The figures below are from Felsenstein's paper "Phylogenies and the Comparative Method". I was wondering if there was a specific name for this effect where there is an apparant correlation that is actually the result of the data being structured into two separate groups, where there is no correlation within groups but an apparent correlation between groups. "Phylogenetic non-independence" doesn't seem specific enough.

enter image description here

  • $\begingroup$ Is this similar to Simpson's Paradox? $\endgroup$ – Oreotrephes Jun 22 '15 at 19:03
  • $\begingroup$ @Oreotrephes Not really, I'm wondering if there is a term for this specific phylogenetic effect $\endgroup$ – C_Z_ Jun 22 '15 at 19:12
  • $\begingroup$ @kmm "Phylogenetic pseudoreplication" is exactly what I was looking for, I just couldn't remember the name of the term. Thanks! $\endgroup$ – C_Z_ Jun 22 '15 at 20:57
  • $\begingroup$ what is X and what is Y? I mean what is plotted against what in the figures? (and no, the answer won't be coming from me ;-) $\endgroup$ – AliceD Jun 23 '15 at 0:38
  • 1
    $\begingroup$ @AliceD The axes are arbitrary. X and Y don't matter, nor do their units. The point is to show that the apparent moderate positive correlation between X and Y in the upper figure masks underlying structure. When you separate by 2 monophyletic clades (boxes and Xs), there is clear separation and within groups there isn't any correlation. $\endgroup$ – kmm Jun 23 '15 at 1:09

I've seen this termed "phylogenetic pseudoreplication", but I can't remember offhand where. I'll see if I can find it. Without a tree, the boxes and Xs essentially represent 2 data points. As Remi.b suggests, this is really just high phylogenetic signal.


The fact that species are not independent observations because of their specific phylogenetic relationship is sometimes called phylogenetic dependence (see its use in this paper for example).

Most often though, we talk about the existence of such phylogenetic dependence for a particular trait and we talk about phylogenetic signal for this trait.

I think (correct me if I am wrong) you're searching for a name to describe a situation where two quantitative traits correlate along a phylogeny (when you have one data point per species) but not within any of the species of this phylogeny. If this is the case, I don't think you will actually find a name for this pattern as it it is already quite specific. You could eventually say that there is an inter-species correlation between the two traits but not an intra-species correlation.

  • $\begingroup$ Thanks for the response. I'm looking for a term to describe what happens when there appears to be a correlation between two traits when you just look at the extant species, but actually the correlation is a result of having two separate clades, neither of which has a correlation within the clade. "Phylogenetic signal" doesn't really describe this, but it could be that there actually isn't a term for what I'm describing, in which case phylogenetic dependence is probably as close as I'm going to get. $\endgroup$ – C_Z_ Jun 22 '15 at 19:31
  • $\begingroup$ I think I got your question. See edit (which won't be super useful unfortunately). $\endgroup$ – Remi.b Jun 23 '15 at 1:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.