Re: Declining pterosaur diversity

[David Marjanovic <david.marjanovic@gmx.at>]

----- Original Message -----
From: "John Bois" <jbois@umd5.umd.edu>
Sent: Sunday, November 28, 2004 4:47 PM

> > Nobody but you has even tried for the last ten years or more.
>
> Thank you...but let me decline the compliment.
> Penny, D. and M.J. Phillips 2004 The Rise of birds and mammals: are
> microevolutionary processes sufficient for macroevolution? _Trends in
> Ecology and Evolution_ Vol.19 No.10 pp.518-522.

Oh. Let me rephrase... "no pterosaur expert has tried"... :-)

> ...in which is discussed five models for the biotic turnover at and/or
> before the K/T boundary.  The authors propose inverting the traditional
> question of "When did the last pterosaurs and dinosaurs become extinct?",
> to "When did mammals and birds start replacing small dinosaurs and
> pterosaurs?" (This is a question I have often asked on this list.)

Fine, fine. Let's try to answer that question.
- An SVP meeting abstract from this year says that there probably never was 
an adult pterosaur with less than 40 cm wingspan. So no pterosaurian 
hummingbirds, sparrows, blackbirds, thrushes, titmice, and lots, lots more 
have probably ever existed.
- The biggest known Mesozoic mammals are from the middle-late Early 
Cretaceous. You have tried to build arguments on the amount of large-sized 
mammal species in later times, but the fossil record is still so bad that 
not only we don't have such numbers, we still risk making gross mistakes in 
inferring the ecology of "the tooth, the whole tooth, and nothing but the 
tooth". Remember the Stagodontidae, such as *Didelphodon vorax* from the 
Hell Creek Fm, which I and seemingly everyone else have proclaimed to be 
comparable to the Tasmanian devil in both size and ecology? Looks like we 
were wrong. Another SVP meeting abstract from this year makes a good 
argument for interpreting stagodontids as otter analogs, based on the 
discovery of... isolated tail vertebrae. I don't know of a dinosaur that 
could be interpreted as an otter analog, so it doesn't look like dinosaurs 
occupied that niche before mammals did.

> Fueling this is an observed increase in diversity of birds and mammals.
> They also supply some new data.  OK...the data suffer from the usual
> problems (preservational bias, etc.) but they indicate a reduction in the
> diversity of small dinosaurs from three time periods: E-M Cret.,
> Campanian, Maastrichtian.

Discoveries of dinosaur clades that were thought to have lived only 
elsewhere still happens in the Hell Creek Fm. I will not hazard a guess on 
what small dinosaurs lived in Africa or Australia or even southern Europe in 
the LK, and I certainly won't hazard a guess on what mammals could 
theoretically compete with a compsognathid... short of a fully formed fox 
(note that I'm talking about a relatively cursorial animal here).

> Lastly, the pushing back of placental evolution (you will recognize the
> authors from previous papers re molecular estimates of divergence) insists
> that these questions be asked.

I remember Penny from a paper from 1997... it made sweeping arguments based 
on a catastrophically misrooted molecular tree of Neornithes (don't ask me 
how it passed peer-review... in Science). But enough of this ad hominem 
"argument". Instead let's have a look on whether placental evolution is 
really pushed back by molecular data.

I'll start with deliberately picking the one paper that finds the shortest 
Mesozoic history for Placentalia. (Considering how lengthy the paper 
explains its methods, this subjective choice may not be all that bad.) It is

Peter J. Waddell, Hirohisa Kishino & Rissa Ota: A Phylogenetic foundation 
for Comparative Mammalian Genomics, Genome Informatics 12: 141 -- 154 
(December 2001)

These authors derived four different sets of divergence dates from their 
tree by using one of the four calibration points per set (which is a priori 
a bad thing). When they used the "tarsier/human split" at 55 Ma ago, they 
got only the following divergences in the Mesozoic:

--80--68--Paenungulata
  |   `--Afroinsectiphilia
  `--76--Xenarthra
       `--70--Supraprimates
           `--Laurasiatheria

(The word-monster Afroinsectiphilia includes Tubulidentata, Macroscelidea 
and Afrosoricida; Supraprimates is the same as Euarchontoglires of other 
authors.)

If (!) that calibration is correct, it tells us a few interesting things:
- The deepest divergences of placentals sort of look like they happened in 
anticipation of the K-Pg boundary.
- On a more objective note, not much diversification can have taken place 
during that short time. For example, if we want to find a more immediate 
sistergroup of (Xenarthra + Boreo[eu]theria) than what Afrotheria is, we 
need lots of good luck -- it would have had to appear within no more than 
four million years. When Xenarthra and Boreo(eu)theria diverged, they 
together were _more_ closely related to Afrotheria than chimpanzees are to 
us now.

Now let's ask whether the calibration is correct:
- Being based on fossils of presumed representatives of both lines, the date 
of 55 Ma is automatically too young. There's no way to quantify this, 
however. It might be too young by only a hundred thousand years or two if 
I'm optimistic -- in other words, it might be inside the error margin of the 
date itself.
- Oops, wait. It might be too _old_ -- if omomyids don't belong in the 
tarsier clade. Does someone happen to know how old the earliest anthropoids 
are? IIRC their age is similar, though...
- Regardless, the used model of molecular evolution does, unsurprisingly, 
not take the effects into account that factors like metabolism and 
population size have on the speed of molecular evolution. Small endothermic 
animals have, as a rule of thumb, higher metabolic rates and therefore 
higher mutation rates = faster molecular evolution. It is certainly 
reasonable to assume that K placentals were small -- or at least that only 
small ones survived the K-Pg boundary. Small animals also have, as a rule of 
thumb, larger populations and therefore slower rates of molecular evolution 
than larger animals with the same metabolic rates... it may or may not be 
reasonable to assume that small K placentals had smaller populations than 
small Cenozoic placentals, but it's certainly reasonable that the mass 
extinction event drastically reduced all population sizes! In sum, I can 
imagine that even those last 4 divergence dates will one day be crowded into 
the Cenozoic.
- One of the other calibration points gives ages that are just one to two Ma 
older. It is the rabbit/pika split at 42 Ma ago, which is rather badly 
constrained by fossils, so one could say that both calibration points give 
identical ages and thus corroborate each other -- respectively suffer from 
the same amount of systematical errors.

> The idea that often gets aired on this
> list--that there is infinite ecospace, or rather, an infinite number of
> niches created by division into smaller and smaller units when
> new species come along--must be held to question.

Oh, ecological niches don't always become smaller and smaller. Even if we 
totally ignore competition and predation, extinctions -- including mass 
extinctions -- still happen due to abiotic factors.

> So, to suggest some species become
> extinct upon the arrival of new species--even without evidence of this
> extinction--is not a radical claim.  And I am certainly not the only one
> making it!

That extinction can happen at the literal arrival of species that immigrate 
from elsewhere is indeed not a radical claim. The biggest problem I see, 
however, is that pterosaurs may well have retained a global distribution all 
the way to the end, and certainly had one in the EK -- which means an 
"elsewhere" may not have existed to start with.
       That extinction can happen at the figurative "arrival" of species 
that evolve in situ, however, _is_ a radical claim. (I'm not sure if it's 
your claim.) When an ecological niche cannot be created, speciation cannot 
happen, except perhaps (!) for extreme cases like when species only differ 
in the time of the year when they reproduce.

> By the way, the placental/marsupial split is now at around 125 million
> years ago according to SVP presentation by Luo,Z., Wible, and Yuan.

"Placental-marsupial" split is misleading, because both clades are 
crown-groups; the actually involved sistergroups (with stem-based 
definitions) are Eutheria and Metatheria. That their split has about the age 
that was expected from longer-known fossils tells us nothing about when the 
crown-groups appeared, except that that didn't happen much earlier than 125 
Ma ago. Indeed, fossils of Cretaceous placentals are still not known with 
certainty, and fossils of Cretaceous marsupials are not known at all.

BTW, some molecular divergence time estimates put the Metatheria-Eutheria 
split at a hundred and SEVENTY-five Ma ago, at the base of the Middle 
Jurassic. How's that? Fifty million years off. That's the difference between 
Messel and Rancho La Brea.
       I think some other studies used that crazy date to _calibrate_ their 
trees. But perhaps I'm just getting paranoid. In other cases, however, it 
has repeatedly happened that people used others' molecular date estimates, 
stripped of their margins of error and without the slightest glance at 
morphology, to calibrate their own molecular dates.