[phylocom-user] Opposite results between Blomberg's K and phylogenetic signal in Phylocom (AOT)
humblebee at hotmail.co.jp
Thu Jan 16 18:46:36 PST 2014
Thank you so much for your prompt reply. I downloaded the original paper you mentioned, and had a quick look at it. I think I am slowly getting what K statistics and P-values mean, respectively. This is very helpful, thank you again!
> Subject: Re: [phylocom-user] Opposite results between Blomberg's K and phylogenetic signal in Phylocom (AOT)
> From: steve.kembel at gmail.com
> Date: Thu, 16 Jan 2014 21:06:20 -0500
> CC: phylocom-user at lists.phylodiversity.net
> To: humblebee at hotmail.co.jp
> both methods are telling you the same thing - there is more phylogenetic signal than expected by chance. For the K statistic, your P-value of P = 0.001 indicates there is more signal than expected by chance. Same thing for AOT. The absolute value of K = 0.29 tells you the magnitude of the signal is less than expected under a Brownian motion process, but it.FN"s still more than random according to the P-value. Have a look at the original papers by Blomberg and colleagues for an explanation of how to interpret K and the associated P-value:
> Blomberg, S. P., T. Garland, Jr., and A. R. Ives. 2003. Testing for phylogenetic signal in comparative data: behavioral traits are more labile. Evolution 57:717-745.
> Le 2014-01-16 .AN` 20:21, K O <humblebee at hotmail.co.jp> a Nicrit :
> > I've been trying to test for significant phylogenetic conservatism in some plant traits. I used R-package picante (Blomberg's K) and AOT in Phylocom. So far these two methods yield completely opposite results for the same traits, and I'm confused. Is it possible to get such very contrasting outputs from these two methods? In both cases, I am assuming that all the branch lengths = 1.
> > For example, one trait shows Blomberg's K = 0.2892 (P = 0.001), suggesting that the evolution of this trait overall has not been greatly constrained by phylogenetic relatedness. In contrast, for the same trait AOT yields variance of standardized contrasts across tree = 0.006, significance of contrast variance relative to null model (low rank) = 1, and significance of contrast variance relative to null model (high rank) = 1000 (i.e., P = 0.002), suggesting that treewise phylogenetic conservatism is significantly greater than the expectations of random tip swapping. (if I am correctly understanding)
> > I've just started looking at these phylogenetic signals, so I may be overlooking something critical. I will greatly appreciate if you could kindly give me directions, thank you very much!
> > _______________________________________________
> > phylocom-user mailing list
> > phylocom-user at lists.phylodiversity.net
> > http://lists.phylodiversity.net/listinfo.cgi/phylocom-user-phylodiversity.net
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the phylocom-user