[Taxacom] More on Map of Life

John Grehan jgrehan at sciencebuff.org
Thu Nov 18 15:25:57 CST 2010

More comments on Kidd's Map of Life method.


It took me a couple of hours to digest the methodology section (I admit
to not being a quantitative theorist so what I write below might be a
load of rubbish).


In reading on the The Geopghyletic Data Model section I noticed that the
first two and a half pages of text were largely devoted to technical
question of georeferencing distributional data - a challenge that is
general to any method interested in location (including


The cartographic section is followed by the placement of 'internal
nodes' (p. 745) and this is where the method starts to get a bit hairy -
at least for me. As the author notes, the node could be placed sung
direct evidence (of what is not said) from the fossil or historical
records (not sure that those are yet), or using output from an
analytical historical biogeographic reconstruction algorithm (not
defined yet).


Those intangible possibilities aside, Kidd decided to use for his
example the same centroid positioning approach - at the geographical
centroid of their immediate daughter nodes. While this is
straightforward enough, the empirical nature of that node is not (to
me). Kidd describes this method as "very crude", but one that "can be
justified in the absence of other information if, for example,
diversification is the consequence of vicariance and the isolating event
is postulated to have occurred somewhere between the geographical ranges
of the sister clades". 


So the inference seems to be that the node in a phylogeny has some
specific spatial locality with respect to various information that
includes (but is not limited to, the others not being specified yet) a
theorized isolating event "somewhere between the geographical ranges of
the sister clades".


Or if Kidd postulates that they represent centers of origin that are
"most likely" (no empirical evidence given) to have existed between the
daughter clades".


But then in the absence of these Kidd proposes weighting nodes by branch
lengths, but why this would be informative about anything was not said
(or was not apparent to me).


Having characterized nodes as something that can be placed according to
any number of criteria, Kidd next moves to "the spatial path of links
between nodes" which again are to be determined by a variety of ways.


He notes that geophyletic branchs trace paths through space and through
time and thus may be composed of a number of vertices (segments or
parts). In his worked examples four vertices were inserted along each
link which (and this is where it gets interesting) "simply followed the
shortest geographical path between nodes". I found this quite
astonishing given the opposition Kidd received to acknowledging
panbiogeography and yet here he is being allowed to publish on a
panbiogeographic technique (the minimal spanning link). The contrast is,
of course, that the minimum distance links in panbiogeography are made
between localities. In Kidd's method the links are the minimum
geographic distances between theorized nodes that are themselves placed
in a hypothetical intermediate distance between the subordinate sister
taxa. Thus, Kidd's method attempts to graph a non-empirical entity (a
geographically placed phylogenetic node) with a non-empirical location.
At least that is what it seems to boil down to.


Additional non-empirical elements are introduced such as dispersal
theories to reject the minimum distance link (in panbiogeographic
mapping the minimum distance link may also be modified with respect to
specified topographic features, but not with respect to theorized
historical events).


In his worked example for bears Kidd takes the view that since the
biogeographic pattern is 'believed' to be the result of diversification
in Pleistocene refugia, ecological niche and palaeoclimate theories
could be used to somehow locate the node positions and branch paths. 


I regard the mapping of phylogenies onto geographic distributions as a
legitimate challenge for biogeographic analysis and while Kidd's
approach utilizes some nice GIS technology, it seems to confuse or imbed
empirical observations with highly theoretical assumptions that seem to
render the result very problematic indeed. 


Kidd notes that "GIS and other visualization technologies are
revolutionizing the exploration and presentation of large complex data
sets" but I wonder if this will be another case (like that in molecular
clock theory) where technological advances are confused with conceptual
advances, as if the former automatically promotes the latter. In Kidd's
approach there are elements of promising interest, but their potential
is lost among the traditional theorizing of Darwinian biogeography that
fails to recognize distribution location as an independent empirical
foundation for biogeography. As in Darwinian biogeography, Kidd's method
subordinates empirical space/time (as distribution) to theorized


Interestingly, Kidd seems to recognize the problem of data integration
resulting in a lack of information in a "confusing spaghetti". He
suggests breaking the geophylogeny into subclades or use other criteria
to break the data. And yet when panbiogeographic methods are able to
combine individual distributions into coherent composite patterns
(generalized or standard tracks) they are ridiculed (at least in the
English-speaking world).



Dr. John R. Grehan
Director of Science and Research
Buffalo Museum of Science
1020 Humboldt Parkway
Buffalo, NY 14211-1193

email: jgrehan at sciencebuff.org
Phone: (716) 896-5200 ext 372
Fax: (716) 897-6723


Ghost moth research

Human evolution and the great apes


More information about the Taxacom mailing list