Peter,

Both Kmult and its associated significance level are not effect sizes, so 
comparisons of these can be dicey. I would suggest examining their associated 
Z-score as obtained from the RRPP distribution and the observed value (see 
geomorph’s help file).

As to your second question, if one has species-level data, they are not 
independent. Thus one should take phylogeny into account, even if the 
phylogenetic signal is not overly strong.  To that latter point (having 
significant phylogenetic signal but less than Brownian motion), please see the 
review by Adams and Collyer 2019 (Ann. Rev. Ecol. Evol. Syst.). In that we 
explicitly discuss that issue. One alternative not often considered by 
empiricists is that such patterns could be obtained when the phylogenetic 
signal is concentrated in one or a few trait dimensions. We show some simple 
simulations demonstrating this is possible, which are helpful to keep in mind 
when interpreting such findings.

Best,

Dean

Dr. Dean C. Adams (he/him)
Distinguished Professor of Evolutionary Biology
Department of Ecology, Evolution, and Organismal Biology
Iowa State University
https://faculty.sites.iastate.edu/dcadams/
phone: 515-294-3834

From: morphmet2@googlegroups.com <morphmet2@googlegroups.com> On Behalf Of 
Peter Rühr
Sent: Wednesday, August 23, 2023 9:21 AM
To: Morphmet <morphmet2@googlegroups.com>
Subject: [MORPHMET2] Phylegenetic Signal in Graphs

Dear all,

I am studying bite curve shapes of 650 insect species. The bite curve graphs 
are described by polynomial models with six degrees, of which I can get 100 
graph point coordinates using the function predict(). Currently, I have two 
main questions regarding phylogenetic signal in that data:

1) Can I test for phylogenetic signal given the high non-independence of 
consecutive graph points?  I see two possible inputs for the 
geomorph::physignal() function, of which the latter seems to be more 
problematic in that regard: I could use the six polynomials + intercept per 
species, or the 100 predicted coordinates per species. The difference, however, 
seems marginal:

data                                       p                             Kmult  
  iterations
polynomials                        0.0001   0.105     10000
predicetd values               0.0001   0.113     10000

If you have an opinion on this, I would be very glad to hear about it.

2) As you can see, while being statistically significant, the Kmult-values are 
very low, indicating that species are less similar to each other than with a 
Brownian trait evolution. However, the Kmult-signal itself is in such a low 
range that some authors have decided to not take phylogeny into account 
(phylogenetic 'correction') in their own subsequent analyses. Should I correct 
for phylogeny or not? And again: would the polynomials instead of the predicted 
coordinates be a better choice for running phylogenetically informed analyses, 
or is such kind of data not suited to be phylogenetically corrected at all?

Best regards and many thanks in advance,
Peter

--
Peter T. Rühr

Bonn Institute of Organismic Biology (BIOB)
Section Biodiversity of Animals
University of Bonn
An der Immenburg 1
53121 Bonn, Germany

Phone: +49 (0) 228 73 5115
--
You received this message because you are subscribed to the Google Groups 
"Morphmet" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to 
morphmet2+unsubscr...@googlegroups.com<mailto:morphmet2+unsubscr...@googlegroups.com>.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/morphmet2/485d9c02-c088-477a-ab38-863316d19708n%40googlegroups.com<https://groups.google.com/d/msgid/morphmet2/485d9c02-c088-477a-ab38-863316d19708n%40googlegroups.com?utm_medium=email&utm_source=footer>.

-- 
You received this message because you are subscribed to the Google Groups 
"Morphmet" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to morphmet2+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/morphmet2/CO6PR04MB842723EED22EE076BE9571E7A2E1A%40CO6PR04MB8427.namprd04.prod.outlook.com.

Reply via email to