Ideally I would have introduced the concept of ghost lineages here before but I never quite got round to it and as something is coming up soon-ish that requires some discussion of the situation it seemed a good time to get stuck in. This is one of those areas of palaeontology that seems to be more controversial that it should be, and though I am not sure why, I suspect it is down to a misunderstanding of the fossil record in general or phylogenetics specifically.
Those who are not up on their ability to read phylogenetic trees might want to take the opportunity to revisit this post of mine on the subject and with special reference to illustration number 8. The crux of the ghost lineage problem is where we have disparity between where we find certain organisms in the fossil record, and where we think we should find them. Everyone agrees that plants are older organisms than animals and thus we should find plants earlier in the fossil record than animals and we do, but we also think that birds evolved from derived maniraptoran theropod dinosaurs, yet we find birds (well, OK, Archaeopteryx) *before* those dromaeosaurs and troodontids that they supposedly evolved from.
This then is a ghost lineage – the gap of time in which an organisms supposedly existed (according to our phylogentic hypothesis) but for which there are no fossils. Looking at the simple figure here I’ve made here (sorry for the poor quality – Powerpoint strikes again) should help clarify this. The thick lines represent fossils, the thin ones the relationships determined by our phylogenetic analysis and the vertical lines are time. As you can see these vary a fair bit but are mostly consistent. D is the most basal one (branches off first) which is what we see – it is also the oldest (over 40 time units old). A and B are sister taxa and despite the discrepancy between them, their position is consistent – they both appear after D. A and B are slightly different in age range but this is no real problem – the fossil record is not perfect and we are not necessarily going to find them both at exactly the same time, so this is fine. C however has a ghost lineage (marked in red) since it only appears after A, B and D (here) yet phylogenetically we expect it to appear after D but before A and B and it does not. The ghost lineage itself is of an uncertain length since it only really needs to go back to the same age as B (since they may have split off almost simultaneously) but could go all the way back to when D branched (since this may have happened simultaneously). Either way though it has a ghost lineage of at least 10 time units and possibly much more.
This could of course change with later finds – if we find A or B back at 30 time units then C’s ghost lineage would extend since it must predate this. Alternatively, if we find C at 35 units, then it would dramatically shrink the ghost lineage (though note eliminate it). Only if we find C between the origin of D and B can we remove this problem.
But just how much of a problem is this? Are these really very important and what do they mean? Well that all (inevitably) depends on the exact situation which is a combination of phylogeny and preservation. Generally however I would say that they are interesting but not necessarily that important –they tell us that there is potentially a problem with either our understanding of the fossil record or our phylogenetic hypotheses and neither of these is especially enlightening – we know this already – and some circumstances are pretty much unavoidable.
Let’s take (a rarity for the Musings) an example from the invertebrate fossil record, or rather what is not there. It is quite clear that things like rotifers and tartigrades have been around for a very long time – hundreds of millions of years. However their fossil record is effectively zero and thus their ghost lineage runs from the present to their origin. That is truly enormous, but no surprise and nor a *problem* for phylogenetic analysis since we would not expect them to be present – they are incredibly small and do not preserve well. There is no real issue with them not being in the fossil record and thus the ghost lineage is effectively irrelevant. As ever, the fossil record is full of all kinds of biases and if all you are doing is showing up one of the more obvious ones then it’s not really an issue. (While we are here lampreys are another good example who have just a few bits known from the Early Cretaceous but are otherwise effectively absent from the fossil record despite the fact that they must predate bony fishes).
Ghost lineages can also tell you that your phylogenetic analysis may be wrong which again is no great surprise or issue. Cladistic trees are our *hypotyheses* of relationships and not statements of relationships (a subtle but important distinction) so the idea that we may have a taxon in the wrong place is not an issue often. Pterosaurs, as I have oft lamented, have a patchy fossil record and are hard to analyse systematically since they have few characters to piece together their relationships (both between pterosaur lineages and comparing pterosaurs to other clades). As such any ghost lineage for pterosaurs can tell us that we have a mistake in our phylogeny or their already patchy fossil record means we simply haven’t found something that probably does exist.
The real problems begin where we have a good fossil record and a well established phylogeny but the two are still at odds. We can’t immediately think that our phylogeny is amiss nor that our well scoured rocks are failed to preserve the well represented (or we have not found it). The truth however is that these situations are incredibly rare and to be honest I cannot think of an obvious candidate to use as an example here though they abound for others with either phylogentic issues or preservational ones. Haldane’s rabbit is the only think I can think of and that itself is a hypothetical example. Since we think that most clades when they originate are represented by only a few, small taxa it is no surprise that clade origins are hard to pin down – their early representative would be few in number, small in size and older that their descendents and thus by definition are rather harder to find and may even be hard to tell that they are related to later taxa if the differences are great enough. In short, the odds are against us finding or identifying early members of clades. While this should hold true for all clades (and thus relieve much of this issue) it means that we only need to find one early representative of a one late clade to throw up a bunch of ghost lineages for all of its relatives, but it need not mean our phylogeny is wrong, merely that we have been luck with one find we do have.
That pretty much covers everything here that I can think of while keeping things short and relatively interesting. Big ghost lineages *are* a worry, but before you think that the mere existence of one is problematic, look at the factors that may affect this. Are there various competing phylogentic hypotheses which might conflict with each other and one of which might fit the fossil record far better, are the rocks well know, have all the older rocks been checked, are the rocks available of the right age or represent the right environment (there’s no point looking for amphibians in marine rocks) and do these organisms even preserve well, or were numerous enough to likely be found? Only when this has all been taken into account can a ghost lineage be considered a real issue and even then, who knows what you may find next? Our understanding of the fossil record continues to grow and new finds are new surprises are inevitable so even the most problematic gaps will likely be filled in eventually.
Share this Post