pubmed.ncbi.nlm.nih.gov

Orbital pacing of carbon fluxes by a ∼9-My eccentricity cycle during the Mesozoic - PubMed

  • ️Thu Jan 01 2015

Orbital pacing of carbon fluxes by a ∼9-My eccentricity cycle during the Mesozoic

Mathieu Martinez et al. Proc Natl Acad Sci U S A. 2015.

Abstract

Eccentricity, obliquity, and precession are cyclic parameters of the Earth's orbit whose climatic implications have been widely demonstrated on recent and short time intervals. Amplitude modulations of these parameters on million-year time scales induce "grand orbital cycles," but the behavior and the paleoenvironmental consequences of these cycles remain debated for the Mesozoic owing to the chaotic diffusion of the solar system in the past. Here, we test for these cycles from the Jurassic to the Early Cretaceous by analyzing new stable isotope datasets reflecting fluctuations in the carbon cycle and seawater temperatures. Our results document a prominent cyclicity of ∼9 My in the carbon cycle paced by changes in the seasonal dynamics of hydrological processes and long-term sea level fluctuations. These paleoenvironmental changes are linked to a great eccentricity cycle consistent with astronomical solutions. The orbital forcing signal was mainly amplified by cumulative sequestration of organic matter in the boreal wetlands under greenhouse conditions. Finally, we show that the ∼9-My cycle faded during the Pliensbachian, which could either reflect major paleoenvironmental disturbances or a chaotic transition affecting this cycle.

Keywords: Mesozoic; carbon cycle; eustatism; orbital cycles; paleoclimate.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Fig. 1.
Fig. 1.

Compilation of δ13C and δ18O data (A) measured on NW Tethyan belemnites represented in their global (B) and regional (C) paleogeographic contexts. (B) Modified from ref. , with permission from Elsevier and courtesy of Ron Blakey (Colorado Plateau Geosystems). (C) Image courtesy of the Commission of the Geological Map of the World (57). The smoothed isotopic curves are LOWESS regressions with respective coefficients of 0.0016313 and 0.0016667 for δ18O and δ13C data. Aal, Aalenian; Apen, Apennines; Apt, Aptian; Ast, Asturias; Baj, Bajocian; Bako, Bakony and Gerecse Mountains; Bar, Barremian; Bat, Bathonian; BC, Basque–Cantabrian Basin; Berr, Berriasian; Bet, Betic and Subbetic basins; Cal, Callovian; Cala, Calabria; Crim, Crimea; Do, Dorset; GC, Grands Causses Basin; Ha, Hauterivian; Ib, Iberian range; Kim, Kimmeridgian; Lus, Lusitanian Basin; Mid, Midland Platform; Oxfor, Oxfordian; Paris: east and northeast of the Paris Basin; Pliensb, Pliensbachian; Pol, Polish Jura Chain; Sax, Lower Saxony Basin; Sco, Northeastern Scotland; SE, Sud-Est Basin; Sine, Sinemurian; Sk, Isle of Skye; Swa, Swabian Basin; Swiss, Swiss Jura Chain; Titho, Tithonian; Val, Valanginian; Vo, Vocontian Basin; and York, Yorkshire.

Fig. 2.
Fig. 2.

The 2π-multitaper LOWSPEC-treated spectra of belemnite δ13C (A) and δ18O (B) time series. (C) The 2π-multitaper spectrum of the amplitude modulation of the 2.4-My band from the astronomical models (i.e., La2004 and La2010d) (3, 4) over the studied interval. The spectra shown in A and B were performed after running 1,000 Monte-Carlo age-random simulations. The gray lines represent the spectra of the 1,000 simulations and the black line represents the average spectrum. Only the 9.1-My cycle is significant with respect to Bonferroni-corrected confidence thresholds.

Fig. 3.
Fig. 3.

Comparison of belemnite δ13C signals with other paleoenvironmental records and astronomical solutions. (A) δ18O time series (this study) reflecting seawater temperature and salinity. (B) First-order (Left) and second-order (Right) cycles of eustatic transgression/regression in the Mediterranean domain (41). (C) Belemnite δ13C time series with Taner low-pass filters showing the general trend (cutoff frequency: 0.04883 cycles per million years, roll-off rate: 103) and the 9.1-My cycle (cutoff frequency: 0.1587 cycles per million years, roll-off rate: 103). (D) Detrended belemnite δ13C times series with low-pass filter of the 9.1-My cycles numbered from C1 to C9. (E) Amplitude modulation of 2.4-My eccentricity cycles showing the ∼9-My envelope [calculated from La2010d (4)]. (F) Main oceanic anoxic events (56). (G) Durations and surface areas of large igneous provinces (56). (H) Results from time-frequency Fourier transforms performed with 20-My-width windows showing temporal changes in the expression of the 9.1-My period. The black contours indicate the intervals exceeding the 95% confidence level in the evolutive harmonic F-test. Six positive peaks of δ13C of nine correspond to maximum flooding events indicated by purple arrows. See Fig. 1 for stratigraphic abbreviations.

Fig. S1.
Fig. S1.

Influence of smoothing methods (red curves) of the raw δ13C and δ18O series (black dots) on the multitaper spectra. (A) δ13C data smoothed by averaging the values having a same age and (B) resulting 2π-multitaper method (MTM) spectrum. (C) δ13C data smoothed by applying medians to the values having a same age and (D) resulting 2π-MTM spectrum. (E) δ13C data smoothed by robust LOWESS regression and (F) resulting 2π-MTM spectrum. (G) δ18O data smoothed by averaging the values having a same age and (H) resulting 2π-MTM spectrum. (I) δ18O data smoothed by applying medians to the values having a same age and (J) resulting 2π-MTM spectrum. (K) δ18O data smoothed by robust LOWESS regression and (L) resulting 2π-MTM spectrum. The frequencies and the periods having high-power peaks are labeled in cycles per million years and million years, respectively. The uncertainty of frequency is calculated from the Rayleigh frequency of the padded series. Note that the smoothing method has no influence on results. BW, bandwidth.

Fig. S2.
Fig. S2.

Influence of detrending methods (red curves) on the multitaper spectra of the LOWESS-smoothed δ13C series. (A) Without detrending. (B) With linear trend. (C) With piecewise linear trend. (D) With LOWESS trend (coefficient = 0.3). (E) With Taner low-pass filter trend based on a frequency cut of 4.883 × 10−2 cycles per million years. (F–J) Corresponding detrended and standardized δ13C series (average = 0 and SD = 1). (K–O) Power spectra corresponding to the above standardized series. The frequencies and the periods having high-power peaks are labeled in cycles per million years and million years, respectively. The uncertainty of frequency is calculated from the Rayleigh frequency of the padded series. Note that the 9.1-My peak is always present.

Fig. S3.
Fig. S3.

Influence of detrending methods (red curves) on the multitaper spectra of the LOWESS-smoothed δ18O series. (A) Without detrending. (B) With linear trend. (C) With LOWESS trend (coefficient = 0.3). (D) With piecewise linear trend. (E) With Taner low-pass filter trend based on a frequency cut of 7.324 × 10−2 cycles per million years. (F–J) Corresponding detrended and standardized δ18O series (average = 0 and SD = 1). (K–O) Power spectra corresponding to the above standardized series. The frequencies and the periods having high-power peaks are labeled in cycles per million years and million years, respectively. The uncertainty of frequency is calculated from the Rayleigh frequency of the padded series.

Fig. S4.
Fig. S4.

The 2π-MTM spectra of δ13C (Left) and δ18O (Right) time series with conventional (A and B), robust (C and D), and lowspec (E and F) red-noise confidence levels before age error simulation. The robust red-noise confidence levels are based on Meyers (53). (G and H) Summary of red-noise significativity levels for δ13C (Left) and δ18O (Right) spectra. (I and J) The 2π-MTM harmonic F-test significativity levels of δ13C (Left) and δ18O (Right) series. The color code of the significance thresholds is the same for all figures. BW, bandwidth.

Fig. S5.
Fig. S5.

The 2π-MTM spectra of δ13C (Left) and δ18O (Right) series with conventional (A and B), robust (C and D), and lowspec red-noise (E and F) confidence levels after age-error simulation. The robust red-noise confidence levels are based on Meyers (53). (G and H) Summary of red-noise significativity levels for δ13C (Left) and δ18O (Right) spectra. (I and J) The 2π-MTM harmonic F-test significativity levels of δ13C (Left) and δ18O (Right) series. The color code of significance thresholds is the same for all figures. From A–F, the gray curves represent the spectrum of each of the 1,000 simulations and the black curve is the average spectrum. The confidence thresholds are calculated from the average spectrum. BW, bandwidth.

Fig. S6.
Fig. S6.

Coherency and cross-phase spectral analysis between the 2.4-My cycle envelope calculated from the La2010d nominal solution and the 9.1-My δ13C cycle. (A) Coherency spectrum. The horizontal dashed line indicates the corrected 95% confidence level for the coherency between red noise and a narrow band signal. The vertical dashed line indicates the position of the frequency of 0.1099 cycles per million years (period of 9.1 My). (B) Cross-phase spectrum. The horizontal dashed line indicates the 0° phase, and gray shading indicates the approximate 95% confidence level. We used the cross-MTM method in the Matlab routine of Peter Huybers (

www.people.fas.harvard.edu/∼phuybers/Mfiles/

; e.g., ref. 53).

Fig. S7.
Fig. S7.

Time-frequency spectral analyses of δ18O (Left) and δ13C (Right) series. (A and E) LOWESS smoothing of δ18O and δ13C series (in blue) with the long-term trends (in red). (B and F) The 2π-MTM spectra of δ18O and δ13C series such as in Fig. S4 E and F. (C and G) Spectrograms of δ18O and δ13C series with 20-My windows. Black contours indicate the zones above the 95% confidence levels in the evolutive harmonic analysis, as shown in D and H. Low and highest powers are in blue and red, respectively. (D and H) Evolutive 2π-MTM harmonic confidence levels (the highest significativity levels are in red and black contours indicate the zones above the 95% confidence levels).

Fig. 4.
Fig. 4.

Scheme of paleoenvironmental changes linked to maxima (A) and minima (B) of eccentricity. In this model, the carbon cycle is highly influenced by the seasonal dynamics of hydrological processes and sea-level changes.

Fig. S8.
Fig. S8.

Illustration of the Monte Carlo Markov chain Bchron simulations. (A) Model for the whole time interval. The gray zone represents the location of 95% of the randomized ages. The red band represents one of the simulations. (B) Focus on the Aptian Stage. (C) Difference between simulated ages and their corresponding original ages for the 1,000 simulations. Each gray line corresponds to one simulation and the red dashed lines are the lower and upper boundaries of 95% of the simulated ages.

Similar articles

Cited by

References

    1. Berger A, Loutre MF. 1994. Precession, eccentricity, obliquity, insolation and paleoclimates. Long-term Climatic Variations, NATO ASI Series, eds Duplessy JC, Spyridakis MT (Springer, Berlin), Vol 22, pp 107–151.
    1. Paillard D. Climate and the orbital parameters of the Earth. C R Geosci. 2010;342(4-5):273–285.
    1. Laskar J, et al. A long-term numerical solution for the insolation quantities of the Earth. Astron Astrophys. 2004;428:261–285.
    1. Laskar J, Fienga A, Gastineau M, Manche H. La2010: A new orbital solution for the long-term motion of the Earth. Astron Astrophys. 2011;532:A89.
    1. Hinnov LA. Cyclostratigraphy and its revolutionizing applications in the earth and planetary sciences. Geol Soc Am Bull. 2013;125(11-12):1703–1734.

Publication types

LinkOut - more resources