Wednesday, January 31, 2007

When phylogenetic names would be useful

To avoid being charged with being consistent (unlikely, I know), despite being underwhelmed by phylogenetic names in the context of TreeBASE (see conversation with David Marjanović in the previous post), I think they could be very useful in annotating phylogenetic trees. One of the things that impresses me most about Google Earth is the community that has developed around this tool (take a look at the latest issue of Sightseer). Lots of people are contributing content to Google Earth, and I think this is driven in part by the ease of creating the KML files that Google Earth uses, and the ease of referring to a geographic location (i.e., latitude and longitude).

Imagine having a similar tool for phylogenetics, where people could add annotations (such as "wings developed here", "oldest fossil found at this site", "pollination mode changed here", "change in base composition took place here", etc.). A bit like MacClade on steroids, or a community driven version of the TaxonTree tool developed by the University of Maryland's Human-Computer Interaction Lab (see their page on Biodiversity Informatics Visualization).

Leaving aside the issue of file format, we'd need a way of unambiguously locating a point on a tree. An obvious solution is to use least common ancestors (same concept as most recent common ancestors), as used in phylogenetic nomenclature (although not in apomorphy-based definitions, which strike me as a disaster waiting to happen).

It's in this sort of context that I think phylogenetic names show great promise. For some nice examples of using LCA's to identify nodes, take a look at the mor server (seemingly down at present, but described in a paper by David Hibbett and colleagues doi:10.1080/10635150590947104), and Michael Sanderson's program r8s. The mor package uses phylogenetic definitions of names to locate the corresponding groups in a large phylogenetic tree, and r8s has the MRCA command to attach age constraints to nodes in a tree.

10 comments:

  1. ...although not in apomorphy-based definitions, which strike me as a disaster waiting to happen.

    Having just used an apomorphy-based definition for a group of cephalopods (it's currently under review -- we'll see if the reviewers agree with you), I'd like to see your thoughts on them. Apomorphy-based definitions, that is, not cephalopods.

    Or maybe I wouldn't?

    ReplyDelete
  2. Andy, clearly I need to remember to engage my brain before sounding off (i.e., I hope people don't take my comments too seriously).

    My concern about apomorphy-based definitions is the uncertainty associated with mapping characters onto trees. Given a tree and two or more taxa, I can find the LCA without ambiguity. Given a character and it's states, I can't always guarantee this (alternative optimisations, alternative methods, stability when additional taxa are added). I don't deny there is a lot of appeal in being able to say that group x has feature y, I'm just not sure that this is a suitable way to attach names to a tree.

    Cephalopods, of course, are way kewl.

    ReplyDelete
  3. TaxonTree looks very interesting...

    My take on apomorphy-based definitions is that in a few cases they are preferable over the other types, but usually they should be avoided because missing data can mean that a clade cannot be unambiguously assigned to a larger taxon even though its phylogenetic position is fully resolved.

    The perhaps best example of where an apomorphy-based definition is feasible and preferable are the turtles (Testudinata sensu Joyce et al. 2004): in amniotes (whose ribs are always long enough), the shoulder girdle is either inside or outside the ribcage, there can't be an intermediate. Of course, few characters are that simple, and few names (Diapsida, Tetrapoda...) are so closely tied to a character that anything but an apomorphy-based definition would be inconsistent with historical usage and thus with Article 10.1 of the PhyloCode.

    ReplyDelete
  4. Turtles might be a fun case to explore in the context of phylogenetic databases, although sadly there aren't many turtle studies in TreeBASE. Thanks for the Joyce et al reference (doi:10.1666/0022-3360(2004)078<0989:DAPFTC>2.0.CO;2.

    ReplyDelete
  5. I take everything you say seriously!

    FWIW, I now remember it wasn't an apomorphy-based definition; it was an apomorphy-modified node-based definition for Decabrachia, which refers (if I have my way) to the most inclusive crown clade exhibiting a pair of tentacles synapomorphic with those in Sepia officinalis.

    I think this was preferable because we are confident in the monophyly of the group, but we don't know for sure who the sister group is (so it's hard to establish a good external specifier), and we don't have a good handle at all on relationships within the clade (thus, it's hard to pin down good internal specifiers that would ensure the name would refer to the appropriate set of organisms and only to that set). Too bad tentacles (along with most other cephalopod parts) don't fossilize well!

    I agree, though -- for what you're talking about here, an apomorphy-based or apomorphy-modified definition wouldn't be optimal.

    Cephalopods, of course, are way kewl.

    I obviously agree. But then why aren't there more cephalopodiatrists?

    ReplyDelete
  6. Well, I'm sure the prospect of being called a cephalopodiatrist would be enough to put most people off...

    ReplyDelete
  7. Well, I'm sure the prospect of being called a cephalopodiatrist would be enough to put most people off...

    Maybe they'd prefer "teuthologist"? That makes them sound like dentists...

    ReplyDelete
  8. But then why aren't there more cephalopodiatrists?

    Because there are too many cephalopods. You can't oversee them all. That drives people away. Why aren't there 10,000 times as many entomologists as there are?

    ReplyDelete
  9. Because there are too many cephalopods. You can't oversee them all. That drives people away. Why aren't there 10,000 times as many entomologists as there are?

    There are probably fewer than 1000 extant cephalopod species, so I actually think we have a decent number of cephalopod systematists to deal with that level of diversity. Plus, there actually are quite a few people that work on cephalopod fisheries issues, and many of them would consider themselves teuthologists.

    And the idea of 10,000 times as many entomologists -- justified by current levels of insect biodiversity or not -- sends chills down my spine...

    ReplyDelete
  10. And the idea of 10,000 times as many entomologists -- justified by current levels of insect biodiversity or not -- sends chills down my spine...

    Hehe. Same for me! :^)

    ReplyDelete

Note: only a member of this blog may post a comment.