There are couple of major problems here:

One is the representability of the fossil record. Imagine a species accounting for billions of organisms during the species existence for example for 10MY and producing fossil record. Often this species would be represented by few specimens scattered in time. I believe that for an individual the probability to be preserved in sediments AND to happen to be examined later by a paleontologist would be 1-E10 or less. If a speciation takes but a small faction of time (according to all modern models of speciation), this would mean that capturing the transition is next to impossible.

Still, if a species was so numerous and well represented, that there is still significant hope to detect the transition, we still would be in trouble claiming success in observing the missing link due to the peculiarity of ALL definitions of the term "species": it lacks time dimension. Although we may delimit co-existing species by their morphological differences (in spite of polymorphism) and/or sequences of their genes, we have no means to delimit them in time.

The answer to your question depends on what do you mean by "common ancestor". If the coalescent one, the answer is obvious, if something else, please specify.

I am not referring to genetics as such. I am referring to the fossil record alone, and the putative common ancestors of various species that are determined on the basis of shared morphology, dating and other factors. Like I say, where is the fossil record of the species ancestral to all great apes, or all mammals, or all amniotes..etc?

There are couple of major problems here:

One is the representability of the fossil record. Imagine a species accounting for billions of organisms during the species existence for example for 10MY and producing fossil record. Often this species would be represented by few specimens scattered in time. I believe that for an individual the probability to be preserved in sediments AND to happen to be examined later by a paleontologist would be 1-E10 or less. If a speciation takes but a small faction of time (according to all modern models of speciation), this would mean that capturing the transition is next to impossible.

Still, if a species was so numerous and well represented, that there is still significant hope to detect the transition, we still would be in trouble claiming success in observing the missing link due to the peculiarity of ALL definitions of the term "species": it lacks time dimension. Although we may delimit co-existing species by their morphological differences (in spite of polymorphism) and/or sequences of their genes, we have no means to delimit them in time.

That is basically an argument akin to"the dog ate my homework". Sure, very few individuals from every species that ever existed has a chance of being fossilized, but we should expect at least one individual specimen to show up. If the common ancestor of humans and chimpanzees lived 5-7 mya (according to phylogenetic inference), then some fossilized remains should be found. If not, then we should be able to find remains of the common ancestors of other species, but hardly any have so far been recovered and identified. The only one I can think of is Homo hiedelbergensis which is the species ancestral both to modern humans and neanderthals.

First of all, a recognition of a difficulty in science is an instrument rather than an argument.

Indeed, the dog-spoiled homework situation may describe the first point of my previous message, especially if a dog would do this systematically (that is we agree that speciation is but a short instane on the evolutionary timescale)

But why the second reason is wrong?

Part of the reason that 'common ancestors' are not identified in the fossil record is, as Dmitry said, because the fossil record simply does not have a very good chance of recording the common ancestor. However, even if the common ancestor is preserved in the fossil record it is nearly impossible to tell if it is the actual common ancestor or simply a close relative.

This is a basic assumption of strict cladistics that all taxa being analyzed have sister-species relationships and that none are directly ancestral to the others. For fossil species where the record is exceptional, such as graptolites, we can in fact see species evolving into another form and there are methods currently being worked out to build trees which allow the possibility of direct ancestors being present (paleotree package in R).

Yet the fossil record seems to be quite good at recording species other than the putative common ancestors. But I would agree that differentiating the common ancestor from a close relative is not an easy task, but surely dating methods would be critically important in determining this. And here I think Dmitry's phylogenetic methods would come in handy: if we predict that an ancestor lived 10 MYA and then date remains to that time period, that should clear up any questions related to the morphology of the specimen itself.

Even so, perhaps the reason why common ancestor species are not abundantly represrnted in the fossil record is not to do with the problem of fossilization per se but because some species are in fact directly descended from each other (i.e anagenesis). Certainly in human evolution, the idea that there is a direct lineage of human ancestor species represented in the fossil record was prevalent until relatively recently. A century ago, for example, it was believed that chimps were the direct ancestors of humans: Some chimpanzees evolved into humans while others basically stayed the same. This is now what is believed in the case of "Homo habilis" which has been shown to be contemporaneous with "Homo ergaster" for half a million years despite being touted as the direct ancestor of the latter species.

The fossil record is just very patchy. In environments like the continental slope the record of marine invertebrates is probably quite complete for short windows of time. But in other areas, particularly on land, the record is very sparse and we are likely missing an enormous amount of data.

The paleotree package I mentioned incorporates stratigraphic position as a major component of estimating whether a species is ancestral but that requires a very well-sampled system and we simply do not have it for most groups of organisms. Even in cases where we have a good record the time resolution is in tens to hundreds of thousands of years in error bars leaving plenty of time for unobserved speciation and extinction in quickly evolving lineages.

Anagenesis is exactly what I was referring to in the last post about cladistics assuming no direct ancestors are present in a data set. At a fine enough level all evolution is going to be gradual and somewhat anagenetic but the fossil record does not record time at that resolution normally.

For human evolution the events have been recent enough that the rock record has not compacted as much yet and we can get finer resolution. Therefore we can observe the gradual transition of one species to another. If we were able to wait several million years and then try to investigate the evolution of humans at some point in the future the transition from H. erectus to H. ergaster could very easily be missed due to either erosion of sediment containing the fossils or a lack of time resolution.

Lauri,

1) This question is sound and the discussion is sane so far. The question is important indeed and concerns "grey area" where our terms do conflict with each other (cladistic dao vs the coalescent one etc)

2) I would refrain from being politically or ideologically motivated while discussing a scientific topic.

Dear Lauri, send my regards to your colleague, Dr. Matti Leisola, who has made a joke of Finnish scientists with his apologetics for intelligent design: http://creation.com/matti-leisola-interview

If you don't have anything to contribute, then please leave.

Why Leisola? He thinks that non-avian dinosaurs lived with humans.

Did the avian ones live with angels? Or they were the angels themselves? Actually i followed the ref. Sad story.

Well, I am glad you realize just how absurd Leisola is. My question relates to the absence or the non-identification of common ancestor species in the fossil record, not whether common descent actually happened. I think the two contributors here. James and Dmitry, seem to have a plausible argument which is that the common ancestor species may not have been around for very long, or that they existed in very small populations, or that not enough time has elapsed in the case of recently extinct species or, quite possibly, that they lived in an environment that was not conducive to fossilization - like forests. Even so, the issue of common ancestor species (especially in human evolution) is a nagging problem for palaeontology.

You did not mention the following: when speaking about a common ancestor, people silently assume that all states of all morphological traits of that organism were in between its descendants. While this is likely to be true for neutral molecular markers, selectively valuable characters would rarely follow this rule due to arms race, character displacement etc. Therefore I may suppose, that even if a paleontologist finds a fossilized "missing link", he would hardly be able to prove it if the detailed model of every trait evolutionary transitions is missing or highly speculative.

I disagree, Dmitry. The common ancestor of humans and chimps was presumed to be more like a chimp than a human until very recently when it was realized that the MRCA of both species was more monkey-like and arboreal in nature. Molecular evolution studies, which is your focus, have shown that chimps have been evolving perhaps more strongly than humans which also detracts from the notion that the MRCA of humans and chimps was just a primitive panin. Part of the problem of fossils is not their paucity, but the fact that the specimens are incomplete, deformed, or of poor quality. Just finding a tooth or a broken jawbone is insufficient to draw exact conclusions. But I still expect there to be something rather than nothing.

I did not mean humans, but had in mind my recent discussion with a paleontologist studying mostly plants. For him the ancestor was always in between.

OK. But that is where molecular dating methods may prove useful. Even if the character traits are not intermediate between the descendant species, the date ascribed to the putative common ancestor fossil should be in congruence with what is predicted by phylogenetic inference.

With molecular dates you have several potential sources of error. They require a model of evolutionary rates to calculate the timing of branching. I honestly don't understand exactly how the models are chosen but I do know that they are often calibrated using dated fossils which have their own errors mentioned already.

The models are going to be dependent on taxa that are sampled so you really need the majority of living species to be included to have high confidence in a model. However, if there is a branch or branches of a clade that are entirely extinct for which we have no molecular data but who are included in the group you are trying to estimate the timing of the common ancestor for I think that could introduce a lot of noise. These errors together should mean that any molecular clock estimate will have a variance greater than the precision needed to pin down a single species as ancestral.

Even if you are confident in a molecular clocks date and you find a taxa that seems to be a good fit for a common ancestor given the lack of resolution in the rock record you usually have no way of determining if that species was actually ancestral or a closely-related offshoot to the crown group. You can only hypothesize that the species is close to the most recent common ancestor of your group based on the congruence of fossils and molecular dating data.

True. But I am not asking for absolutely certainty, just plausible possibility. It would be sufficient just to say "This is a likely but not definite candidate for a common ancestor of two or more species."

I guess we couldn't decide, even if we had a perfect fossil record. As James argues, it would not be possible to tell which fossil was last common ancestor on principle.

Well, I am saying that we don't need to be certain but that we need to have plausible candidates that can be reasonably dated to the expected/inferred time. The fossil record is quite good - hundreds of milions of fossilized remains have been found - so the common ancestor species should have been excavated from the rocks by now even if they haven' been identified as being ancestral.

Is there a specific system or reason why you want to identify a specific common ancestor?

Of course there is. Evidence for common ancestry should not be limited to inferring the existence of putative ancestors from DNA sequences. There should be physical evidence of them that can be obserevd in the rocks. Also, it is useful as far as any debate over the merits of anagenesis versus cladogenesis. I gave the example of Homo habilis, long assumed to have evolved as a population into Homo erectus, now being seen as merely sharing a recent common ancestor with erectus. Alternatively, just a few habilines could have evolved into erectus and the rest of the population remained the same. Perhaps the absence of evidence for a common ancestor species can be explained in terms of there being a direct descent.

According to "A Framework for Post-Phylogenetic Systematics" by R. Zander, available on Amazon, "shared ancestors" may well be one of the sister groups. Peripatric evolution should make this a common occurrence.

Joseph, I think that the problem is not so much a lack of fossils that *could* be common ancestors (i.e. share/combine traits with younger sister species) but rather a lack of unbroken lineages of such fossils through time such that common ancestry can be unambiguously assigned. Post-Darwin, there was a bit of an obsession with finding "missing links" that were literally half-way between modern taxa (as with the Creationist fallacy of the "crocoduck") but we are much more sophisticated with our understanding of evolution now. Evolution - particularly *morphological* evolution - is neither constant nor unidirectional through time, and it is therefore impossible to extrapolate back a predict exactly what a common ancestor would look like. Without a full vertical record, the best that can usually done with confidence is to identify something as consistent with being a common ancestor but not definitely THE common ancestor. For the reasons already given, such unbroken chains of fossils through time are extremely unlikely. Thus, even if the odds are in favour of *any* common ancestor being preserved, the chances of *enough* common ancestors for robust assignment of ancestry are vanishingly small.

Richard, I understand the points you make - which are perfectly valid - but the issue of fossil preservation of identifiable common ancestors still boils down to the excuse of "the dog ate my homework". In human evolution, we have lots of hominid fossils purported to be extinct direct ancestors of our species, i.e transitional intermediates, but not one that is claimed to be a common ancestor of us and extant chimpanzees.

@Joseph, you are entitled to your opinion, of course. If, however, my points *are* "perfectly valid" then your "dog ate my homework" analogy is false unless you live in a kennel and your homework was an open bowl of pedigree chum. (In which case it is disingenuous.) You are looking for an explanation for something that has already been explained - If you want to see this as unlikely and in need of further explanation, I think you will be disappointed.

Richard, it is a matter of testability and falsifiability. Yes, I understand that fossilization is a precarious process, and that trying to identify potential common ancestors is very difficult, but this also detracts from the predictive power of palaeontology and evolutionary biology where "the dog ate my homework" is the excuse. One alternative is to propose that the common ancestors were not very successful and that is why they didn't hang around long: "adaptive radiation" is just such a scenario.

@Joseph, an "excuse" and an "explanation" are not the same thing. It is true that in circumstances for which we have a "complete" fossil record and expected fossils are missing, alternative explanations may be required. It is also true that when we *don't* have a complete record, we cannot rule out alternative explanations as being the actual natural history. Perhaps I have misunderstood but I still fail to see (a) where your surprise (if any) lies and (b) why you consider very real evolutionary, geological and biogeographical processes (as we currently understand them) constitute an "excuse" as if there is something embarrassing to brush under the carpet or cover up.

Do you have a specific example where the fossils are expected - i.e. (1) a specific geographical location with the right age rocks of the right type and strong reasons to believe that a specific common ancestor should be present, and (2) closely-related ancestral and descendant fossils such that strong predictions can be made regarding the form of the common ancestor - and yet such fossils are missing despite extensive searching?

IMO, that which you says "detracts from the predictive power of palaeontology and evolutionary biology" simply constitutes being realistic about said predictive power. One of the reasons that fossils of potential common ancestors are so interesting is because it IS so hard to predict exactly what they should be like - thus they have the potential to tell us something new about the natural history of the organisms concerned!

Richard. on the question of expectation, it was predicted that the remains of the ancestors of whales would be found around the mouth of the Indus in Pakistan where terrestrial mammals took to the water and gradually became more aquatic. Fossils of extinct species like Ambulocetus and Pakicetus were indeed found, and these are regarded as plausible ancestors of all cetaceans. They may well not be, and there is definitely cause to doubt, but at least they are possible candidates. However, far from being the rule in paleontology, as it should be, this is one of the few exceptions. In human/primate evolution, for example, attempts to find common ancestors has become something of a sad joke - the "Ida" fiasco being a salient example of this (Ida was initially claimed to be an ancestor of all apes). Lots of predictions have been made about the location, rock strata, specimen type and time period where primate ancestors should be found. but they have failed to live up to expectation. The typical excuse is that primates lived in forests where the soil was acidic and dissolved the remains quickly which prevented the chance of fossilization.

I previously referred to the reconstruction of the appearance of the putative common ancestor of all placental mammals (based on an analysis of character traits from eutherian species) that has recently been completed: http://www.livescience.com/26929-mama-first-ancestor-placental-mammals.html It is predicted to have lived 60-65 MYA. Great! But, unless the population of this ancestor was very small, and its time on earth very short, a fossil which broadly matches this reconstruction should have been found or will soon be found. If this does not happen, then the usual excuses will be made about the "imperfection of the fossil record" that Darwin made 150 years ago (when the geological record was actually quite extensive).

I think that the fact that "lots of predictions have been made..." indicates that no one really knows for sure exactly what to expect. Specific hypotheses/assumptions will clearly be weakened by the absence of expected fossils but I sincerely doubt that anyone can make robust, high confidence, specific predictions about even the common ancestor of humans/chimps, let alone all mammals. The world is a big place and geological time is long. However disappointing this reality is - recognised indeed since Darwin's day - it is not an "excuse". There is an extensive fossil record, true, but it's surely but a *tiny* fraction of past biodiversity in both time and space.

Richard, you are right that nobody is sure where to look: the first chimp fossils were found in East Africa, right alongside Homo erectus: http://www.nature.com/news/2005/050829/full/news050829-10.html This refuted an assumption that chimps and humans split apart because one remained in the forests while the other took to the savannah. That shows the problem with evolutionary biology in general: there are too many preconceptions and assumptions, and there is not enough induction. The data is often used to fit in with the existing model rather than determining how the model should take shape. Frankly, I doubt that the "reconstructed" ancestor of all mammals will ever show up in the fossil record no matter how hard we look. I don't think that can be properly explained in terms of the odds of successful fossilizaton.

Here is one;

http://www.nature.com/nature/journal/v502/n7470/full/nature12617.html?WT.ec_id=NATURE-20131010

Mollusks often provide extraordinarily complete records. In revising the classification of Bivalvia (Carter et al., 2011) we proposed to stop playing the silly game of not wanting to see what we have right in front of us. We see ancestral species. But cladistics says you can't see that. So, we proposed to use cladistic analysis (or course), but refrain from ignoring stratigraphic and ancestor-descendant evidence. There are common ancestors, the problem is finding them.