Technical Comments

Response to Comment on “Sensitivity of seafloor bathymetry to climate-driven fluctuations in mid-ocean ridge magma supply”

+ See all authors and affiliations

Science  17 Jun 2016:
Vol. 352, Issue 6292, pp. 1405
DOI: 10.1126/science.aaf2021


Huybers et al. present new bathymetric spectra from an intermediate-spreading ridge as evidence for a primary contribution of sea level cycles to the morphology of the seafloor. Although we acknowledge the possibility that sea level–modulated magmatic constructions may be superimposed on a first-order tectonic fabric, we emphasize the difficulty of deciphering these different contributions in the frequency domain alone.

We thank Huybers et al. (1) for continuing the discussion on the processes that shape the seafloor and their sensitivity to sea level cycles. Prefacing any discussion on this topic, we again emphasize that the dominant source of relief at slow- and intermediate-spreading mid-ocean ridges consists of scarps that are axis-facing, are 50 to 1000 m high (Fig. 1A), and are the unambiguous manifestation of lithospheric-scale, seismogenic normal faults (25). The average spacing of these faults can be measured by identifying fault scarps along cross-axis bathymetry and side-scan sonar profiles (4, 6). Such data yield a characteristic length scale, which also emerges from statistical analyses of bathymetry that are agnostic to the local geology (7, 8). Further, this length scale is known to decrease with increasing spreading rate (8, 9) and thus cannot be explained by a model in which the dominant seafloor wavelengths are controlled solely by time-periodic processes such as climate cycles. In (10), we developed a quantitative model that explains the observed variations in fault spacing across all spreading rates without requiring temporal variations in magma supply.

Fig. 1 The spectral signature of seafloor-shaping processes.

(A) Fault scarps generating sawtooth-like topography at the Chile Ridge. Sawtooth topography (or any periodic signal) can be decomposed into a sum of harmonics. (B) Spectrum of cross-axis bathymetry (black curve) at the Australian-Antarctic Ridge from (11). The red curve shows the spectra of seafloor older than 700 ky and younger than 1.2 million years. Dashed blue lines indicate Milankovitch frequencies. (C) Bathymetric spectra from the Chile Ridge. From the uppermost curve: A spectrum shown by Huybers et al. in their comment (reference profile), which is calculated using bathymetry measured along the ship’s central acoustic beam (centerbeam). A spectrum of the same profile using (gridded) bathymetry produced from processed multibeam data (13) is shown just below and shows appreciable differences at high frequencies. Additional spectra shown below use gridded bathymetry and correspond to (from top to bottom) the conjugate profile located on the eastern side of the ridge axis opposite the reference profile; a profile 10 km south of—and parallel to—the reference profile; and stacked spectra of five cross-axis profiles on the western side of the same segment. The color code for the spectra and the dashed lines is the same as in (B).

Huybers et al. criticize our analysis as being incomplete by pointing out that features such as seamounts and volcanic ridges indicate that seafloor magmatism is punctuated in time and space and can generate large-amplitude topography in a manner that is not accounted for in our models. Although it is true that we neglected sudden or magnified magma pulses, we stress that our goal was to assess the topographic manifestation of changes in melt supply attributable to sea level cycles. Thus, we assumed that magma supply to the crust followed the continuous fluctuations predicted by (11) and, as such, that the fluctuating portion of the magma supply is emplaced in the same way as the background melt supply. Specifically, we assumed that magma is primarily (≥70%) emplaced as intrusive units that produce an upward load on the base of a lithosphere of finite thickness. This intrusive-to-extrusive ratio is consistent with seismic observations of a thin layer 2A relative to the overall crustal thickness (12), and we see no clear reason that the fluctuating portion of magma supply should be entirely extrusive. Combining this assumption with the effects of lateral smearing due to magma residence in a kilometer-scale melt reservoir leads to our prediction that volcanic topography caused by sea level cycles should be of much lower amplitude than the faulting signal, especially at intermediate-spreading rates.

Our study (10) also highlighted a key issue in looking for a sea level signal in the bathymetry of intermediate-spreading ridges such as the Australian-Antarctic Ridge (AAR) or the Chile Ridge (CR). In these systems, the observed spacing of major faults (~3 km) happens to correspond closely to the periodicity of glacial-interglacial cycles [~100,000 years (ky)], thereby making it difficult to decipher a sea level–modulated magmatic contribution to the ~100-ky spectral peak from that of the strictly tectonic morphology. By contrast, the 41- and 23-ky peaks reported at the AAR (10) and CR (in the Comment by Huybers et al.) do not correspond to any obvious tectonic length scale, leaving sea level–driven volcanic fluctuations—if proven to occur on a global scale—as one possibility.

However, because the frequency-domain manifestation of even a periodic process can be complex, a simple exercise exposes a key weakness of a strictly spectral view of seafloor bathymetry. For example, if seafloor bathymetry entirely consisted of steep fault scarps separated by regions of gentler slopes with a fault spacing S (Fig. 1A), this signal could be decomposed into a sum of Fourier harmonics and produce a spectrum with significant energy at wavelengths S/2, S/3, and so on. In this framework, the strongest energy peak at 82 ky at the AAR (Fig. 1B) could reflect a mean fault spacing of 2.7 km, which would in turn produce spectral overtones at 41, 27, and 21 ky—close to the observed high-frequency climate peaks (Fig. 1B).

Nonetheless, there could be a mechanism not accounted for by our models that enables the expression of climate periodicities in seafloor bathymetry. Robust evidence for such an expression would require an unambiguous spectral signal with the following characteristics: (i) manifestation on both sides of the ridge axis, (ii) along-axis continuity, and (iii) presence on a global scale. Yet, recalculating the CR spectra shown by Huybers et al. reveals sensitivity to relatively small uncertainty in bathymetric data, marked differences between the spectral signature of conjugate profiles, and sharp changes between profiles that are only 10 km apart (Fig. 1C). Further, when stacking the spectra of five profiles from the west side of the axis, peaks in the 20- to 80-ky period range tend to destructively interfere, and clear Milankovitch peaks cannot be recovered. This is also true for spectra computed exclusively from seafloor older than 700 ky and younger than 1.2 million years, which do not systematically reveal a shift in frequency content to a dominant 41-ky periodicity either at the CR (Fig. 1C) or the AAR (Fig. 1B). All spectra, however, show energy in a range of wavelengths consistent with a fault spacing of 2 to 3 km.

In summary, exercises such as those discussed above show that global bathymetric analyses in both time and frequency domain are needed to move this debate forward. Further, direct seafloor geological observations would be a straightforward way to establish the nature (i.e., magmatic versus tectonic) of topographic features that have amplitude and spacing consistent with Milankovitch periods.

References and Notes

  1. Acknowledgments: The authors thank P. Huybers for sharing his spectral analysis codes. Funding was provided by NSF grants OCE-1154238 (M.D.B), OCE-1155098 (G.I. and S.H.), EAR-1009839 (W.R.B), CNRS support to J.E., and an LDEO Postdoctoral Fellowship for J.-A.O.
View Abstract


Navigate This Article