The Future of Upper Triassic Vertebrate Biostratigraphy and Biochronology

Bill is currently reviewing the most recent issue of Journal of Vertebrate Paleontology, which contains three papers that directly bear on Upper Triassic archosaurs: descriptions of the weird little basal archosauriform Doswellia (Dilkes and Sues, 2009) and the postcranial skeleton of the “rauisuchian” Batrachotomus (Gower and Schoch, 2009), and a methodological critique of Spencer Lucas’ “Late Triassic land vertebrate faunachrons” (Rayfield et al., 2009). This makes this possibly the BEST JVP EVER in spite of the fact that there is a f-ing shark on the cover.

The Rayfield et al. paper touches on a lot of the same issues I covered in my dissertation (and hope to get into print before too long), and I want to spend a few posts talking about it. It is a short but direct criticism of the way in which Lucas and his colleagues use Triassic vertebrates for intercontinental correlation. Specifically, how they squeeze alpha taxonomy to most conveniently fit their correlations (by lumping when it aids correlation and splitting when it helps subdivide superimposed strata), ignore the effects of endemism, cannot seem to keep track of their own reasoning about how stratigraphic units are correlated, and make arbitrary and largely unexplained decisions in resolving overlaps between supposed index taxa.

As I discussed in a previous blog, putting the events in the history of life in correct chronological order is the ultimate goal of the biostratigrapher. The Law of Superposition helps us with this goal when looking in particular study areas (in my case, the Dockum Group in southern Garza County, West Texas, and Petrified Forest National Park in northeastern Arizona). If the lithostratigraphy is correctly worked out by walking out contacts, and vertebrate localities have been correctly tied into the lithostratigraphic model, superposition allows us to place vertebrate occurrences, and therefore evolutionary, migratory, and/or extinction events, in the correct chronological order. However, it is far more problematic to work out the relative ages of strata and their constituent fossils in geographic areas too widely separated to physically walk out lithologic units in between. If you cannot place physically separated lithologic units in the correct superpositional order, how do you determine what their relative ages are?

The longest-utilized method, biostratigraphic correlation, relies on the fossils themselves. If you can identify the same individual taxa and/or faunas, and ideally the same succession of individual taxa and/or faunas in different areas, then it might be assumed that equivalent taxa and/or faunas (and therefore the strata containing them) are the same age. However, as wise and realistic biostratigraphers working on both invertebrates and vertebrates have regularly pointed out for the past century, this is sketchy assumption which comes dangerously close to treating biostratigraphic units as chronostratigraphic units. The ranges of taxa in different areas may be diachrononous, and if the fossil record is incomplete (as it always is), you run the risk of identifying incomplete ranges of taxa living in different areas as being exactly equivalent in age. In reality, the boundaries of lithostratigraphic, biostratigraphic, and chronostratigraphic units do not have to coincide with each other in any way.

Mammalian biostratigraphers largely have gotten around this problem thanks to the rampant volcanic activity in western North America during the Cenozoic, which has provided them with a plethora of radiometric dates, augmented by extensive magnetostratigraphic zonation of Cenozoic strata. They therefore have a system of correlating isochronous strata completely independent of biostratigraphic correlation, which can, and has, been used to determine whether or not biostratigraphic units are in fact isochronous.

Unfortunately, there are relatively few volcanic deposits in Upper Triassic rocks in western North America, and little work has been done on the magnetostratigraphy. Consequently, chronostratigraphic correlation has mostly relied on vertebrates and pollen. However, with current advances in radioisotopic dating techniques making it easier to extract absolute age dates from a wider variety of strata (and run the samples in less time for less money), biostratigraphic correlation as a way of determining time-equivalent strata is in danger of becoming passé. Why rely on biostratigraphic correlations hampered by pitiful sample sizes and questionable assumptions about isochronous taxon ranges on different continents when you can just use lots of radiometric dates, augmented by magnetostratigraphy, to determine which strata are the same age?

So this is the most likely (and most promising) future of chronostratigraphic correlation for Late Triassic vertebrate=bearing strata: Provincial lithostratigraphic and biostratigraphic models, in which the lithostratigraphy and biostratigraphy is worked out in detail for particular areas. The identification of isochronous strata is based almost entirely on radiometric dating and magnetostratigraphy, and the alpha taxonomy of vertebrate specimens is done solely based on morphological criteria regardless of the age or geographic distribution of specimens. In this way, the chronostratigraphic correlation of strata tells us whether or not the ranges of similar taxa on different continents was isochronous or not, alpha taxonomy tells us whether taxa were geographically restricted or widespread instead of being dictated by biostratigraphic convenience, and we avoid the sort of circularity in reasoning discussed in the Rayfield et al. paper. The age equivalence of vertebrates is derived from the ages of strata established by other methods, not vice versa. If the correlation of Upper Triassic terrestrial deposits indeed proceeds along these lines, this will make some of the issues discussed in the Rayfield et al. paper irrelevant.

Incidentally, marine invertebrate biostratigraphy is a bit more complicated, because it is intricately intertwined with the chronostratigraphic time scale. The current correlation of stage boundaries, at least those for which a boundary stratotype (“golden spike”) has NOT been driven in, is based almost exclusively on the biostratigraphic correlation using, particularly, ammonoid and conodont biozones. The correlation of ammonoid and conodont biozones is therefore tied directly to the correlation of chronostratigraphic units, and necessary to find the most suitable locations for chronostratigraphic boundary stratotypes. However, one a “golden spike” has been driven in, a geologic stage pulls free of biostratigraphy, and chronostratigraphic correlation can be based entirely on radioisotopic dating and magnetostratigraphy (if available).