The mother of all extinctions and the perils of correlation (2)

The previous post, instalment one of this saga, looked at what a geologist must do, faced with rock exposures in, say, a cliff face, to read the ledgers of our planet's past. How the sequence of layers records different and changing sedimentary environments, and how these layers must be traced and correlated laterally to build up a picture of what the surface of the earth looked like and how it changed with in time and place. But then we must appreciate that the simple correlation, joining up the pieces of the puzzle by matching rock characteristics (lithostratigraphic correlation) is not the same as a chronostratigraphic scheme because essentially all sedimentary layers are diachronous - the timing of their deposition varies from place to place. But a lithostratigraphic approach is the necessary starting point - unless the geologist is incredibly lucky and finds a real equivalent of my imaginary "pulpite," a layer whose actual age can be determined, the rocks are all he or she has - what you see is what you get.

From time immemorial, it has been the convention that the pile of rock layers is described and depicted as if looking at the cliff face in profile - a column is drawn, showing the different lithologies of the different layers, and reflecting their toughness - rocks resistant to erosion, for example, sandstones, are drawn as ledges, "cliff-formers," while shales, easily eroded away, are less prominent. Here's an example from a study of the 140 million year old Dakota Sandstone in Colorado by workers at the University of Kansas. Sections from different places (I've only included two here for simplicity), annotated with details of the rock type, are put next to each other and the proposed correlation lines drawn between them; in this example, the interpreted environment of deposition of the successive "packages" of sandstones and shales is also included. The correlation lines are drawn with certainty - but note how different the two piles are, only 4.5 kilometers away from each other. Nature abhors continuity, but the geologists can argue for their interpretation from the evidence and the data, and tell a coherent, sensible, geological story - it may not be exactly right but it works. Of course, what the scheme doesn't tell us is when the front of the delta was at what is now Billy Creek and when it was at Chaffee Creek - we must be careful what conclusions we draw.

And the conclusions we draw may differ significantly depending on how such correlations are made and what we infer the significance is for timing and diachroneity; a single sandstone layer, perhaps only a few centimeters thick, depending on how it's correlated from place to place, can change the geological story dramatically. And that is exactly where I've been trying to get to with all this pre-ambling.

The Karoo Basin of South Africa is a spectacular place. Its landscapes are ones that seduce a writer into the old clichés of endless skies and panoramic vistas - because that's the way the place is. The rocks of the Karoo that make these wonderful landscapes are the ledgers of a long period of earth's history; hundreds of millions of years of change are recorded there, including one of the most dire and catastrophic changes that life has endured, the Permian-Triassic extinction. The reasons and details are still unclear and the subject of vigorous and exciting debate, but the fact is, as I wrote in instalment one, that 250 million years ago, 90 percent of marine species and 70 percent of land life were obliterated, they disappeared forever.

As I described in the Alaska river example in the previous post, walk along a river system today and you'll eventually get to the shoreline, the beach and the marine realm beyond. So it was in the Permian period in the Karoo - mountains to the south and west were being worn away, the detritus carried by rivers to the sea, and, in the deep parts of the marine basin, sheets of sand and mud were deposited by the same processes we see, for example, in the Monterey submarine canyon today. Those layers are now preserved in the landscape shown below, which I had the pleasure and privilege of roaming around some time ago. Stratigraphy and correlation is relatively straightforward in these marine sediments - the layers are more or less continuous and can be traced in three dimensions around the terrain. But in detail, even those layers are not continuous - the lithology and the architecture changes laterally.

But even with lateral changes and diachroneity, the great advantage of marine deposits is that not much erosion occurs on the sea bed - deposition is the order of the day, and the entries into the ledger are made continuously - the story is a joined-up one. And, within the marine sedimentary pile of the Karoo are beds of volcanic ash that represent geologically instantaneous events - and contain zircon minerals that can be dated radiometrically (see my earlier post on Magic crystals - tales zircons tell). Recent international research collaboration has used these ash layers to pin down exactly where the boundary between the Permian and Triassic periods - the time of the great extinction - is in the sedimentary pile. But travel towards the Permian shoreline and up onto the land where the rivers were flowing, and things become much more complicated. Of course they do - the land surface is dynamic and constantly swept and excavated by those rivers: lateral continuity is abhorred. But that's where the record of life's catastrophic experience is, nevertheless, preserved. It's in those sediments that we see the diversity of vertebrate life on the land decimated. In the sands and muds of the latest Permian is a thriving ecology, dominated by the dynoconodonts, mammal-like reptiles of whom a good example is rambling unsuspectingly around the bottom of the image at the head of these posts - it was as happy as a turkey the day before Christmas. In the sands and muds of the Triassic period, just a little higher up the pile, his place is taken by one of the few survivors, smug lystrosaurus, who would go on with his extended family to found the dinosaur dynasty. Other vertebrates that disappeared at the end of the Permian or only first appeared in the Triassic "recovery" after the extinction tell the detailed story. Even the character of the sediments changes dramatically - the hallmarks of the sands of the late Permian rivers suggest that they were relatively tranquil, meandering across their floodplains, as on the left of the image below. The rivers of the early Triassic were far more violent, braided channels constantly changing course and carrying huge cargoes of sediment (right) - quite possibly because the land was no longer protected from erosion by plant life.

In between these clearly Permian and clearly Triassic sediments, things are a mess. But it's a geological mess that should shed light on how the extinction occurred, when it occurred, and how long it took, and these rocks have long been the subjects of devoted scrutiny by teams of researchers from around the world. The image below is taken from a 2003 paper by Greg Retallack of the University of Oregon, Roger Smith, of the South African Museum, and Peter Ward from the University of Washington - leading workers in the field. It summarises neatly and graphically the typical stratigraphy and the extinctions across the Permian-Triassic boundary in the Karoo.

The difficult and controversial part of the sedimentary pile is the section marked "laminites." These layers, sometimes purple, sometimes red, and shown in the image at the top of this post, are thin layers - laminae - of fine sandstone, siltstone and shale that contain no evidence of life whatsoever - they are referred to as the "event" zone or the "dead" zone, the pages of the ledger that record the extinction. The work has built up a picture of this zone being correlatable across long distances in the Karoo, comparable with the stratigraphy of the extinction elsewhere in the world and consistent with the definition of the boundary in the equivalent marine sediments; and it's not just lithostratigraphic correlation - evidence from the well-defined timing of reversals of the earth's magnetic field, the chemical signatures of isotopes in the sediments, and at least one radiometric date, support this. And an important implication is that the extinction took place in what is geologically a very short period of time.

Now, earlier this year, Robert Gastaldo and students from Colby College in Maine, together with Johann Neveling of the South African Council for Geosciences, published a paper suggesting a radical revision, and titled it "The terrestrial Permian-Triassic boundary event bed is a nonevent." Working in part in the same area as Retallack and his colleagues, across the kind of outcrops of the "event zone" shown at left, they identified a single sandstone layer at the top of the laminite zone and correlated it, lithostratigraphically, across a valley to a different level - 10 meters different - than proposed in the earlier work. From their paper, I've reproduced the two different correlations below (slightly simplified and annotated) - the piles of sediment, the lithostratigraphy, from four different locations are the same, but above, A, is Gastaldo's interpretation of the previous correlation, below, B, his proposed revision. This whole issue is far more complex and involves far more detail and data than I have been able to include here. The key point is that the substantial difference hinges on the interpretation of the correlation of the one single thin bed of sandstone - but the implications are huge.

In scheme A, the "event zone" is a common and roughly synchronous set of sediments across the Karoo, and the extinction was rapid; in scheme B, different things were going at different places and times and the extinction took place over a long period of time; and the implications cascade onwards, for what species survived for how long and for how life recovered. But, as Bryan at In Terra Veritas pointed out shortly after the paper was published, it ends up relying on lithostratigraphy and adds no chronostratigraphic data, no time framework. It's for this reason that I have spent so much time trying to set out the pitfalls of lithostratigraphic correlations and the importance of establishing a chronostratigraphic framework - as I have said many times, diachroneity is everywhere and nature abhors continuity.

Now Gastaldo's paper has, as I'm sure was intended by the melodramatic title, stimulated all kinds of attention in the science news press - including the heading "The catastrophe that wasn't" on a highly reputable site. In an article in the June edition of Earth magazine, more soberly titled "Permian extinction layer not actually extinction layer?," Retallack comments that correlations have been misinterpreted:

The problem, Retallack says, is that Gastaldo incorrectly assumed that he and his colleagues had matched a laminite on the western side [of the valley] composed of fine sandstone with a laminite on the eastern side composed of shale......

Although Retallack and his colleagues disagree with some of Gastaldo's findings, they agree that the laminated interval, or any other geologic feature that occurs at, above or below the P-T boundary, cannot be used on its own to mark the boundary itself. Instead, Retallack says, scientists should use the fossils themselves to mark the extinction and later the recovery, as he and his colleagues have done. Gastaldo's work on the stratigraphy, therefore, "doesn't affect where we place the boundary. It doesn't affect the extinction."...... Peter Ward...... agrees. "It's much ado about nothing," he says.

It may or may not be much ado about nothing, but it is a dramatic illustration of not only the perils of correlation, but the distortion of science. This distortion is precipitated by two factors: first, the provocative title used for a scientific paper, that is a declaration in the interests of publicity, and second, a dispiriting apparent lack of understanding that this is how geology, how science, works. That there is never an absolute truth but rather a process of observation, testing, dispute, and refinement that should lead to an increasingly better understanding - but with multiple working hypotheses and the arguments that accompany them along the way.

And it's very much up to us to broadcast and defend this process, not simply for general scientific appreciation, but because we also add grist to the mill of the "Answers in Genesis" crowd who gleefully reported as follows (excuse me for quoting it verbatim and in total - I couldn't bring myself to summarise it):

The Permian extinction is considered by old-Earthers to have been the largest mass extinction in Earth history, with 90 percent of marine species and 70 percent of land species dying out. It has also long been thought of as a sudden, catastrophic event—until now.
A team led by Colby College paleontologist Robert Gastaldo takes a different view. Gastaldo led students on six trips to South African locations purported to be home to evidence of the Permian extinction—specifically, a thin sedimentary layer separating the Permian period from the Triassic that followed it. But according to Gastaldo, that layer “couldn’t be traced more than about 100 meters laterally,” showing that it wasn’t a global event. “We spent days walking kilometers throughout the [sites] trying to trace it from every angle and couldn’t,” Gastaldo said.
Yet in other places, the team found the sedimentary layer eight meters below the Permian–Triassic boundary! Gastaldo’s conclusion? “Because the boundary event bed doesn’t occur at the same position in the rock record there can be no one, unique event.”
The creation model considers many of the sedimentary layers we observe—including most of the layers that have fossils—to have been laid down by geological events associated with the global Flood (including volcanic activity) along with the Flood itself. It’s therefore very easy to incorporate ideas such as the Permian extinction, or the better-known K–T (Cretaceous–Tertiary) extinction event, with the Flood model. Even if evolutionists decide a certain event took place millions of years earlier or later than what was once thought, that translates in the Flood model to likely mere months (or less) of difference.
We also have to smile whenever longstanding evolutionary ideas are overturned. In this case, a sudden Permian extinction event is now “last year’s fact,” as the CreationWise cartoon says. While a benefit of the scientific method is that hypotheses incorporate new evidence, that’s also the drawback of using science as the be-all and end-all of knowledge.

To anyone who has stuck with me this far, to the bitter end, so to speak, congratulations and heartfelt appreciation. What started off as a little story about a little layer of sand with big stories to tell, ended up as an epic saga - but I hope you'll agree that it's fascinating stuff.

[for news reports on the controversy see http://www.nsf.gov/news/news_summ.jsp?cntn_id=114276&govDel=USNSF_51 and http://www.physorg.com/news155238422.html; Gastaldo's paper can be downloaded from http://www.colby.edu/~ragastal/RAG_reprints/RAG2009a.pdf and examples of the other side of the story can be found at http://www.nasmus.co.za/PALAEO/jbotha/pdfs/Smith%20and%20Botha%202005.pdf, http://www.fsl.orst.edu/wpg/events/S02/RiverMorph.pdf, and http://www.gps.caltech.edu/users/jkirschvink/pdfs/WardKarooScienceFinal.pdf. For several key papers published by the Geological Society of America, and therefore not easily available to non-members, please contact me for PDFs.]