[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Subject Index][Author Index]

RE: Romerograms... Some thoughts...



First up - that 'histomap' thing was amazing - agree with all that PZ said
on his blog (phyarngula), and awful chauvinistic biases as pointed out
between 'human vs. everything else' and 'some humans vs. other
humans'...yikes
...which is a shame, looking at it purely as a graphic, I was thinking how
much work went into it considering in the absence of digital 'cut and paste'
and Adobe Illustrator etc. that we are so used to these days (perhaps only
last 10 years that this really has become _common place_?) - I assume the
author would have had a large area on the floor that he shuffled small bits
of paper around on...

Secondly to stick my neck out on 'romerograms' (see below)...


> On Fri, 7 Dec 2007, David Marjanovic wrote:
> >> [...]
> > Considering avicephalans "thecodonts" was made possible by 
> the absence 
> > of tree-thinking. In precladistic times, all but maybe the real 
> > specialists on any group didn't imagine a group as a twig, they 
> > imagined it as a blob on a romerogram. Such blobs lack internal 
> > structure; the fact that no avicephalan fits into the 
> "thecodont" tree 
> > didn't matter because people simply didn't think of it. At 
> best, blobs 
> > on a romerogram consist of more blobs with stippled lines between 
> > them; adding one more little blob for *Megalancosaurus*, 
> supported on 
> > a stippled line that originates from a free-floating 
> question mark, is 
> > no big affair (if only you believe that
> > *Megalancosaurus* has The Defining Feature, the antorbital 
> fenestra).
> > [...]

I'm not sure if we can say 'tree-thinking' was entirely absent when
romerograms were popular (ye olde precladistic times), I think most people
realised they were crude, rough approximations of phylogenies in the absence
of a better system/method. Speciation, divergence, unlikelihood of
reticulations (successful hybridisation uncommon or unimportant) and
reversals, I'm sure were all important concepts most were aware of, and knew
the tree like pattern this would produce, but had no systematic way of
pulling out that level of detail. So, I've always assumed that the authors
of romerograms have always been aware that there was a 'real tree' that had
occurred, but without quantitative tools to tease that out of the data, they
composed their romerograms that only depicted level of detail of
relationships that they felt their qualitative methods could provide
evidence for - hence the blobs, stippled areas, dotted lines and question
marks. They didn't overstep the mark and say more than they felt they could
(BTW this should apply to both those of the precladistic and cladistic
eras!). The 'blobs lacked internal structure' because that was the
resolution of their confidence in the pattern, not because they didn't think
that that internal structure was actually there (sitting beyond their
methodological grasp).

I think the task romerograms were intended to fill was simple: illustrate
what was known about when fossil groups occurred, which coincided and which
didn't, sometimes with geographical information, to get a broad overview
summarised in a simple graphic. Of course they also intended to illustrate
which groups were related, and give basic phylogenetic information, but only
what they felt confident about. They were also aware of the incompleteness
of the fossil record.

Sure, this may have lead to inadvertently 'shaping thought' (despite an
effort to avoid it) and leading to unnatural 'groupings' (similar arguments
in Phylocode/unranked taxonomy vs. Linnaean System debate) or placing more
importance on a few key characteristics rather than someone using cladistics
would feel comfortable with - but I'm sure many were aware of such problems,
but did the best they could to avoid them.

And the width of a blob I'm sure was something they sweated over, knowing
it's limitations, inaccuracies, prejudices etc. The sorts of things they
wanted to indicate were (a) ecological importance* (which can involve body
size, which is what John Conway was getting at in his early post on this
thread when referring to biomass etc, as well as number of
individuals/species); or (b) diversity (no of species and possibly
problematic evaluations of degree morphological differences); based on (c)
data: amount of fossil evidence for a group (number of fossil species; or
possibly amount of fossil fragments/material in ecological works?).
Cladistics is more about filling the gaps between the data with hypothetical
relationships.

The difference between a romerogram and a modern cladistic phylogeny lies
not in the appreciation of darwian evolution it's author may have, but in
resolution of the hypothetical relationships between the data points each
method allows, and the fact that the latter is quantitative rather than
qualitative. Bear in mind, as we all should, that though cladistics can more
confidently construct more detailed hypotheses about the relationships
between the data points, they also are not the 'real tree' (though we hope
we are approaching it as more work is conducted).

Romerograms and 'Star' clades:
To convert a romerogram into something more like a cladogram (i.e. depicting
the same information using cladistic patterns) would result in each blob
becoming a 'star' radiation, and any question marks becoming basal branches
(I think the appropriate ref here is Falsenstein (1985) and those that cite
this paper). If you knew the number of fossil species that was used in
making the romerogram, you'd hopefully find the wider blobs becoming the
star clades with the greater number of branches (unless body size as in (a)
or morphological differences as in (b) were important in their version of
'width'). To a modern worker we understand these 'star' radiations not to be
what we think actually happened during the real evolution of the group, but
an indication of how much can be hypothesized based on the same data... but
perhaps to the author of a romerogram in pre-cladistic times it would appear
we were trying too hard connect the dots, and the blobs and question marks
would be more meaningful to them in depicting the same data (given the
absence of quantitative method to clarify it any further). I basically think
at this level of resolution either method would be as good as the other to
depict the data, depending on what you wanted show, and at this point
romerogram is probably showing more of what you do know than creating
patterns for what you don't (consider: you can be virtually 100% confident
that the star clades are artefacts, and only meaningful in statistical
analysis) - as here neither is any better than the other at revealing the
'real tree'.

With the benefit of modern quantitative methods (and more data (fossils),
the amount of which seems to be on an exponential rise in recent decades) we
can find that a lot of such 'star' radiations that the blobs represent are
false groupings produced by the previous lack of information or resolution
of the older method, not necessarily by false assumptions of the author. No
doubt some mistakes could have been made, not so much arising from
misunderstanding of evolutionary patterns and processes, but possibly more
from the unfortunate choice of the few key characters that the pattern was
based on (c.f. cladistic analyses can handle many characters - and there's
pages of possible discussion on that point).

No doubt some people did use 'romerogram-thinking' in an incorrect way,
however I would assume most were 'tree-thinkers', that in the absence of
quantitative methods, were being understandably conservative on the level of
detail to depict for the branching relationships when producing romerograms.

*ecological importance - I haven't had to worry about this, as I'm not an
ecologist, but I do know modern surveys struggle with such concepts - e.g.
gut content analysis of predatory sharks and rays will use IRI (index of
relative importance) for various food items where number of prey individuals
can't be determined and prey body mass is an important factor (I think - out
on a limb on this point). 

OK, these are just my thoughts (no doubt some wrong terminology, and stating
the obvious in parts - and just plain wrong elsewhere) - can't say I know
this is what the romerogram authors were thinking (I'm no expert on the
history of this subject) but gather that since many of the reasons I've
given here occurred to me as an average school kid, who hadn't heard of
cladistics, reading books of dinosaurs and evolution that had such diagrams,
I assume it occurred to these authors too. E.g. I'd figured that there would
be a 'real' twiggy pattern in those blobs and between them, but that those
details were beyond what was known - I actually remember thinking this
through after looking at the very superficial similarity of the crest of
Pteranodon and Parasaurolophus and having a brief mental hiccup 'hey they
must be related - hang on, of course not, they can't be' (OK, I was very
young after all)... that lead to lots of scribbling trying to work that
out...and so on... long story, not much sunlight involved.

Cheers,
Chris G

Reference:

Falsenstein (1985). Phylogenies and the comparative method. The American
Naturalist. 125: 1-15..