"Close correspondence between quantitative- and molecular-genetic divergence times for Neandertals and modern humans"

This week PNAS publishes a study* by Weaver, Roseman and Stringer on Neandertal and modern human divergence via exploration of the morphological data, which is a follow up to an earlier paper by the same team**. The paper claims to show no conflict between morphology and molecules: this is a subject that interests me greatly as not only is it a big biological question but also relates to my own study of New World monkeys. The morphology vs molecules argument is basically this: with DNA sequencing abundant information came available for inferring evolutionary relationships of taxa/groups and molecular studies absolutely routed those based on morphology. Often the molecular approach showed morphology-based assumptions to be misleading at best or wrong at worst. This has led to an amount of arrogance in the molecular genetic community, with undergrad courses pretty much saying that the molecular is vastly superior to morphology as the basis for figuring out evolutionary relationships. The problem with this is that; (1) molecular approaches are far from perfect (unknown element of error in sequencing, a gene tree is not a species tree, over emphasis on single genetic markers like mtDNA, etc); (2) morphological study has come a long way since 30 years ago (and in the future geometric morphometrics may yield a lot of power); (3) DNA is (generally) lacking from fossil groups, therefore whether you like it or not we rely on morphological study to fit past groups into the scenario of the present. Of course morphological analysis has all sorts of problems, especially with homoplasy (i.e. similarity between groups that is not a result of common descent) but that may also be a factor in DNA based study to a degree not to mention the problem of horizontal genetic transfer.

Now the "conflict" is clearly that sometimes analysis of the two sources of information gives alternative results. Often this means that morphologists are running around like headless chickens trying to recapitulate the results of molecular studies. One area where there is a little more respect for morphological study, and some worries about the molecular clock (the alleged constant change in DNA sequence by mutation over time that can be used to predict how long separates individuals or groups from a common ancestor), are divergence of lineages/groups. The fossil record is needed to calibrate the molecular clock, and there seems to be a growing tide of people unhappy with the "constant rate" of the molecular clock (e.g.***). To get to the point: morphology is useful. Hurray!

The Weaver et al. (2008) paper relates to a bit of a minefield, namely the separation between the two most recent "human" groups namely Neandertals and modern humans. Although I could probably drone on about this all day the most well supported view, as originally proposed by Chris Stringer who is co-author on this paper, is that of Neandertals as a European hominid radiation distinct and separate genetically from the modern human Homo sapiens sapiens that originates from the second 'Out of Africa' radiation. Although there are sites indicating overlap and probable co-existence, some of the multi-regionalists and admixture fans such as Milford Wolpoff and John Hawks favour a different view: more gene flow between groups, deeper ancestry, a regional continuity in morphology representative of interlinked genetic units throughout the past million plus years. John Hawks also favours, with his recent paper**** on "accelerated human evolution" (which in my opinion shows only evidence that our species has had a massive population increase and says nothing about accelerated evolution... but that's a whole different rant!).

Hopefully the background to the paper is now set in a certain context: morphology-molecules and Neandertal-human. Well there is also a more specific element: are craniofacial differences in humans a result of evolution by natural selection/adaptation or random change namely genetic drift? As mentioned earlier the Weaver et al. trio have published on this issue before, as have Rebecca Ackermann and James Cheverud***** (who looked at early Homo and the australopithecines and found evidence for morphological evolution mainly by genetic drift- somehow Weaver must have missed this paper as they don't cite it). Of course selection and drift can both have an effect, but the question is more how much of morphological change is caused by each.

Weaver et al. (2008) first test a neutrality model of morphological evolution in Neandertals and modern humans, and once they show neutrality they use that very neutrality to estimate divergence between the two groups. The techniques are pretty much taken from those used in quantitative and population genetics, but altered for use with morphology rather than molecules. I find it hard to comment on these procedures as I honestly don't have a good enough understanding, but it will interesting to hear from some population geneticists about the suitability. If we do accept that the cranial measurements use measure neutral regions between the two taxa (thats a big if, and the question of two taxa against one requires another big if) then we can open our minds to the calculation of divergence time: 311ky or 435ky depending on mutation-drift assumptions (plus pretty big confidence intervals of 150ky in each direction).

The good thing about the hominid fossil record is that we can go out and test this big divergence range, although that will depend on how well preserved other fossil groups are. Also, the 311ky estimate that I think the authors prefer falls very close to recent(ish) genetic analysis****** of 370kya. So perhaps this is some strong congruence between morphological and molecular analysis. But it may just be congruence between specific measurements used and the genetic, although the authors point this out and call for more work. My worry is that those people with alternative views will take whichever divergence time they prefer so that this doesn't necessarily take us closer to agreement on topics of neutrality and divergence. One good idea could be to examine this as a test case in modern primate groups where the molecular dates are strongly supported, and then develop morphological tests of congruence (something I am considering doing with New World monkeys perhaps). Also, whilst I can see why you would use traditional craniometric measurements (because a lot of past data has been taken in that form) I think we need to start thinking about geometric morphometric shape analysis- yes I am obsessed because I work with it, but I feel like it gives that little bit more information that analysis can then handle.

Overall it is a very interesting paper. As a final thought (and this is a really fuzzy idea that I'm only just coming too) I wonder whether this sort of analysis could integrate comparison to an outgroup to ensure you aren't over inferring neutrality when really you are sampling two groups of the same taxonomic unit?

NB Kambiz# and John Hawkes## also have good posts on this, the former especially dealing more with the experimental issues that I sort of skimmed over because they scared me. Although John Hawks has a vested interest in the area his post is definitely worth reading.

* http://www.pnas.org/cgi/content/abstract/0709079105v1
** http://dx.doi.org/10.1016/j.jhevol.2007.03.001
*** http://dx.doi.org/10.1016/j.tree.2006.11.013
**** http://www.pnas.org/cgi/content/short/104/52/20753
***** http://www.pnas.org/cgi/content/abstract/101/52/17946
****** http://www.pnas.org/cgi/content/abstract/101/52/17946

# http://anthropology.net/2008/03/19/correlating-genetic-and-morphology-evidence-to-pinpoint-a-divergence-times-for-neandertals-and-modern-humans/
## http://johnhawks.net/weblog/