I have read that most scientist working in the field of human evolution date the split between man and chimp lineage to about 5-6 MYA but I read this post on John Hawks blog
http://johnhawks.net/weblog/reviews/gen ... -2009.html
which indicates that going by the molecular clock it could not have happend earlier than 4.5 MYA. I am just wondering does this mean that molecular clock evidence conflicts with fossil evidence? if so why is this? How accurate is the molecular clock?

If in doubt always bet on palaeontology

Net Traveller wrote:I have read that most scientist working in the field of human evolution date the split between man and chimp lineage to about 5-6 MYA but I read this post on John Hawks blog
http://johnhawks.net/weblog/reviews/gen ... -2009.html
which indicates that going by the molecular clock it could not have happend earlier than 4.5 MYA. I am just wondering does this mean that molecular clock evidence conflicts with fossil evidence? if so why is this? How accurate is the molecular clock?

Molecular clocks do not tick as regularly as some might think. There are numerous biases and sources of fluctuation and inaccuracy, and so they need to be regularly recalibrated as you go back in time - typically by reference to fossils, whose strata can be dated by the more regular tick of a nuclear decay clock. Even such factors as body size (which may actually be a proxy for a generation time effect) cause variations in the clock (for vertebrates if not for other groups). Temperature, false assumptions about the neutrality of the sites examined, species-specific variations in error-correction and detection or germline mutation differentials between the sexes, metabolic rate, multiple serial substitutions being counted as fewer (A->G->T shows up as one change, A->G->T->A as none), all cause the clock to be a real, but tricky method of dating - good for relative dating, not so good for absolute.

Allan Miller wrote:

Net Traveller wrote:I have read that most scientist working in the field of human evolution date the split between man and chimp lineage to about 5-6 MYA but I read this post on John Hawks blog
http://johnhawks.net/weblog/reviews/gen ... -2009.html
which indicates that going by the molecular clock it could not have happend earlier than 4.5 MYA. I am just wondering does this mean that molecular clock evidence conflicts with fossil evidence? if so why is this? How accurate is the molecular clock?

Molecular clocks do not tick as regularly as some might think. There are numerous biases and sources of fluctuation and inaccuracy, and so they need to be regularly recalibrated as you go back in time - typically by reference to fossils, whose strata can be dated by the more regular tick of a nuclear decay clock. Even such factors as body size (which may actually be a proxy for a generation time effect) cause variations in the clock (for vertebrates if not for other groups). Temperature, false assumptions about the neutrality of the sites examined, species-specific variations in error-correction and detection or germline mutation differentials between the sexes, metabolic rate, multiple serial substitutions being counted as fewer (A->G->T shows up as one change, A->G->T->A as none), all cause the clock to be a real, but tricky method of dating - good for relative dating, not so good for absolute.

I've actually once been told by a geneticist that the molecular clock dating of divergences is now so good, that palaeontologists should reevaluate our dates on fossils to match those indicated my genetics - one of those moments when you barely know where to start.

Indeed, given that biologists have known for some time that evolutionary processes can take place at variable speed, purported 'molecular clocks' need to be handled with particular care. One cannot simply assume that the rate of acquisition of mutations of any one single gene was constant, because there may have been circumstances in the past when that rate ran at a different speed. For example, a gene that is neutral in one set of ecological circumstances, may suddenly become strongly positively selectable when those circumstances change. Or, the reverse could happen - a gene that was once subject to strong positive selection could suddenly become neutral.

Of course, one can try and eliminate the effect of this, by considering many genes, and noncoding sequences, on the basis that tracking the mutational acquisition of many such sequences may show up instances where some genes in the considered set underwent changes in mutation acquisition rates. But there is a limit to how far one can take this approach. When you have evidence that, for example, there was wholesale duplication of important sections of a genome (which has been established to be the case for Actinopterygian fishes), then unless you can find non-duplicated genes unaffected by that event to use as part of your proposed 'clock', that clock is only valid after that duplication event. Unless of course someone can point me to papers containing a solution to this problem.

Well, the solution is carefully choosing neutral substitutions and there are statistical methods to do so. But the crux here appears to be the calibration. Fossil calibration ages tend to be too young, simply because you have the speciation event that they calibrate, but generally it then takes time for the diagnostic criteria we use to identify a member of one of the resultant groups as a fossil to evolve and then we don´t really expect the first individual that would be diagnosable to become a fossil and actually be collected. It´s basically Signor-Lipps in reverse here. And of course this means that any molecular clock date has to be taken with a heavily skewed error - it´s generally a tad too young again.
It´s also worth noting that a good calibration generally requires several good fossil dates of closely related groups to the one you are interested in and that the errors above are rather constant in an absolute sense, which means the relative error grows for younger splits.