Müller et al. (2013) present an interesting hypothesis that quantifies a relationship between temporal changes in the ratio of Mg and Ca in seawater and global hydrothermal flux through oceanic crust. The relationship is quantified through estimates of the area of oceanic crust <65 m.y. old for the past 265 m.y., which are then correlated to fluid flow through time (Müller et al., their figures 2 and 3). Estimates of the age distribution of oceanic crust are the most problematic aspects of the paper and the subject of this Comment, which is based on a review by I.O. Norton of an earlier version of the paper for a different journal.

Fast spreading in the Pacific Ocean has resulted in subduction of enormous areas of crust, and restoring this crust results in large geologic uncertainties. By 100 Ma, only 8% of the area of the Pacific Ocean has preserved Chrons, the M-anomalies. The Pacific Ocean, in turn, covers 40% of the Earth at 100 Ma, so there are large uncertainties in extrapolating areas of subducted crust. Müller et al. show bands that represent their estimates of uncertainties in their figure 2, but these grossly underestimate the true geologic uncertainties. Although these geologic uncertainties could be illustrated for all of the time slices used by Müller et al., there is not space in this one-page Comment to discuss more than one, so we will use the 200 Ma reconstruction.

Figure 1 shows the age model for 200 Ma from Müller et al (from their figures 1 and 2). The model requires extrapolation of ages based on the oldest preserved M-anomaly back to 265 Ma. This is an extrapolation of more than 100 m.y. from the oldest reliably dated M-anomaly, M29 (157 Ma). Besides this extrapolation problem, the ridge geometry shown in the Pacific Ocean has a “tectonic impossibility.” The Izanagi-Phoenix rotation pole (“IP Pole”) is located on the plate boundary between these plates. This pole location predicts extension to the northeast of the pole location (diverging arrows) but implies compression along the plate boundary to the southwest (converging arrows). There are no known examples of this type of oceanic plate boundary geometry, which makes it hard to justify proposing it in the Triassic-Jurassic period where there are no age constraints. Shown in Figure 1 is an alternative model for age distribution at 200 Ma. The Izanagi-Phoenix stage pole (“IP Pole alt”) is shifted so that the equator relative to this pole is at the center of the Izanagi-Phoenix plate boundary (a latitude-longitude grid about this pole is shown) and the total spreading at the triple junction is the same as shown by Müller et al. A polygon that follows lines of longitude (i.e., isochrons) relative to the rotation pole is drawn that outlines crust of 0–65 m.y. age. This interpretation is presented as an alternative. There are no data to support it, so from that point of view it is no better than Müller et al.’s version, although it does avoid the “tectonic impossibility” of a rotation pole located on a ridge axis. The area of 0–65 Ma crust in this alternate interpretation is ∼219 × 106 km2. This value, less the ∼4 × 106 km2 area of crust <1 m.y. old, is plotted on inset A of Figure 1. This is the age-area plot from Müller et al.’s figure 2. This implies a much larger area of young oceanic crust at 200 Ma. Inset A shows a symbol at 100 Ma with a smaller area of young crust than that used by the authors. There is not space to illustrate this, but it is from an alternate interpretation of the four-plate model for the southern Pacific used by the authors (from Seton et al. [2012], and Matthews et al. [2012]).

The above example of an alternate interpretation for 200 Ma, and our analysis of the 100 Ma reconstruction, show that there are large geologic uncertainties in the age distribution of oceanic crust. The same uncertainties apply to the other times mapped by Müller et al.; in fact, given the large uncertainties inherent in these maps, it is probably possible to make an age distribution that keeps the area of crust <65 m.y. old almost constant through time, which makes geodynamic sense as it would imply that Earth’s heat loss by seafloor spreading is constant. At the least, though, the correlations proposed by Müller et al. should be questioned.

This research was supported by the PLATES project at the University of Texas Institute for Geophysics. Plate reconstructions use the program PaleoGIS, provided by the Rothwell Group.