In a recent thread about unrooted trees, there was some discussion and disagreement about whether unrooted trees are informative with regard to evolution. One sees a fair number of minimum spanning trees and other networks in the literature, but it is not clear to me how to interpret these things without at least an implicit root (midpoint, long branch, whatever). I am curious to see what others think about this, and in particular how one can interpret an unrooted tree as a hypothesis about evolution.
An unrooted is a much less defined hypothesis of relationships (and thus of evolution) than a rooted tree. It is thus more general, and contains several different more explicit (rooted) hypotheses. For instance, for a problem of 4 taxa only admitting bifurcating trees, there are three possible unrooted trees, which contain fifteen rooted trees depending in which node you place the root. This is old useful cladistics theory. What is the utility of unrooted trees? Since all reconstruction approaches (parsimony, ML, Bayesian) do their searches using unrooted trees and later set the root using the outgroup(s), you might consider that unrooted trees are just part of the computation, thus a sort of methodological issue that you don’t need to understand or worry about. However, in my view, familiarity with unrooted trees might be useful to interpret rooted trees with a more open mind. Examining a rooted solution with this view allows you to consider alternative hypotheses should the root be placed in a different node (if supporting characters for the preferred hypothesis are not so strong). So, I think that being familiar with unrooted trees contributes to minimizing (if ever possible) the “black box” effect of computer programs for reconstructing phylogenies.
It is a good question and based on what I have read and practice it means very little in evolutionary terms. I think you can still say that the linkages between taxa are valid in thier own right but you cannot infer any evolutionary relationship.
From my experience, an unrooted tree determines the common ancestors of the taxinomic units but not the direction of the changes. The direction could be determined by the use of an outgroup.
An unrooted tree means nothing in evolutionary terms. All interpretations about relationships depend on the placement of the root (or assumptions thereof). Without a root, a tree is dead.
I don't believe in unrooted trees. They are just analytical artifacts that have no meaning to me.
An unrooted tree does not define clades but rather split the taxa into two sets that are.
The main reason that you need to understand and use unrooted trees is because most phylogenetic analysis methods yield unrooted not rooted trees.
Not an easy problem.
In theory unrooted trees are useful (and informative) when you wish to draw a network of relationships between units (species or populations). But not in a goal to draw an evolutionary history.
Anyway, if you consider DNA, every actual species possesses exactly the same "time of evolution" It means that you can't define "strictly" an ancestor (you may solve such problems using a clever study using morphological characters). Then, without a certainty of which species have to be include (outgroup) as ancestor, why not to consider that we may only present an image of relationships between species (unrooted tree) and not the "true" phylogeny of a group of species.
My point of view is that when dealing with populations (using microsat, ISSR...) to present data as unrooted tree is the correct way. When you work on species in an evolutionary purpose, trees have to be rooted.
Hoping it helps !!
Luc
Hi Luc,
If you consider an unrooted tree/network (e.g. of populations) as an image of relationships, what do two 'sister' branches mean? These might not be sister at all if the root is placed on either branch. Do you perhaps consider an unrooted tree/network to be an image of overall similarity? In other words, do you consider unrooted trees/networks to be non-historical? If so, it follows that we should avoid the term 'relationships' when discussing branches on an unrooted tree/network, as these are really just degrees of overall similarity.
Sober (2000) Philosophy of Biology, 2nd Edition:
"In ordinary parlance, "evolution" means change. If evolution is understood in this way, then the theory of evolution should provide a global account of cosmic change. Laws must be stated in which the trajectories of stars, of societies, and of everything else are encapsulated within a single framework. Indeed, this is what Herbert Spencer (1820-1903) attempted to do...The idea of a tree of life obviously entails the idea of evolution. If human beings and chimps have a common ancestor, then there must have been change in the lineages leading from that ancestor to its descendants. But the tree of life hypothesis says more than just that evolution has occurred...If we described the tree of life in some detail, we would say which species are descended from which others and when new characteristics originated and old ones disappeared. What is left for evolutionary theory to do, once these facts about life's pattern are described? One task that remains is to address the question of why. If a new characteristic evolved in a lineage, why did it do so? And if a new species comes into existence or an old one exits from die scene, again the question is why that event occurred. Answers to such questions involve theories about the process of evolution. As we move from the root to the tips of the tree of life, we see speciation events, extinctions, and new characteristics evolving. What processes occur in die tree's branches diat explain these occurrences?"
If evolution is change, we can only know the "direction" of change through a root.
Hello George and folks
when similarity become...proximity.....relationship....If I understand well your purpose, I think that we've got a rather close position but may be with some different acceptation of words/terminologies.
Anyway, it seems that we are barely agree that unrooted trees are suitable to illustrate historical evolutionary events.
Cheers from Luc
An unrooted is a much less defined hypothesis of relationships (and thus of evolution) than a rooted tree. It is thus more general, and contains several different more explicit (rooted) hypotheses. For instance, for a problem of 4 taxa only admitting bifurcating trees, there are three possible unrooted trees, which contain fifteen rooted trees depending in which node you place the root. This is old useful cladistics theory. What is the utility of unrooted trees? Since all reconstruction approaches (parsimony, ML, Bayesian) do their searches using unrooted trees and later set the root using the outgroup(s), you might consider that unrooted trees are just part of the computation, thus a sort of methodological issue that you don’t need to understand or worry about. However, in my view, familiarity with unrooted trees might be useful to interpret rooted trees with a more open mind. Examining a rooted solution with this view allows you to consider alternative hypotheses should the root be placed in a different node (if supporting characters for the preferred hypothesis are not so strong). So, I think that being familiar with unrooted trees contributes to minimizing (if ever possible) the “black box” effect of computer programs for reconstructing phylogenies.
Some excellent answers above. I just want to make one point explicit. When you draw an unrooted tree, you aren't really depicting evolution (in the sense of change over time) but only relationships among the taxa. When you have a rooted tree, however, you have an explicit depiction of time - which taxa emerged before which in the lineage you are studying. That is why you typically use an outgroup because you know that it emerged before any of the taxa in the ingroup, from other studies. Furthermore, if you have data for calibration, you can then even estimate times of divergence between taxa of interest.
Jan is correct in clarifying that the direction of the time element is missing in an unrooted tree, and not change over time itself. The direction of time is what I was talking about when I referred to the order of emergence, but thanks for clarifying my supposedly explicit answer :-). Clearly, change over time is depicted even in an unrooted tree, as internal branches can differ in length. I was referring to the order of emergence when I said change over time.
As for relationships, though, I think you may be splitting hairs. Yes, a monophyly can be undone if the root is within it, and in the same manner the sister relationship between two monophyletic groups can also be lost due to rooting. Thus, what appeared to be a monophyletic relationship in the unrooted tree may not be "truly" so (once a root is determined). But other than the disruption of relationships among the clades (or within a clade) created by the position of the root, the relationship among all the taxa within each clade still remains the same as in the unrooted tree. My point was that these relationships are all that an unrooted tree depicts. The addition of a root gives us the order of emergence of (groups of) taxa, but the relationships (in the sense of who is more or less closely related to whom within each clade) is seen in the unrooted tree as well. Finally, my reasoning for calling your objection "splitting hairs" can be summarized by this - doesn't a "split" (sorry about the pun here!) refer to a relationship - the two taxa are related due to the divergence from a common ancestor?
A common ancestor does imply a root, though, and every split has a common ancestor, by definition. The presence of a common ancestor also automatically gives directionality to the descendant taxa (again, by definition). In a rooted tree, each common ancestor is now the root for the descendant taxa, but my point is that this is true for an unrooted tree as well, except where disrupted by the root.
But maybe we should be having this conversation off-line? :-)
OK, let's try this with an illustration of an unrooted tree (see attached picture). I have annotated it with dots and arrows. Now as far as the blue dot is concerned, you are absolutely right that we do not know the direction of evolution (time), and that it could be in either direction as depicted by the two blue arrows. However, if you consider the red dot, as far as species 15, 9, 6, 7 and 8 are concerned, the direction is only one way - the one shown by the red arrow. These five species form a clade, irrespective of whether it is a rooted or unrooted tree, unless of course, it is disrupted by the root.
Sudhindra
In your figure, it is rather evident that the evolutionary direction goes from your species 4 as the ancestral one. Finally in this case to present the tree rooted or unrooted is just a question of design.
The problem is more then you've got trees such as the one attached. We compared there various populations/subspecies/sister species among a genus (to please Andrew) of butterflies (Nymphalidae) using molecular markers. The individuals in red and green are supposed to be the ancestral ones. With such results, we will never claim that they are the ancestral ones just that considering molecular relationships between these taxa.....all are exactly the same and the artificial splitting of names is not valid.
Cheers from Luc
I'm afraid I don't quite follow, Luc. Are all your OTUs extant? If so, how are the tips ancestral?
- Badges
- Science method
- Similar topics
- Biological Science
- Evolutionary Biology
- Phylogenetics
- Phylogenetic Analysis