Late Miocene Atmospheric CO2 Concentrations and the Expansion of C4 Grasses

See allHide authors and affiliations

Science  06 Aug 1999:
Vol. 285, Issue 5429, pp. 876-879
DOI: 10.1126/science.285.5429.876


The global expansion of C4 grasslands in the late Miocene has been attributed to a large-scale decrease in atmospheric carbon dioxide (CO2) concentrations. This triggering mechanism is controversial, in part because of a lack of direct evidence for change in the partial pressure of CO2(pCO2) and because other factors are also important determinants in controlling plant-type distributions. Alkenone-based pCO2 estimates for the late Miocene indicate that pCO2 increased from 14 to 9 million years ago and stabilized at preindustrial values by 9 million years ago. The estimates presented here provide no evidence for major changes in pCO2 during the late Miocene. Thus, C4 plant expansion was likely driven by additional factors, possibly a tectonically related episode of enhanced low-latitude aridity or changes in seasonal precipitation patterns on a global scale (or both).

Instability in Miocene climates is detailed by extensive stable isotope records (1) and is associated with turnovers in marine (2, 3) and terrestrial biota (4), sea-level variability (5), and changes in surface- (2) and deep-water circulation (6). Short- and long-term climatic patterns are thought to reflect changes in CO2 concentrations (7–9) or tectonically driven readjustments in ocean circulation (6, 10). In particular, high-latitude climates gradually warmed during the early to middle Miocene [∼24 to 15 million years ago (Ma)] and then rapidly cooled as the East Antarctic ice sheet expanded during the middle Miocene (11).

Evidence suggests that C4 grasses expanded rapidly during the late Miocene (∼8 to 4 Ma) (7). Characterized by the Hatch-Slack photosynthetic pathway, C4 plants (largely but not exclusively represented by grasses) can internally concentrate CO2 before carbon is fixed by way of the Calvin cycle and subsequently avoid the energetic costs of photorespiration (12, 13). This physiology provides C4plants with a competitive advantage over C3 plants (which lack a CO2-concentrating mechanism) when the ratio of atmospheric CO2 to O2 concentrations is low (12, 14). Furthermore, the ability to increase internal leaf CO2 concentrations allows C4 plants to decrease their stomatal conductance, which effectively increases their water-use efficiency (13). Such adaptations provide an advantage under hot, high-irradiance, water-stressed conditions (12). The distribution of modern C4 grasses on a global scale is most strongly correlated to minimum growing season temperatures, with high minimum temperatures favoring C4 grasses (15).

The above physiological considerations led Cerling et al. (7) to attribute a global expansion of C4 plants in the late Miocene to a decrease in the partial pressure of CO2 (pCO2) and to argue that crossing a critically low CO2/O2threshold triggered an ecological response. Here we provide a record of alkenone-based pCO2 estimates for the late Miocene (10 to 5 Ma) in combination with apCO2 record for the early to middle Miocene (16), and evaluate the role ofpCO2 as a mechanism forcing this ecological change.

The alkenone approach to estimating pCO2uses records of carbon isotopic fractionation during marine photosynthetic carbon fixation (ɛp). ɛp (expressed in per mil) for many marine algae is largely a function of the concentration of aqueous CO2 in the growth medium ([CO2aq]), cellular growth rate (17), and cell geometry (18). The isotopic composition of sedimentary organic carbon derived solely from specific photosynthetic marine organisms is best evaluated by isolating molecular biomarkers. When these biomarkers are unique to a particular group of organisms, one avoids the “noise” resulting from the integration of isotopic signals from an array of photosynthesizers with varying geometries, growth characteristics, and carbon fixation pathways. Long-chained unsaturated ketones (alkenones) represent one such class of biomarkers that are exclusively produced by some haptophyte algae in the modern ocean (19). ɛp records constructed from diunsaturated alkenones (ɛp 37:2) have been used to determine variations in paleocean dynamics and QuaternarypCO2 (20). Recent work, however, has provided an empirical expression of ɛp 37:2 as a function of surface-water [PO4 3−] and [CO2aq] (17). Accordingly, reconstruction of paleo-pCO2, which requires knowledge of past growth rates, can be constrained by estimating surface-water [PO4 3−]. Our approach was to obtain ɛp 37:2 from an oceanographic setting with long-term nutrient-limited conditions, such as those found in oligotrophic regions of mid-ocean gyres and inferred for our sample location. This approach minimizes the effect of growth rate on ɛp 37:2, thereby leaving [CO2aq] as the primary control (16).

We studied samples from Deep Sea Drilling Project (DSDP) site 588 (26°06.7′S; 161°13.6′E; southwest Pacific). The low sedimentation rates (∼2 cm/1000 years), characteristic low organic-carbon contents (<0.1%) (21, 22), and uniform deposition of nannofossil-foraminiferal oozes indicate deposition under oligotrophic water masses similar to conditions that characterize this site today. An age model was developed by linearly interpolating between magnetostratigraphic datums (23).

Our results (Figs. 1 and2) show that pCO2steadily increased from a low at ∼14 Ma [∼180 parts per million by volume (ppmv)] and stabilized at concentrations between 320 and 250 ppmv during the late Miocene (9 Ma). These uniformly lowpCO2 values are consistent with middle to late Miocene alkenone-based pCO2 estimates and trends from other localities (16), as well as late Miocene atmospheric CO2 concentrations estimated from stomatal parameters of fossil oak leaves (24). In addition, other alkenone data suggest that pCO2 decreased to about preindustrial levels near the end of the Oligocene (16). Therefore, assuming that models for C4 versus C3 plant competition are correct, lowpCO2 should have favored C4over C3 floras by the early Miocene. However, because nearly 15 million years elapsed between the onset of lowpCO2 and the major C4 plant expansion, it appears that pCO2 level alone was not a sufficient trigger of the late Miocene event.

Figure 1

(A) δ13C values of heptatriaconta-15E, 22E-dien-2-ones (diunsaturated alkenones) from DSDP site 588 (53). PDB, pee dee belemnite. (B) δ13C values of shallow-dwelling planktonic foraminifera from site 588. (C) δ18O values for shallow-dwelling planktonic foraminifera. (D) ɛp record derived from diunsaturated alkenones (53). ɛp = [(δd + 1000/δp + 1000) − 1] × 103, where δd is the carbon isotopic composition of CO2aq calculated from planktonic foraminifera and δp is the carbon isotopic composition of haptophyte organic matter enriched by 4.2 per mil relative to alkenone δ13C values (54). Points represent the average of values measured for each sample. ○, data from Pagani et al. (16); •, results from this study.

Figure 2

Maximum pCO2 estimates calculated on the basis of the ɛp record of site 588, where ɛp = ɛfb/[CO2aq]. The term b represents the sum of physiological factors, including growth rate (17) and cell geometry (18), that affect total carbon isotope discrimination. In the modern ocean, b is highly correlated to surface-water [PO4 3−] (17).pCO2 values represented by the right edge of the shaded band are calculated with a value of 27 per mil for ɛf (the carbon isotope fractionation due to Rubisco), [PO4 3−] = 0.3 μmol/l, and an equation for the physiological-dependent term b calculated with the upper 95% confidence limit from the global data set derived from all available data (17, 55) (b (27 per mil) = 4.35 × [PO4 3−]2 + 125.65 × [PO4 3−] + 108.89). Values on the left edge of the shaded band are calculated with a value of 25 per mil for ɛf, [PO4 3−] = 0.3 μmol/l, and an equation for the physiological-dependent term b calculated with the upper 95% confidence limit from the global data set (b (25 per mil) = 4.17 × [PO4 3−]2 + 113.79 × [PO4 3−] + 88.63). The dashed line represents pCO2estimates calculated with an equation for the physiological-dependent term b calculated with geometric mean regression of the global data set, a value of 25 per mil for ɛf, and a [PO4 3−] = 0.3 μmol/l (b (GM25 per mil) = 116.96 × [PO4 3−] + 81.42). Values of CO2aq were converted to pCO2 by applying Henry's law with KH (the temperature- and salinity-dependent CO2 solubility coefficient) values (56) calculated assuming a salinity of 35 and surface-water temperatures derived from δ18O values for planktonic foraminifera (Fig. 1C). Propagation of errors results in a 15% uncertainty for calculated pCO2 values (57). EAIS, expansion of the East Antarctic ice sheet.

On the basis of phylogenetic data (25), fossil pollen evidence (26), and recent molecular isotopic data (27), the emergence of the C4photosynthetic pathway, as well as C4 grasses, occurred before the Miocene. Accordingly, C4 flora must have represented a component of vegetation in the early Neogene (28). For example, as much as a 10 to 30% C4dietary influence can be inferred from the carbon isotopic compositions of early to middle Miocene mammal tooth enamel (29). Nevertheless, carbon isotopic trends of paleosol carbonate (30–32), fossil mammal tooth enamel (δ13Cen) (7, 30,32–34), and terrestrially derived organic matter (35) from Pakistan, South America, North America, and Africa support an interval of substantial ecological change to C4-dominated vegetation (grasslands) between about 8 and 4 Ma. The character and pace of this change differed among localities (32), suggesting regional controls on the expansion of C4 flora.

If the widespread expansion of C4 plants was not a response to a sharp decrease in pCO2, then we must seek another explanation for this change. In general, it is not necessarily justified to expect an immediate biological response (that is, diversification) after a physical forcing event (36). Therefore, it is possible that the timing of C4 expansion was far removed from the conditions that promoted it. Alternatively, changes in climatic conditions, other thanpCO2, could have forced C4 plant expansion. The most important modern environmental characteristics that favor C4 plants include aridity for C4 dicots and strong seasonal precipitation (that is, warm-season precipitation), with coinciding high minimum temperatures during the growing season for C4 grasses (12, 37). Changes in seasonal patterns of precipitation and temperature in key regions during the late Miocene can be inferred from a variety of data. For example, evolutionary trends in mammals (and other fauna) and floras from the middle to late Miocene suggest a pattern of increasing seasonality and aridity in North America, Europe, Africa, Pakistan, and Australia (4, 38). Soil carbonate δ18O data from Pakistan, Nepal, East Africa, Argentina, and the eastern Mediterranean (31, 32,39), as well as tooth enamel δ18O values from Argentina and North America (32), increase, suggesting increasing evaporation and aridity preceding and accompanying the expansion of C4 flora. Dust fluxes, likely driven by the development of aridity in Asia and South America, increased in the North Pacific at 7.7 Ma (40) and peaked at about 8 Ma in the subtropical South Pacific (41). An increase in regional precipitation rates in North America (central Oregon and the Great Plains) is inferred from an increase in the depth of fossil soil calcic horizons at 7 to 6 Ma (42). We suggest that it was the development of low-latitude seasonal aridity and changes in growing conditions on a global scale, rather than a decrease inpCO2, that led to the sudden expansion of C4 vegetation at ∼7 Ma.

Global climates could have been altered as a result of tectonic processes. For example, Ruddiman et al. (43), among others, have championed the role of late Cenozoic plateau uplift (southeast Asia and the American West) and other mountain-building events as major drivers of global climate change. Model simulations and paleoclimatic evidence imply that these uplifts altered zonal wind patterns, inducing strong seasonality in precipitation and aridity in many mid- to high-latitude regions in the Northern Hemisphere (43).

The timing of the major large-scale uplift events is controversial. Many studies have relied on the marine strontium isotope record, for example, as an indicator of greater uplift and weathering rates in the Himalaya-Tibetan Plateau (9). Trends in strontium isotope ratios indicate that substantial changes in the rate of 87Sr increase occurred during two episodes in the Miocene (44) from about 21 to 17 Ma and 12 to 9 Ma. The first interval shows little indication of increased sediment yield to the ocean basins (45). The second, however, is associated with a number of indicators of uplift and increasing erosion rates on land (40, 41), including a major increase in clastic sediment flux to the Indian Ocean basin (9 to 6 Ma) (45) [as well as a probable global increase (46)], an increase in Ge/Si ratios in opaline silica (8 to 4 Ma) (47), and a marked increase in pelagic phosphorous accumulation rates (8 to 4 Ma) (48). Therefore, we suggest that it was this late Miocene phase of Asian uplift, in conjunction with preexisting low pCO2levels, that caused the critical changes in climate patterns that favored C4 plant expansion.

  • * Present address: Earth Science Department, University of California, Santa Cruz, CA 95064, USA.


View Abstract

Navigate This Article