Coordinates and A Hot High Sierra
A recent paper by Mix et al. seeks to further bolster the story about the Sierra Nevada having already reached essentially modern elevations back in the Eocene. Examining the paper made GG want to play with a few things, and in the end the feeling here is that the new data (oxygen isotopes) don’t really help the story. However reconsidering the whole of this dataset brings up questions about just what is being measured.
OK, first off, the paper appears to have two main goals: first, to show that temperatures were never so high as to have disturbed the ∂D (deuterium-hydrogen) measurements originally put forward by Mulch et al., and second to show that the oxygen isotope ratios support the original inference of near-modern elevations of this region.
The temperature results, which originate in differing fractionation coefficients for hydrogen and oxygen when making kaolinite, produce a very curious pattern:
(Note that “upriver” is distance from shoreline in the original paper, which turns out to be measured along paleorivers). Basically, if you take the temperatures at face value, it would seem that temperatures increased as you went upstream–that higher areas were hotter. Perhaps as curious, the spread of temperatures at a single site seems to be quite large. Although these results were used to argue that the hydrogen results had not been contaminated, the authors declined to interpret these temperatures as reflecting the local climate for several reasons, the most interesting being “uncertainty in the kaolinite- water fractionation at low temperatures (see Sheppard and Gilg, 1996) is likely greater than the resolution necessary for temperature-based paleoaltimetry reconstructions, at least across this modest climatic gradient.” One might take that to mean that the temperatures have no meaning at all, yet the mean temperature of all these is taken to be a significant piece of evidence supporting the Eocene origin of these isotopic patterns. This just feels like a bit of situational ethics–the temperatures are meaningful when they support your hypothesis (no problems with ∂D, matches expected Eocene temperatures) and not when they don’t (higher elevations seem to be warmer).
In playing with plotting, made this plot, the significance of which (if any) remains unclear to GG, being a grumpy geophysicist and not a grumpy geochemist:
Again, at least at face value, this is backwards: more depleted (more negative) ∂D values should be colder; if the temperature estimates were wholly random, you might not expect the rather noticeable correlation. But maybe this makes sense, just seemed strange to GG.
OK, but what about supporting the isotopic gradient story?
Well, there is an error in the plots in the paper: one site, Gold Run, is shown as being 84 km from shoreline when it should have been 64 km. Replotting the data as was done in the paper yields this:
Fixing that one site drops the R² from 0.50 as published to 0.26, which to GG’s mind really doesn’t support the original observation so much as fail to violate it. But this plot suggests a number of questions, the main one being, just what water were these rocks in equilibrium with? Why is there so much scatter where there are more samples?
Consider for instance the collection of points near 55 km in this plot.Most of these are from the You Bet Diggings from three different collection sites. The first (09-, 10-, 11-UB05) have ∂D measurements of -89 to -91 and ∂18O of 14.7 to 18.3. The second site (02- and 05-UB05) has ∂D of -96 to -98 while ∂18O is 17.5-18.5. The third (07- and 08-UB05) has an identical (to the meter) location as the first site and has ∂D values of -87 to -94 and ∂18O of 14.7-17.9. In other words, this one locality–and two of the three collection sites–has oxygen isotope variations that very nearly span the entire range of observations made throughout the region. The story is nearly the same at the locality at 64 km, which is the Gold Run site. It seems that the scatter at a given site is so large that you could draw a horizontal line across the plot above and justify it as honoring the data. This does not seem to be a compelling case supporting the isotopic gradients.
So again, why so much scatter? In many ways, we are in the dark. We don’t know if the separate samples were from the exact same horizon or tens of meters apart vertically. Most of the scatter is between samples, not within them. In Mulch et al., the scatter is attributed to the kaolinites in some cases equilibrating with river water and in some cases with meteoric water. It seems an untestable hypothesis. One can then only assume that the highest values here represent meteoric water and the lowest river water–quite a spread. A perfectly defensible alternative is that this simply reflects the noise in the system. Certainly if you accept the meteoric vs river explanation, then which source created the values at more poorly sampled localities? And why is there such a high jump from Gold Run (at 64 km) and You Bet (at 55 km)? There is at present only 5 km distance between these and 60 m of elevation difference. Yet the entire gradient in oxygen isotopes comes down to the difference between these two sites. It seems to GG that some other explanation than variation in paleoelevation is in play here.
If, for instance, you take the highest values at each site as reflecting meteoric water, then you want to look at either elevation or distance from the paleoshoreline. In both cases, these two sites are virtually identical, being less than 3 km different in distance and no more than 60 m different in elevation, yet the difference in the highest ∂18O is about 4‰, which should represent some 1600m difference in elevation according to the tables in Mix et al. This cannot be paleoelevation, it is unlikely to reflect differences in paleorainfall unless these are samples from profoundly different times. And if that is the case, then all of these data are uninterpretable until their ages are greatly refined (which is not an easy trick).
OK, so the oxygen story kind of sucks. What of the original hydrogen story?
Here’s the plot of ∂D vs river distance, with Gold Run fixed again:
Again we notice a pretty broad range at any individual site with a lot of samples. So You Bet, for instance, runs from -87 to -98; Gold Run–again just next door–varies from -103 to -106 (the other values at 64 km distance are from Malakoff Diggings). So the difference in the meteoric water should be no less than 5‰ and could be as high as 16‰, so a paleoelevation interpretation of a difference of 330m to about 1 km. This doesn’t seem to be much of an improvement over the issues with the oxygen data. And if you plot this data with airline distance from the paleoshoreline or versus elevation, the patterns get uglier as north-south variations along the one stretch of the paleoYuba collapse:
If the high end of the measurements reflects meteoric water, then there is a deep hole in the You Bet region 45-50 km from the shoreline. If the low end reflects river water, then there is no variation over 600m of modern elevation before a sudden drop near 900 m modern elevation with no further change in elevation below that.
Arguably that is over interpreting these data, but then so is drawing a line through these points. The intra-locality scatter of nearly 10‰ limits any interpretation that does not resolve possible temporal variations or possible variations in the kaolinitization environment. Certainly it is possible that there are large variations in the isotopic composition of Eocene rivers or rainfall, but with that scatter and the strong lateral differences over very short distances, it seems equally probable that there are no resolvable differences in rainfall or river water.
So do we know there was a “hot, high Eocene Sierra” as the title of the Mix et al. paper promises? No, we do not. That might be the case, but there are enough internal inconsistencies in the geochemical data to make the case far from certain.