Reptile & Amphibian Forums

The reliability of molecular clocks in reptile phylogenies?

Hi folks,

I read about molecular clocks and the Neutralist Theory and, the more I read, the more I’m getting curious about how reliable these models are when applied to reptile phylogeny to determine the evolutionary age of lineages (at least when using mtDNA for phylogenies).

I understand that the basic principal on which the molecular clock theory is based is the constant rate of neutral mutations (those that do not affect the phenotyp) in genes. Furthermore, rate of neutral mutation is also independent from population size (sensu Kimura), whereas in advantageous mutations population size has to be considered as well…

Well, today we know that there is nothing like a “universal molecular clock”. Slopes and upswings in mutation rates must be calculated for every lineage and a priori assumptions have to be made to “set” the clock.
I recall some papers on reptile phylogeny worked with the pseudo-standard rate of 2% MYA for mtDNA (e.g. Keogh et al., 2001), although the authors mentioned that more recent studies revealed slower or faster clocks (=levels of gene divergence) (e.g. Zamudio & Greene (1997) for allopatric populations of Lachechis (0.47 – 1.32% /My) or more recently Wüster et al. (2002) with 1.09 – 1.77% for the cyt b of the crotaline genus Porthidium. The latter rates were also used by Malotra and Thorpe (2004) for Trimeresurus.). So, at the end, how relieable is the usage of the 2% rate in Keogh et al. (2001). I recall a paper (ARBOGAST, B. S. and J. B. SLOWINSKI 1998) taking apart a priviously published paper from Klicka and Zink due to errornous assumptions and for not providing measures of errors...

What about all these side-effects that may have more or less impact on the rate of mutation in genes such as the negative body size correlation and some life-history traits (temperature), gene repair capabilities, selective pressure, advantageous mutations rather than neutral mutations and implied within population size, evolving geographic barriers and diseases that may kill most of a population, other traits of population dynamics, high UV-radiation causing more mutations, and others). How do we handle these? There are lots of stochastic models around but, who decides which of these side-effects (if any) may have taken place?

Cheers,
Wulf
-----
http://www.leiopython.de - the white-lipped python site -
http://www.herpers-digest.com - herp related eBooks search -

The short version:

Whenever a molecular clock is used, it needs to be calibrated for the taxa being used. There is no standard rate that can simply be assumed. This means you need fossils that can be used to date nodes on the tree, to give you a calibration that can then be extended to areas of the tree for which we have no fossils. However: fossils are in short supply and accurately placing a fossil on a tree can be problematic. Fossil dates are also often misinterpreted; for instance, some researchers make simple mistakes like assuming that the age of the oldest fossil of genus X is the date of the origin of genus X, when it is really only the minimum possible age of the genus. Generally, establishing maximum possible ages, or exact dates of origin for lineages, is difficult or impossible (you can't use a sparse fossil record to establish that a genus *wasn't* around), and as a result molecular clock estimates are generally minimum possible ages.

Nodes on a tree are also sometimes estimated based on geological or geographic events; for instance, if we have two sister genera that are separated by a mountain range, we can hypothesize that the two diverged at about the time the mountain range rose. This presents its own problems, but may be useful nonetheless. We also know that rates of molecular evolution vary not only between but also within lineages; we have no good way of eliminating this as a confounding factor in molecular clock estimates, short of including larger numbers of dated nodes (of course, if we had a lot dates on the nodes to start with, we wouldn't need the molecular clock estimates!).

Patrick Alexander

>>The reliability of molecular clocks in reptile phylogenies?
>>
>>Hi folks,
>>
>>I read about molecular clocks and the Neutralist Theory and, the more I read, the more I’m getting curious about how reliable these models are when applied to reptile phylogeny to determine the evolutionary age of lineages (at least when using mtDNA for phylogenies).

Different people have different believes about the accuracy of the molecular clock. In general, I think most scientists would agree that it is fairly accurate.

>>I understand that the basic principal on which the molecular clock theory is based is the constant rate of neutral mutations (those that do not affect the phenotyp) in genes. Furthermore, rate of neutral mutation is also independent from population size (sensu Kimura), whereas in advantageous mutations population size has to be considered as well…

Neutral mutations are those that do not affect the fitness of an organism. A classic example is the mutation of the third nucleotide in a codon, since the net result is no change in the amino acid being coded. Of course another well known example are some of the mitochondrial genes.

>>Well, today we know that there is nothing like a “universal molecular clock”. Slopes and upswings in mutation rates must be calculated for every lineage and a priori assumptions have to be made to “set” the clock.

Not all genes are suitable for ascertaining phylogenies, and certainly not all of them are suitable for use as molecular clocks. Adaptive characters are subject to natural selection, with chance playing a relatively small or even nonexistent role, so these characters are poorly suited for ascertaining phylogenies and/or for use as molecular clocks. I think many people simply erred in ignoring this simple fact, which is well known even to Darwin and his contemporaries, who distrust adaptive characters when dealing with systematics.

>>I recall some papers on reptile phylogeny worked with the pseudo-standard rate of 2% MYA for mtDNA (e.g. Keogh et al., 2001), although the authors mentioned that more recent studies revealed slower or faster clocks (=levels of gene divergence) (e.g. Zamudio & Greene (1997) for allopatric populations of Lachechis (0.47 – 1.32% /My) or more recently Wüster et al. (2002) with 1.09 – 1.77% for the cyt b of the crotaline genus Porthidium. The latter rates were also used by Malotra and Thorpe (2004) for Trimeresurus.). So, at the end, how relieable is the usage of the 2% rate in Keogh et al. (2001). I recall a paper (ARBOGAST, B. S. and J. B. SLOWINSKI 1998) taking apart a priviously published paper from Klicka and Zink due to errornous assumptions and for not providing measures of errors...

Molecular clocks have to be calibrated using the fossil record. So, a disagreement on rates is understandable. I think it is more important to figure out how these authors come up with the mutation rates they are quoting than how fast or slow they think a particular gene is mutating. The reliability of a particular molecular clock is therefore dependent on the soundness of the method of calibration.

>>What about all these side-effects that may have more or less impact on the rate of mutation in genes such as the negative body size correlation and some life-history traits (temperature), gene repair capabilities, selective pressure, advantageous mutations rather than neutral mutations and implied within population size, evolving geographic barriers and diseases that may kill most of a population, other traits of population dynamics, high UV-radiation causing more mutations, and others). How do we handle these? There are lots of stochastic models around but, who decides which of these side-effects (if any) may have taken place?
>>
>>Cheers,
>>Wulf
>>-----
>>http://www.leiopython.de - the white-lipped python site -
>>http://www.herpers-digest.com - herp related eBooks search -

These are all very good questions. When most individuals of a population has perished, the result is a genetic bottleneck, and it is well known that great changes in the genotype can be expected when a population goes through a bottleneck. Another factor that you mentioned, such as increased UV radiation causing more mutations, may not be important as these mutations may not be neutral. A frog egg that has been exposed to too much UV radiation, for example, may not survive to reproduce. Do remember that it is neutral mutations that are most likely to be useful for calibrating the molecular clock.

All of the postings before this are quite right. Unfortunately, there is no real basis as to the beginning of this molecular clock. The only thing it is based on is the rate of mutation, which is still highly up in the air. Some things can easily be traced/tracked, but we still have no idea or even suspicion of when other traits occurred. There are huge gaps in these molecular clocks that cannot be explained as of yet, and it is more of a divergence tool, which I think is what the original posting was concerned about. As to the various speciation events that occurred, we can only make an extremely rough geusstimate as to when they diverged. Hopefully, some day we will be able to create a rate of mutation molecular clock and be able to get some handle on it all.

Forgive me if this is a stupid question but I only have a high school education. Assuming we could even correctly calibrate this molecular clock all we can determine from it is rate of divergence of the non nuetral genes, correct? Then what does this really tell us? Does it only tell us how long there has been no gene flow? What good does that do? Isn't it the nonneutral genes that will ultimately allow the animal to adapt, change and become something else? Couldn't these changes occur anywhere along that timeline on the molecular clock?

>>Forgive me if this is a stupid question but I only have a high school education. Assuming we could even correctly calibrate this molecular clock all we can determine from it is rate of divergence of the non nuetral genes, correct? Then what does this really tell us? Does it only tell us how long there has been no gene flow? What good does that do? Isn't it the nonneutral genes that will ultimately allow the animal to adapt, change and become something else? Couldn't these changes occur anywhere along that timeline on the molecular clock?

Answer:

As I said before, only non-adaptive genes can be used for calibrating a molecular clock, because changes of adaptive genes (non-neutral genes) are constrained by natural selection. An adaptive gene can change if 1) the change is adaptive to a new environment, or 2) if the change has no effect on its function. An example of the second type of change would be a substitution of the 3rd nucleotide in a codon, which does not change the amino acid that this codon represents.

It is true that the molecular clock only tells us how long two populations have become isolated from one another. In fact, it does not tell us whether these two populations represent different species. In the simple case when two populations co-exist in the same area, the fact that they are isolated from one another is the strongest evidence yet that they represent two different species. Difficulties arise when two populations are geographically isolated from one another so that there is no chance for them to meet under natural conditions. In these cases, biologists must determine whether the two populations represent the same or different species. And indeed it is these cases of geographic isolation of morphologically similar species which is the source of a great deal of taxonomic controversy in the current literature.

Many biologists subscribe to the view that if two geographically isolated populations differ by a single morphological character (no matter how trivial), then this is evidence that these two populations are on "different evolutionary trajectories," which means they will in time diverge into two different species. Many other biologists do not subscribe to this view, because they do not accept the claim that a single, possibly adaptively neutral mutation or morphological variation as evidence that these populations are on different "trajectories." There is of course no middle ground between these two schools of taxonomic thought and therefore no possibility of compromise.

For this reason and others, taxonomy is currently in a chaotic state, and it is unlikely to be resolved soon.

Gist is, if we have a couple of fossils in a group of species we're interested in, but we want to know more about when species in this group diverged, we can use amounts of genetic divergence, coupled with molecular clock estimates, to fill in those dates.

Regarding neutral vs. non-neutral genetic loci: neutral (or nearly neutral) loci are good at letting us determine relationships. This is precisely because they are neutral; they change at more or less the same rate regardless of the rate of morphological or non-neutral site evolution, and so they give us unbiased estimates. Using genetic loci that are adaptively important lets us answer completely different questions. The main interest in studying non-neutral sites is in linking them to particular evolutionary changes. Those loci aren't good for inferring relationships, and, because they evolve at varying rates they can't be used in molecular clocks.