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

Re: Cladograms reflecting the newest data



Roberto Takata wrote-

What about Pisani et al. 2002 supertree approach to genus-level
dinosaur phylogeny? (Well Langer is one of the co-authors...)

Ah! A supertree! Run! ;)

You know, at first I was going to go on another spiel about how supertrees fail when used to analyze phylogenetic relationships, but then I realized they are ideally just the thing Evangelos was looking for- a summary of phylogenetic hypotheses. After all, I highly recommended my website's cladogram, which is just a summary of phylogenetic hypotheses with my bias influencing it. So why do I recommend my cladogram over Pisani et al.'s?

First, their supertree is outdated, as it only includes literature up till 2000 (except for three 2001 refs). The ones I recommended were from 2002 or later, and mine incorporates literature that isn't even published yet.

Second, it doesn't incorporate any preference for hypotheses based on newer data (source trees go back to 1980). Luckily, the number of published phylogenetic trees has increased greatly since the mid-nineties, so it's not as bad as it could be.

Third, the authors didn't incorporate any preference for hypotheses based on more or better data. So Currie's (2001) cladogram of Mongolian theropods gets counted as equal to Holtz's huge 2000 analysis of theropod relationships, even though the former was just supposed to illustrate how a few Mongolian taxa were related and wasn't specifically based on an analysis at all.

Fourth, it's constrained to only use relationships illustrated as a phylogenetic tree. So Sarcosaurus is a ceratosaurid, because Paul (1988) thought so, and only he placed it in a cladogram. For the same reason, we get Gongbusaurus and Agilisaurus as fabrosaurids (after Peng, 1997 I assume).

Fifth, my cladogram is more similar to one resulting from an explicit analysis of characters than a supertree, since I don't base its structure on what a majority of published trees say. Instead, I (admittedly subjectively) weigh the character evidence.

Of course, I recommend any of the studies I listed over Pisani et al.'s supertree, as they are actual analyses of character data, and the problems with supertrees have been addressed before (Gatesy et al., 2002; Gatesy et al., 2004).

Mickey Mortimer