure 3

(A) Simulated changes in leaf width required to avoid thermal damage to the leaf photosynthetic system from a leaf-scale energy-balance–photosynthetic biochemistry model with (var. clim.) and without (const. clim.) the effects of the reconstructed CO2 and temperature increases from Fig. 2(28). (B) Observations of fossil leaf, leaflet, and dissection widths of 33 species from East Greenland (10). Species 1 to 16 (solid lines) are present only in theLepidopteris and transition zones (Rhaetian), when CO2 and temperature were rising (except 15). Species 17 to 20 (short-dashed lines) were established in the transition zone before the T-J boundary, when CO2 and temperature were near maximum levels. Species 21 to 32 (solid lines) are present only in theThaumatopteris zone (Hettangian), when CO2 and temperature were elevated and then declining (Fig. 2). The leaf width of Equisetites muensteri (33, long-dashed line), one of the only species that occurs in all three floral zones, is notably smaller. 1, Ginkgoites obovata; 2, Pterophyllum hanesianum; 3, Saginopteris serrata; 4,Otozamites sp. A; 5, Saginopteris halli; 6,Dictyophyllum exile; 7, P. schenki; 8,Anomozamites minor; 9, Todites scoresbyensis; 10,G. acosmius; 11, G. fimbriatus; 12, P. xiphipterum; 13, P. pinnatifidum; 14,Cladophlebis scariosa; 15, T. geoppertianus; 16,G. minuta; 17, T. princeps; 18, G. munsteriana; 19, Stenopteris dinosaurensis; 20,Czekanowskia nathorstii; 21, A. hartzi; 22,C. ingens; 23, T. recurvatus; 24,Osmundopsis plectrophora; 25, A. niteda; 26,D. nilssoni; 27, Gleicherites niteda; 28,A. marginatus; 29, G. taeniata; 30, G. hermelini; 31, C. svedbergi; 32, P. subaequale; and 33, Equisetites muensteri.

The assumption of constant CO2 and temperature across the boundary does not correlate with the observed reductions in leaf dimensions (Fig. 3). Furthermore, there is no evidence for aridity (as indicated by no increase in fossil plant δ13C and the presence of coal) or marine incursion (as indicated by the absence of marine fossils) in Greenland, both of which could potentially drive selection for narrower or more dissected leaves. The fact that the observed reductions in leaf width parallel those predicted (Fig. 3) under an fourfold CO2 rise and a 3° to 4°C warming of global climate independently corroborates these reconstructions. We therefore suggest that CAMP-induced “super greenhouse” conditions drove selection to avoid lethal leaf temperatures, resulting in extinctions and distributional changes that contributed to the observed >95% turnover of T-J megafloral species in Europe (32).

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


View Abstract

Stay Connected to Science

Navigate This Article