[MUSIC] You've now been through the necessary background stuff. So let's return to the molecular clock, which is essentially just a hypothesis that postulates a linearization between the number of nucleotide or amino acid changes in time as you have seen before. If the clock ticks with a perfect regularity, the length of all branches equal to the number of changes from the root to the tip will be the same, as can be seen on this phylogene. Similar in the sense that the shoulder's length leading from the root to any tip. Say the Araceae on this slide is equal to the length leading from the root to any travel on the tree. In technical terms, the tree is said to be ultrametric. However, more often than not, phylogenetic trees are not ultrametric, as can be easily seen on this slide. Just look at the very large branch, or long branch, leading to a course which is far longer than any other branch on this tree. But what are actually the components that are responsible for branch length? Well, branch length reflects two components that need to be separated, and this is difficult if the tree is not ultrametric. The number of changes and time. Looking only at the tree, there is no way we can separate them. As you perhaps remember, phylogenetic trees, no matter how they are constructed only depicts relative branching order. In the cases we have seen, changes of each branch and the relative branch order does not rely on the number of changes. Thus, because trees are often not ultrametric, we need to be able to separate the two components. To put it differently, we need to accommodate changes in rate equal to the number of changes per time unit in order to estimate the age of all groups. But how can we actually explain the molecular clock? Changes in molecules, like changes is organism's morphology, were originally thought to be only of two kinds, deleterious or advantageous changes. Advantageous changes, changes that increase the organism's survival, were supposed to be very rare. Whereas deleterious mutations, changes that decreases the organism's change the survival were much more common and will be quickly removed by selection. On the other hand, if there exists yet another third category of molecular changes, namely neutral changes. Changes that have no influence on the organism's survival ability which is, at least in part, supported by the partial degeneracy of the genetic code, then the interpretation would change dramatically. If the reason behind linearity is simply that most changes are effectively neutral, then the rate of change would be identical or approximately close to the well-known mutation rate. The number of changes for base and unit time. This interpretation was called the neutral theory. The neutral theory even explained the difference in slope, different rate of change between different kinds of molecules. High histones, who are very much lower than hemoglobins, simply because they had far fewer neutral sites. So let's look at within linear's deviation for linearity. Deviation from linearity, equals a non clock-like behavior, is a fact of nature, but what is causing it? Well, at least two issues might create deviation from a perfect clock. The imprecision of the ticking rate and changes in mutation rate. Let's look first on the clock's slobbiness. Obviously, molecular clocks do not tick with the precision of a standard watch. The molecular clock is, in its very nature, stochastic or random. Thus, variation in the ticking rate can be described as a mathematical function, a Poisson distribution, which allow for a considerable variance. There are many examples of observation events, etc., that follow a Poisson distribution. Such as, for instance, the number of road kills per unit of distance of a road, the number of pine trees in a unit of mixed forest, the number of spelling mistakes you make per page. So let's look at the easy one, spelling errors in pages. If you write one page of a text, it might be error-free or have about, say, 25 errors. All sorts of random factors may influence these. Error rates might increase because you were tired, the light was bad, you were disturbed by your noisy neighbors. There are endless number of reasons. If I write a page, there might always be errors, but perhaps never 25. I'm pretty sure you cannot tell me your average number of errors per page. I cannot tell you mine, at least. If you are to write 100 pages, and we both draw a distribution of how many times, 0,1,2, up to 25 errors occurs, we'll both have created a Poisson distribution. The Poisson distribution has many interesting properties. One is that the expected value, the mean value of the random variable equal to the number of mean errors after an infinitive number of tries is equal to the variance. Our individual, which is equal to the individual spread in the number of our errors, which may, of course, be different. The molecular clock behaves similarly. The main problem is that we never know how long time it takes before we can reasonably assure that we are hitting the expected value or the number of changes expected in a given time. And additionally, that the variance does not follow the classical Poisson distribution. Empirical observations have even suggested that the variance is greater than predicted by a Poisson distribution, and it is described as an overdispersed Poisson distribution. When it comes to the second reason, we will deal with the mutation rate, which directly influenced the clock speed. This rate might shift over time, and the balance between random genetic drift and selection may also change with a changing environment. These changes in rate might hit different sites or positions, whole genes or whole genomes. And changes at one place in the genome might strongly influence rate in other parts because of interactions between molecules and part of molecules. Interaction we often know very little about. Such changes are often called compensatory changes. One example is the three-dimensional structure of a protein, which may have profound influence on the functionality of the protein. Where there's changes in one part of the molecule increases the likelihood of compensatory change in another if the structure is maintained. Even bottlenecks effect. Rapid fluctuations in population size may change the selection of different traits, thus the fixation of neutral alleles is more likely in a small population than in a large. And the effect is enhanced during repeated interest through bottlenecks. Deviation from linearity is also line-specific. If you look across the tree where some of the branches deviate from clock-like behavior, this deviation is often restricted to certain branches, which themselves may show clock-like behavior. One possible reason might be the different groups have different DNA-repair mechanisms. They show simply different ability to repair mutations or other damages. Even different genomes may show different repair mechanisms. Last mitochondrial genome in both animals develop much faster than the mitochondrial genome in higher plants. And the mitochondrial genome higher plants devolve much slower than the plastic genomes. So there's a lot of difference. Mutation rates are also a function of generation time, body size, and metabolic rates. Organism with a short generation time have higher rates than organisms with a longer generation time. The same applies to body size. Smaller organisms have a higher mutation rate but often have a smaller generation time than larger ones. Additionally, exothermic or cold-blooded animals, which have a body temperature that is similar to their surroundings, have a slower rate than the endothermic or warm-blooded animals like us. Bottlenecks also tend to speed up the mutation rate, as was discussed previously. As you have now seen, there is a number of reasons why the clock is not necessarily exact. However, we have not yet torn apart the incremental factors behind branch lenght's change in time. We have to calibrate the clock to do that, to put absolute time on the tree. How do we put real time on the molecular clock? Well, two points fixed in time is needed to calibrate the clock. One of them, we will usually get for free, as we most often deal with extant organisms, but it is not necessarily so. To obtain the other fixed point, two very different methods are available. The fossil record and geological events, typically tectonic changes. Having a fixed point, a fossil of known age that can be placed with reasonable precision in the phylogeny, makes it possible to use a tree-wide point of reference, which can, in turn, be used to turn relative time into absolute time. But only with a reasonable precision if the clock is automatic, which it is not. It could, of course, be better to have more fixed point than just one, but that's rarely the case as fossils are rare, and fossils that can be placed with confidence in a phylogeny are even rarer. If the phylogenic tree is not ultrametric, you will have the model the tree shape such that you can approximate changes in branch length across the whole tree. This is a difficult problem which needs an array of assumptions that we'll not go into detail with. Interesting, most fossils usually do not have any traces of DNA left, and hence can only be placed in a phylogeny based on morphology. One example is a trilobite, as we have seen before, which is a huge group of organisms only known as fossils. Additionally, no trace of any soft tissue is present either, hence a result, very little information usable for placing a fossil in a phylogeny. Especially as most modern phylogenies often use character from old parts of the organisms. Another point worth raising is the fact that fossils only give minimum estimates of rising points. The branch may be way older than we think. One thing is that the literature is ramp with missing links. And in fact, the overwhelming majority of all fossils do not qualify as such. Typically, because they have characteristics not shared with any extinct organism. Thus, placing a fossil correctly is far from trivial. One of the very few fossils that qualify as a missing link is Archaeopteryx lithographica from the Solnhofen slate in Germany. This fossil can be placed anywhere around the branching point even on an alternative branch to birds. However, unfortunately, there is no time stamped on fossils. Time has to be determined from geological evidence, often using radiometric data or dating, which imposes two further uncertainty on time estimates. The equipment-related area and the [INAUDIBLE] difference between two datable layers within which the fossils is deposited. If you look at the covers at the top of this figure, you can easily see that they are surrounded by layers, 495 and 510 years old, giving a range of 15 million years in security. So this has to be added to the equipment-related uncertainty, which is fortunately far smaller. Datable geological events, like tectonic movements, raise of volcanics or volcanic islands, may also be also be used to date branching if one firmly believes that the geological events in question predates separation of two branches on the phylogenetic tree. But such age estimates are often a moot point. Thus, the traditional age of volcanoes may be relatively precisely determined, but if the volcanic island arises above a geological hot spot, there might have been islands on the present location before. However, sometimes the match is close to perfect, as you see on this slide with Hawaiian Honeycreepers. An even more shaky approach is to use previous determined ages based mostly on as other tags as fixed points than the ones that are being used on the phylogenetic trees or at least overlapping once from another tree. This is what [INAUDIBLE] has a high risk of building a house of cards. However, it is important to bear in mind that biologists generally tends to believe that geology represents rock solid evidence, but it is not so. Geology is a hypothesis driven as biology. You have to remember that it took geologists ages to be convinced of continental drift. Whereas biologists, and some geologists, were convinced partly based on biological evidence. However, in certain areas of the globe, like the Caribbean or the Indo-Malaysian area around Wallace's line, part of the evidence is being subducted. And New Caledonia is an island with largely unknown relationships to surrounding landmasses. Often, there is even considerable difference between the age determined by molecular evidence and fossil evidence. So what can the conclusion of this be? There is no doubt that a molecular clock hypothesis is much more complex than originally believed and that the single relationship between change and time is too naive. A lot of issues have profound influence on time estimates. What is not always presented to the reader is confidence intervals on the time estimates. These are often extremely large. However, it's beyond doubt that the molecular clock hypothesis is too good a tool to investigate evolutionary history to be abandoned. We will remain forever interested including age on our own lineage, dinosaurs, etc. And we would still like to know when birds and mammals branch, or from the big diffused conglomerate, we call reptiles, or even the age of Darwin's finches. The list is endless. Thank you for listening. I hope I have convinced you that both the biological clock and the molecular clock cannot replace your own watch. I hope you have enjoyed the talk, thank you. [MUSIC]
