
edit: I dug up an old thread I started bout the undulatus split and the boundary line isn't clear where I found this--could be virgiatus (sp) I think, or consobrinus. How would I tell?
Moderator: Scott Waters
Rather than "avoiding the debate", I think you're assuming a very particular position in that debate. If we (rather naïvely) take Leaché's results at face value, around 20% of the individuals in Sceloporus undulatus sensu lato are introgressed between the various segregate species. This suggests a very high rate of gene flow for us to consider these separate species.ritt wrote:paalexan: The different results between the nuclear and mitochondrial datasets simply shows something that has been shown time and again: gene flow among distinct species (which could be ongoing or could be historic). Due to hybridization between some of the species, mitochondrial haplotypes have introgressed into the other species, resulting in the pattern of a few samples from near the boundaries that appear to be in the 'wrong' clade on the nuclear tree: these are samples that have introgressed mitochondrial DNA. This could have occurred thousands of years ago and may or may not be ongoing. Avoiding the debate on a strict biological species concept or a more relaxed BSC that allows some level of gene flow, this does not mean that these are not distinct species. Leache actually has some pretty good evidence that they are different species.
Well, I disagree. I consider mtDNA-only studies a non-starter when it comes to delimiting species (the data aren't relevant to the question). Once it's been further established, as it is in the 2009 paper, that mitochondrial and nuclear results show substantial incongruence, personally I consider the mtDNA-delimited species defunct.However, it is clear that the mitochondrial clades are not perfect representations of the species boundaries, which is something that pops up all over the place. As such, its necessary to conduct more thorough sampling, combining nuclear data with morphological data to figure out exactly where the boundaries between the species are. Regardless, its definitely not the case that the species boundaries from the mtDNA don't hold up, actually with the exception of those few samples from near the boundaries they hold up quite well (particularly considering how 'slow' nuclear DNA evolution is relative to vertebrate mtDNA evolution).
In what way are they not relevant? Any single locus study is far from ideal, be it based on a mtDNA or nucDNA, but I don't see how mt data are irrelevant to the question of species delimitation. There are clades of highly divergent mitochondrial haplotypes in these guys (and loads of other groups). How did these divergent clades arise? Most likely they, at some point in the past, occurred in isolated pockets without gene flow among them, that then spread to the current distribution. While these were likely associated with bottlenecks that accelerated the loss of other mt haplotypes (thus resulting in the single clade for each species), but the point is they were almost certainly isolated at some point. I have yet to hear an argument that isolated populations will not speciate given sufficient time in isolation. I'm not saying that studies should be based solely on mtDNA data, in fact I don't feel taxonomic revisions should be suggested based solely on mtDNA, but multiple nuclear loci and / or morphology (preferably both) need to be considered for taxonomic changes. But, what I am saying is mtDNA is relevant to species delimitation. Not sufficient for elevation of species in my opinion, but certainly relevant.I consider mtDNA-only studies a non-starter when it comes to delimiting species (the data aren't relevant to the question).
As I stated above, the only reason such high incongruence was observed is Leache was specifically targeting samples from near the clade boundaries. Drop out those 3 samples from near the boundaries, then all the sudden the only congruences are at the deeper levels. The clades within the undulatus complex are the same except for those 3 samples from near the mt clade boundaries.mitochondrial and nuclear results show substantial incongruence
I'm not sure if you have access to the supplemental, but there is a supplemental figure with all the individual gene genealogies. Incomplete lineage sorting is all over the place. This is to be expected with nuclear loci at this level, particularly for critters like Sceloporus that have massive effective population sizes. It takes around 4 +/- 2 N generations for any given locus to sort to reciprocal monophyly. That N generations is the standard coalescent branch length, basically to convert that to years you take the effective population size (Ne) and multiply it by the generation time. Now Ne is very tough to accurately estimate, but say you've got an effective population size of 100,000 individuals. If the generation time was 2 years, it would take an average of 800,000 years for a locus to sort to monophyly. Given that there's almost certainly way more than 100,000 Scelops out there that are reproductively active and part of the Ne, you can imagine it takes a super long time. The strong support for its particular placement in the concatenated tree is a bit odd, but by concatenating them you're assuming they all share the same signal (i.e. that the topology is the same for each locus), which is clearly not the case here when you look at the gene trees. So the concatenated tree isn't really useful, I've been looking at the BUCKy trees in Fig. 5, which basically looks for concordance among the gene trees, and doesn't assume that they've got the same topology. The different placement in the concatenated tree might be due to strong signal at some loci showing that relationship (which drove the pattern), and weaker signal at other loci, or something along those lines. But getting back to my original point, if you look at the gene trees in the supplemental, there is a ton of incomplete lineage sorting.regarding Sceloporus woodi...
Well, not really. What he says in the paper is that samples were chosen to maximize geographic coverage, and that "some" specimens were sampled from mtDNA clade boundaries. Just take a look at the map; samples are pretty evenly distributed, not conspicuously concentrated at mtDNA clade boundaries.ritt wrote:First off, Paul, sorry for hijacking your thread a bit here.
By avoiding the debate I meant to try to avoid a debate on whether gene flow needs to be completely ceased for things to be considered different species. Regardless, trying to say 20% of the samples have introgressed mtDNA is, as you put it, naive, though I'd argue "naive" is an understatement. Rather, such a value is misleading and ignorant of the sampling strategy and the purpose of the study. Leache was targeting samples near the mt clade boundaries, looking for incongruences between mt and nuc clades. So of course he found a relatively high rate of it, he was specifically looking for it.
There will be some set of hierarchical mtDNA relationships among any set of individuals. So, you sample a bunch of individuals from Sceloporus undulatus, yes, there will be mtDNA clades (which you can accurately recover to a greater or lesser degree depending on various aspects of the data and analyses). This isn't a novel empirical observation that tells you anything about the number of species involved. There will be some set of mtDNA clades whether it's one species, two species, etc.In what way are they not relevant? Any single locus study is far from ideal, be it based on a mtDNA or nucDNA, but I don't see how mt data are irrelevant to the question of species delimitation. There are clades of highly divergent mitochondrial haplotypes in these guys (and loads of other groups). How did these divergent clades arise? Most likely they, at some point in the past, occurred in isolated pockets without gene flow among them, that then spread to the current distribution. While these were likely associated with bottlenecks that accelerated the loss of other mt haplotypes (thus resulting in the single clade for each species), but the point is they were almost certainly isolated at some point.I consider mtDNA-only studies a non-starter when it comes to delimiting species (the data aren't relevant to the question).
On this point, I think I agree. I've been arguing that mtDNA alone (and this applies to other single loci as well, but in the herp world single locus analyses are all mitochondrial) is not relevant. It could be part of a more comprehensive set of data that would be informative, absolutely, but on its own, I don't think the observation of a particular set of mtDNA clades is relevant to inferring a number of species.But, what I am saying is mtDNA is relevant to species delimitation. Not sufficient for elevation of species in my opinion, but certainly relevant.
Ah, OK, I haven't looked at the supplemental figures.I'm not sure if you have access to the supplemental, but there is a supplemental figure with all the individual gene genealogies. Incomplete lineage sorting is all over the place.regarding Sceloporus woodi...
I looked at the BUCKy trees a bit, too, but there are some problems with interpreting them relative to the entire 29-locus dataset that places Sceloporus woodi within Sceloporus undulatus; the BUCKy analyses are limited to the 8 loci with complete coverage, so I wonder to what extent the different topologies from concatenated ML/Bayesian analyses and BUCKy analyses are due to the different analytical methods or the different data matrices used. I guess staring at all the single locus analyses might clarify that, but I don't think I've got time at the moment for the rather tedious tree comparisons that would be involved...So the concatenated tree isn't really useful, I've been looking at the BUCKy trees in Fig. 5, which basically looks for concordance among the gene trees, and doesn't assume that they've got the same topology. The different placement in the concatenated tree might be due to strong signal at some loci showing that relationship (which drove the pattern), and weaker signal at other loci, or something along those lines. But getting back to my original point, if you look at the gene trees in the supplemental, there is a ton of incomplete lineage sorting.