Molecular Clock - Non-constant Rate of Molecular Clock

Non-constant Rate of Molecular Clock

Sometimes only a single divergence date can be estimated from fossils, with all other dates inferred from that. Other sets of species have abundant fossils available, allowing the MCH of constant divergence rates to be tested. DNA sequences experiencing low levels of negative selection showed divergence rates of 0.7-0.8% per Myr in bacteria, mammals, invertebrates, and plants. In the same study, genomic regions experiencing very high negative or purifying selection (encoding rRNA) were considerably slower (1% per 50 Myr).

In addition to such variation in rate with genomic position, since the early 1990s, variation among taxa has proven fertile ground for research too, even over comparatively short periods of evolutionary time (for example mockingbirds). Tube-nosed seabirds have molecular clocks that on average run at half speed of many other birds, possibly due to long generation times, and many turtles have a molecular clock running at one-eighth the speed it does in small mammals or even slower. Effects of small population size are also likely to confound molecular clock analyses; cheetahs for example, having gone through at least 2 population bottlenecks, could not be adequately studied based on a molecular clock model alone. Researchers such as Ayala have more fundamentally challenged the molecular clock hypothesis. According to Ayala's 1999 study, 5 factors combine to limit the application of molecular clock models:

• Changing generation times (If the rate of new mutations depends at least partly on the number of generations rather than the number of years)
• Population size (Genetic drift is stronger in small populations, and so more mutations are effectively neutral)
• Species-specific differences (due to differing metabolism, ecology, evolutionary history,...)
• Change in function of the protein studied (can be avoided in closely related species by utilizing non-coding DNA sequences or emphasizing silent mutations)
• Changes in the intensity of natural selection.

Molecular clock users have developed workaround solutions using a number of statistical approaches including maximum likelihood techniques and later Bayesian modeling. In particular, models that take into account rate variation across lineages have been proposed in order to obtain better estimates of divergence times. These models are called relaxed molecular clocks because they represent an intermediate position between the 'strict' molecular clock hypothesis and Felsenstein's many-rates model and are made possible through MCMC techniques that explore a weighted range of tree topologies and simultaneously estimate parameters of the chosen substitution model. It must be remembered that divergence dates inferred using a molecular clock are based on statistical inference and not on direct evidence.

The molecular clock runs into particular challenges at very short and very long timescales. At long timescales, the problem is saturation. When enough time has passed, many sites have undergone more than one change, but it is impossible to detect more than one. This means that the observed number of changes is no longer linear with time, but instead flattens out.

At very short time scales, many differences between samples do not represent fixation of different sequences in the different populations. Instead, they represent alternative alleles that were both present as part of a polymorphism in the common ancestor. The inclusion of differences that have not yet become fixed leads to a potentially dramatic inflation of the apparent rate of the molecular clock at very short timescales.