User:Carl Boettiger/Notebook/Comparative Phylogenetics/2010/08/05

Comparative Phylogenetics

Main project page Previous entry      Next entry

Batch Data Some of the nexus files don't seem to read into R. Opening and saving the file in figtree solves this. Data Sets Labrid traits --Carl Boettiger 00:52, 11 August 2010 (EDT) Note: labels on parameter bootstrap distributions have reversed wrasses and parrotfish. I've updated the software so that the plotting function takes care of the labeling automatically (when these were made it had to be added manually, while the automation of the data conversions and the painting by clade reversed the label assignments). See User:Carl Boettiger/Notebook/Comparative Phylogenetics/2010/08/10. Gape distance (weighted by body mass) Gape distance (weighted by body mass) Protrusion distance weighted by body mass Protrusion distance weighted by body mass Body Mass Body Mass Jaw closing lever ratio Jaw closing lever ratio Jaw opening lever ratio Jaw opening lever ratio Jaw kinematic transmission Jaw kinematic transmission Based on nboot = 100 Various muscle mass traits do not seem to converge. Carnivora Traits Carnivore size by habitat: aquatic vs terrestrial Carnivore size by habitat: aquatic vs terrestrial No preference for any model. OU.1 fit has non-zero, though small alpha, which cannot be distinguished significantly from the Brownian motion model. No evidence for either of the regimes models. (Might compare ws2 to ou1 though) Brownian motion: <syntaxhighlight lang="rsplus"> sigma squared: [,1] [1,] 3.922782 theta: NULL loglik deviance aic aic.c sic dof -317.3379 634.6758 638.6758 638.7225 645.7971 2.0000 </syntaxhighlight> OU.1: <syntaxhighlight lang="rsplus"> alpha: [,1] [1,] 0.2056982 sigma squared: [,1] [1,] 4.099564 theta: $size 9.39714 loglik deviance aic aic.c sic dof -318.9566 637.9133 643.9133 644.0070 654.5953 3.0000 </syntaxhighlight> The coding of marine and freshwater separate, as well as a terrestrial regime does not converge under the wrightscape simplex method using the hansen initial conditions, but does just fine using the standard 1,1 initial conditions. Running now. Mammalia The huge mammalia data set takes a while even fitting the hansen model on the painting. bm loglik returns INF, ou1 returns loglik of -3727. the multiple regime model simplex routine fails to converge using the seed from ou2. Note the dramatic differences in the log body size optima found by the regime model. Certainly these differences must accomplish most of the difference in likelihood of this model. <syntaxhighlight lang="rsplus"> alpha: [,1] [1,] 0.01144819 sigma squared: [,1] [1,] 6.277772 theta: $size 1062 aquatic terrestrial volant 8.961080 254.067332 8.059478 -217.152860 loglik deviance aic aic.c sic dof -3721.305 7442.610 7454.610 7454.634 7491.439 6.000 </syntaxhighlight>