Cretaceous cormorants -- really?

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

I've recently reported on the paper that makes a good argument against 
Cretaceous loons. From this perspective, it could be interesting that 
cormorants seem to be highly derived members of whatever fragment of 
"Pelecaniformes", according to both molecules (Fain & Houde) and morphology 
(http://www.senckenberg.de/files/content/forschung/abteilung/terrzool/ornithologie/balaeniceps.pdf). 
The oldest Cenozoic cormorants are from the late Eocene, like the loons and 
many others (source: Feduccia 1996). Naturally I haven't tested if 
calibration with Cretaceous cormorants makes their relatives impossibly old 
(or just the tongue twist). But let's have a look at the Cretaceous material 
referred to cormorants.

For, perhaps, some reason I always imagined in my infinite wisdom that this 
material consisted of tarsometatarsi. Only today did I check this in 
Mesozoic Birds. Behold, although I had read the whole book before, I was 
wrong! Everything Hope refers to Phalacrocoracidae is a scapula from the 
Nemegt Fm and a femur from the Lance Fm. Here's what she writes about these 
two bones (p. 367):

"Limited diagnosis -- Specimens are placed in Pelecaniformes based on the 
following feature: iliac facet of the femur entirely convex (unique). In 
addition, the lack of elevation of the trochanteric crest of the femur 
differentiates Pelecaniformes from many other birds."

So in case the scapula turns out not to be phalacrocoracid, there's no 
reason to consider it pelecaniform.

"Limited diagnosis -- The specimens are placed in Phalacrocoracidae based on 
the following derived features:"

Then follows a list of 3 to 6 characters of the scapula, depending on how 
one counts, and 6 to 14 or so unquantified characters of the femur; 
afterwards comparisons are made of the scapula to Pelecanidae and of the 
femur to Anhingidae and the late Eocene to early Miocene Plotopteridae; the 
differences sound like they were adaptations to foot-propelled diving in 
cormorants, and those between Phalacrocoracidae and Anhingidae are 
explicitely ascribed to this lifestyle.

The scapula from the Nemegt Fm and the femur from the Lance Fm (p. 368)

"Remarks -- Kurochkin (1995a, b) determined that this specimen came from a 
very large cormorant. My comparisons with diverse cormorants and other 
pelecaniforms are consistent with this determination. [...] The specimen is 
not yet described."

Therefore it is not illustrated, and this means that there is no indication 
of how complete the scapula is. Personally I would be interested of what a 
scapula of, say, *Judinornis* (a hesperornithian known from an isolated 
vertebra from the Nemegt Fm) looked like... One neornithine synapomorphy of 
the scapula is mentioned on p. 348, but not on p. 368: "the humeral facet of 
the scapula faces laterally or craniolaterally".

"Remarks -- In the absence of known derived features of the femur in 
earliest neornithines, the present specimen is referred to Neornithes based 
on unique features of Phalacrocoracidae."

Hear, hear.

"The head of the femur is damaged,"

This means that one of the two "pelecaniform" characters cannot be coded. 
I'm not sure whether the trochanteric crest counts as "elevated" or not 
(fig. 15.9 A), but then this one character is already not considered unique 
to "Pelecaniformes", see above, so I don't even need to mention the recent 
dismantling of this group.

"and the shaft is broken off just proximal to the condyles, with a fragment 
remaining that indicates the very large size and lateral orientation of the 
lateral condyle;"

Judging from fig. 15.9 A, I'll buy that. But fig. 15.9 C shows an extant 
cormorant femur. Its lateral condyle is large -- perhaps smaller than what 
the Cretaceous fragment indicates --, but only slightly laterally oriented 
from the long axis of the femur, and slightly _medially_ oriented compared 
to the sagittal plane of the animal! The latter is the case because the 
femur is strongly bent in cranial view. The Cretaceous femur, on the other 
hand, is completely straight in caudal view. I wonder if this counts. 
Mediolateral curvature is not mentioned in the list of phalacrocoracid 
characters, though, so perhaps it's an autapomorphy of some smaller clade 
that includes *Phalacrocorax pelagicus*.

"part of the popliteal fossa remains."

But this part doesn't contain any diagnostic characters of cormorants.

"A small but probably significant difference from extant cormorants is the 
angular truncation of the trochanteric crest. This feature precludes 
referral to either of the two major extant lineages of Phalacrocoracidae 
recognized by Siegal-Causey (1988)."

Well. I for one don't think we should expect a _crown_ cormorant in the 
Cretaceous anyway. :-}

Fig. 15.9 shows the presence of most or all of the other diagnostic 
characters. But I wonder how many of them are adaptations to foot-propelled 
diving, and therefore partially correlated with each other. Only 2 to 4 of 
the 6 to 14 femoral characters in the list are said to be unique to 
Phalacrocoracidae, but it isn't mentioned where else they occur. Certainly 
Hope would have noticed if the femur came from a hesperornithid or 
*Baptornis*, even though she doesn't mention any comparisons to them; but 
considering the fact that only two characters of the femur are diagnostic 
for "Pelecaniformes", the femur in question could easily drop out of 
Neornithes once it would be removed from Phalacrocoracidae. So what other 
foot-propelled diving birds were there in the Maastrichtian?

I see three possibilities. One are basal, perhaps flying hesperornitheans, 
like what *Potamornis* seems to have been. No femora or for that matter 
practically nothing at all is known of these animals. The other are the 
apparently-not-loons like *Polarornis* and *Neogaeornis* -- if the femur 
from the Lance Fm belongs in there, it doesn't need to be removed from 
Neornithes. The third is *Yungavolucris*, according to some interpretations 
of this isolated tarsometatarsus.

CONCLUSIONS

To settle this question, someone should:
- Find out where cormorants, and many others, sit on the family tree of 
Neoaves. This would have to be done both with molecules -- so that the 
calibration experiment can be repeated for both loons and cormorants -- and 
with morphology -- so that fossils from the Cretaceous to the Eo- or 
Oligocene can be safely inserted into that tree. Well, don't we all wish. 
Fortunately both approaches are being pursued and have yielded promising 
preliminary results (Fain & Houde; Livezey & Zusi).
- Find more fossils of Cretaceous birds, so we get an idea on how realistic 
my alternative suggestions are. Well, don't we all wish.
- Find out which parts of Hope's diagnosis of Phalacrocoracidae are 
adaptations to foot-propelled diving and can therefore be partly expected to 
occur in other foot-propelled divers. This can be done with today's 
knowledge of phylogeny, and has probably been done years ago.

While I am at it, one important feature that was said to tie *Polarornis* to 
the loons was its cnemial crest: in hesperorniths it's formed by the 
patella, in grebes by both tibia and patella, and in loons and *Polarornis* 
by the tibia alone. But if foot-propelled diving evolved more often than 3 
times, then some of these possibilities should have evolved more than once. 
It already has evolved 4 or more times within Neornithes. Does someone know 
what the cnemial crest of cormorants and finfoots look like?