To the OP:
The underlying assumption of the Tree of Life is that everything descended only once, and the most parsimonious (simplest) tree is the correct one. The fewest number of changes to evolve from animal A to animal Z is presumed to be the way it happened. This is done by constructing a tree, using a mathematical formula to find the likelihood that it happened, again, based on an inherently biased set of assumptions, and comparing it to ALL OTHER POSSIBLE TREES. When there are a few species, with a few traits, it is possible to be done by hand. Looking at long genomic sequences makes this a much more arduous task. The real boogaloo, though, is the number of species you are comparing. 20 species means you're looking at least 8*10e21 (8 followed by 21 zeros) DIFFERENT trees.
Limitations in computing power only allow a certain set of these trees to be considered. The statistical considerations for these analyses also have their own set assumptions, which are taken to be true most of the time. Of course, with the vast number of species out there, it's bound to be wrong some of the time.
The underlying assumptions of constructing a tree of life are false in some cases:
The simplest explanation doesn't logically follow as the best one. It just seems that, given what we know about reality, most of the time, the shortest way is the way it happened. Loss-gain-loss of function, for instance, is a well known evolutionary phenomena, though parsimony would rule such a hypothesis as false out of hand (that is, without explicit data, it is deemed unlikely and not considered). Likewise, a A->G->A->T nucleotide substitution is, according to parsimony, less likely to occur than an A->T substitution, regardless of whether or not the A->G->A->T nucleotide substitution really occurred.
The other assumption- descent can only occur once- is being revealed to be not quite true, due to lateral gene transfer. This occurs when DNA from one organism successfully integrates with the DNA of another in a heritable fashion (in the case of multicellular organisms, the germ line must be affected). This is a pretty common phenomena in unicellular organisms, where they transfer large parts of their genome to other phyla. The equivalent for us would be an elephant coming up to you, splicing its DNA into you so that you grow a functional trunk, and all your offspring have a functional trunk.
As it is, this sort of genetic transfer isn't taken into account during tree construction, even though a species can have more than one "ancestor". Per the quoted bit in the OP, this information is simply omitted from the analysis.
It just means the the tree has sprung a shoot or shoots from the roots. We are still learning about everything so nothing is ever as clear-cut as it appears.
Not an apt metaphor.
More like one of the roots got up and switched places with a branch.
No they don't and that's a common misconception. Laymen tend to confuse theories with hypotheses, thanks in no small part to the language of science, which is not a very useful tool for communicating with laymen.
A hypothesis is a possible explanation for something that happens or happened in the natural universe. Hypotheses are tested by the scientific method, using empirical observation, experimentation, reasoning, peer review, and all of our other tools. When a hypothesis has undergone sufficient testing (the degree of sufficiency correlating with the extraordinariness of the hypothesis according to the Rule of Laplace, another of our tools), it becomes true beyond a reasonable doubt. (I borrow the language of the law, because as I may have already said, the language of science seems deliberately designed to thwart communication with laymen.)
At this point the hypothesis has become a theory. The essence of a theory is that its probability of ever being falsified is extremely low. Theories do not "come and go just like the wind."
Occasionally theories are falsified, but it happens so rarely that the canon of science does not come crashing down. What's far more common is that a theory is refined on the basis of data that wasn't originally available. An example is Newton's Laws of Motion. They were integrated into the Theory of Relativity, and have become one component of that theory which, for all practical purposes, remains unaltered except at extremely high velocities.
Hypotheses come and go, but theories are durable. Not eternal, but durable.
The Tree of Life is a durable theory to which some new details are being added.
Well, some underlying assumptions to the Tree of Life make it a best guess. The theory is, indeed, being tested, since assumptions that descent only occurred once is crucial to current tree construction. Otherwise the branches would fuse with themselves.
The theory is that the Tree of Life is a an accurate representation of how life probably evolved, and the hypotheses would be examination of any particular limb. Limbs get moved around all the time, though the underlying idea that this is a tree we're looking at typically goes unchallenged.
To state it another way, our theory is that evolutionary relationships can be represented as a tree, and there is a deductive way to realize that. The hypotheses would be the locations of any particular creatures on that tree, or the structure of the branches.
