Tuesday, July 1, 2008

How Old are Turtles? A Paleontological Perspective.

In a recent paper published in Journal of Vertebrate Paleontology, Danilov and Parham present an interesting analysis of two Middle Jurassic fossil turtle lineages. They estimate the phylogenetic relationships of these extinct lineages using discretely coded morphological characters, and bracket the estimated age for the crown node of all living turtles based on the oldest fossils. As the the figure from their paper shows, there has been debate as to the timing of turtle (Testudines) diversification. Previous estimates range from the Triassic-Jurassic boundary to the Late Jurassic.

Recent molecular divergence time estimates from Charles Marshall (presented at the 2008 Evolution meetings in Minnesota) and a team composed of Peter Meylan, Brad Shaffer, and yours truly result in an age for living turtles that dates to the Triassic-Jurassic boundary (our study), or well into the Triassic (Marshall's study). Regardless of this disagreement among molecular estimates and inferences from the fossil record, Danilov and Parham's paper presents a nice summary of the problem and a clever way to investigate the origin of turtles with the fossil record.

Danilov, I. G. and J. F. Parham 2008. A reassessment of some poorly known turtles from the Middle Jurassic of China, with comments on the antiquity of extant turtles. Journal of Vertebrate Paleontology 28: 306-318.


Glor said...

From what little I know of the fossil record, I have the impression that turtles have always puttered along with relatively low species-richness -- I've never heard discussion of the "Age of Turtles." Does the new study address patterns of diversification over time?

Anonymous said...

Not that I am aware of. First, there has not been a dense sampling of extant lineages in phylogenetic trees, and from what I can tell, there is a lot of material in the fossil record that is undescribed and/or unidentified.

That said, I agree with your observation. Given their morphology and uniform patterns of life history, I would guess that turtles have always been long-lived late maturing animals. Now a great question would be to investigate of D-M incompatibilities acumulate slower in turtles than other tetrapod lineages.

Brian R. Moore said...

Even without a complete (or comprehensive) phylogeny of extant turtle species or dense sampling of the turtle fossil record, it is clear that turtles are anomalously depauperate (this can be calculated by comparing the distribution of extant species diversity descended from contemporaneous tetrapod lineages to expectations generated under a stochastic branching model).

The relatively ancient origin and low extant species diversity of turtles—realizing a mere ~300 species over >220 Ma—suggests either an improbably low (and constant) rate of speciation or an improbably high relative extinction rate (which would more likely have led to the complete loss of the group).

Glor said...

I just remembered that turtles were part of the original stochastic diversity simulation analysis by Raup et al. 1973. They don't seem to be particularly anomalous there (see Fig. 7), but I think that's probably just because this study was based exclusively on fossils and turtles tend to be pretty good at leaving those behind.

Brian R. Moore said...

Rich is correct. The study of Raup et al. implies that turtles appear to have diversified at a stochastically constant rate (based on fossil occurrence data).

However, the rate of turtle diversification—although stochastically constant— is indeed anomalously low when compared to that of other contemporaneous tetrapod lineages (which was my previous point).

Glor said...

I didn't mean to disagree with your point that turtles are extraordinarily undiverse relative to other tetrapods. My point was that if one looks only at the fossil record of reptiles, turtles don't appear to have done that poorly (they have one of the fatter spindle diagrams of the 17 reptilian clades Raup et al. examined). Could be several explanations, including preservation bias or the use of family-level (or higher) taxa to reconstruct these diagrams.

Brian R. Moore said...

Rich, I didn't mean to imply that we were in disagreement :-)

I'm a big fan of the Wood's Hole studies, which pioneered the application of stochastic branching processes to inform our null expectations regarding patterns of diversification.

However, I also agree that their use of taxomomic ranks as the units of diversity and other inherent artifacts of the data make it somewhat difficult to interpret any particular empirical conclusions.

Glor said...

Too late dude, we're feuding now.

On problems with Raup et al. - let's not forget the fact that their null model included diversity-dependent cladogenesis. I'm not saying this model isn't appropriate, but the way they used it was arbitrary and based more on computational constraints than anything else.

Brian R. Moore said...

Oh, it is on! (somebody call 'nerd fight')

Granted, it's easy to look back on these seminal papers that were published 35 years ago and pick nits...you might also point to the fact that their simulations were implemented in discrete (rather than continuous) time, that the number of replications were only marginally adequate, that the scope of parameter space was limited, etc.

Despite these quirks and quibbles, you have to remember the state of play at the time. People pointed to any blip or dip in a taxonomic diversity plot/spindle diagram as the basis to weave elaborate causal narratives for 'dramatic radiations' and 'catastrophic extinction events' (cf LTT plotters of today).

From this morass of groundless data over-interpretation, the Woods Hole heroes emerged triumphantly to raise the level of discourse. First and foremost, they pointed out the central premise that diversification is an inherently stochastic process (i.e., such that two lineages diversifying under identical probabilities of speciation and extinction are likely to realize substantially different diversification rates).
And they developed a number of stochastic branching processes to generate expectations under rate constancy.
And wrote programs to implement these stochastic branching process models.
And performed a number of simulations under these models to evaluate empirical pattens.
Oh yeah, and did I mention they did all this in the early 1970's.

So, I'd be deliriously happy if—in 35 years time—any of my papers should fair as well as the Woods Hole studies.

Glor said...

What I love about this debate is that we don't actually disagree. The papers by the so-called "Woods Hole Group" are neglected classics. Way ahead of their time. Of course, the results were almost universally ignored or repudiated by paleontologists (who had a lock on data on diversification over time in the 1970s). I'm sure part of this had to do with their implication of stochastic processes, but I'm afraid that it may also have involved the hubris of the presentation.

Brian R. Moore said...

Geez Rich, you make it awfully hard to keep a 'feud' going...I have to admit that I enjoy the edgy tone of the Woods Hole papers. Their frustration with—and contempt for—paleontological conventions of the day are palpable and make for damned amusing reading. I think my favorite in this regard has to be Gould et al., 1973, but Raup & Gould, 1974, is also good.