A Master’s Class: What I Found

This is it. What you’ve all been waiting for. The stage is set, the major players all identified. If you’ve been keeping up, you know all about what genes are and how they change. You’ve gotten to know a bit about microsatellites and mitochondrial DNA, and what each can tell us when you know how to look at them. You’ve even gotten a quick introduction to population genetics and how important genetic diversity is to an organism’s survival. And so we come to the final question: what is it that I found out there among the fog coated cairns of the Scottish Highlands?

Truth be told, not a lot. But, paradoxically, that’s also finding quite a bit.

As I’ve said, the two main questions I was trying to address with my Master’s were:

How are Scottish populations of Rana temporaria, the common frog, related to other European populations?


What kind of population structure do we see across the Scottish Highlands?

We’ll start with the first question. Looking at cytochrome b from the populations I sampled, along with sequences from previous studies pulled from GenBank (Muir et al. 2013, Palo et al. 2004, and Teacher et al. 2009, to be precise), we identified 15 haplotypes across 75 sequences, sampled from 14 countries. In my Scottish samples, there was only ever one cytb haplotype identified (although Muir et al. 2013 did find a few single nucleotide mutations present at some sites, but these were never found in more than one individual). This dominant haplotype was the same as the haplotype found in populations sampled in Switzerland, Spain, Ireland, Germany, France, and other parts of the UK. The haplotype network below shows the distribution of haplotypes among the countries we found sequences for.  Note that while there are a number of other haplotypes found in Western Europe (represented by the smaller circles surrounding the large central node of the left hand side of the network), they are all separated by a single hash mark, meaning they differ from the dominant haplotype by a single base.

rana cytb.trimmed.TXT
Here we have the cytochrome b haplotype network. Node size indicates the number of samples that share the haplotype represented by that node. So bigger node, more samples. Numbers next to country indicate the number of samples collected from that country. Bolded Scottish sample comes from my study, which I did to differentiate it from the previous study’s sequences. All due credit to the program PopART which I used to generate this network. If you need a quick, flexible, and fairly attractive network construction tool, check it out.

When we look at cytochrome oxidase subunit I (COI), we see a very similar outcome (additional COI sequences were pulled from Vences et al. 2013). Across the 27 sequences we looked at, we identified 12 haplotypes, covering 10 countries. Unlike the cytb network, we can see that there is a ring of unlabeled nodes bridging the gap between Western and Eastern European samples. This ring is made up of “inferred ancestral haplotypes.” Basically, we argue that a network with common ancestors we didn’t sample makes more sense than assuming no common ancestors, because it makes the network simpler and keeps the total number of hash marks (i.e. mutations) lower. The algorithm used to create these networks decides whether or not these extra ancestors are necessary. It’s worth keeping that process in mind because, while we see mostly one or two nucleotide differences between any neighboring nodes, if you want to get from Scottish samples to Spanish samples or French samples, you have to go through a number of links and thus accrue a good number of mutations.

Vences COI plus 3 haps.2.no.geo.TXT.labelled.again
Haplotype network for cytochrome oxidase subunit I. Once again, bolded samples are from my study. Hash marks indicate number of mutations. 

Looking at my Scottish samples (here noted with bold labels), we see that we have three haplotypes, so ever-so-slightly more variable than cytb. One of these was only present in one frog from a single site. However, the other was present across all three sites in a total of 10 frogs (I collapsed those 10 down into one sequence during alignment so that’s why the label indicates only one sample). So we do see some minor consistency there. It is also definitely worth noting all those Spanish samples over on the right hand side. I have no idea why those are so different, but man, Spanish frogs seem to be getting pretty weird so far COI is concerned.

So what does all this mean? Well, looking at both cytb and COI, it seems that Scottish populations of Rana temporaria are very closely related to those in mainland Europe. However, since cytb is so steady across most of Western Europe it is difficult to say where Scottish populations might have originated before they migrated here at the end of the last glacial period. COI is slightly more variable, but still shows close relationships between Scottish populations and those found in Ireland, Germany, and Switzerland. This could indicate a colonization originating from Germany and Switzerland, or it could be a result of incomplete sampling. To draw a solid conclusion, more COI sequences are really needed (I say this mainly because the study that I pulled those sequences had much fewer sequences for COI than the segment of cytb they looked at).

The results we see looking at population structure are a bit more informative. I’ll make a quick note here that there was a second question sort of nested within the work I did looking at population structure: what are the relative strengths and weaknesses of three different Bayesian clustering packages often used for evaluating structure from genetic data? I’m not going to get into that here though, largely because that’s very technical and will require a lot of explanation to understand how Bayesian stats work, the mechanisms of Markov Chain Monte Carlo algorithms, and the nuances of model selection. I’ll probably write about all these things later on in another post, but for now, I’m going to skip that discussion and give you the highlights.

Basically, there is little to no structure in R. temporaria populations looking across the Scottish Highlands. Frogs from Glasgow are not genetically differentiated from frogs farther north in the Blackmount region or even frogs found on top of Cairn Gorm, which is considerably farther east (only one mountain was differentiated, the same one identified in the previous study, Meall nan Tarmachan). What this means is that there are high levels of gene flow between populations. Enough individuals are moving between these far-flung populations that any random shifts in allele frequencies resulting from mutation are carried into the other populations, smoothing out any possible differences that may arise. This is interesting because, as I said in my first post about all this beautiful beautiful science, previous work has found that frogs don’t tend to go far from where they breed (us science folks like to call that high site fidelity). If frogs are moving between populations and breeding with enough frequency to homogenize populations on this scale, then they aren’t sticking around once they’ve made all those tadpoles (which we would call, unsurprisingly, low site fidelity).

This is a sort of population map generated as part of a spatial model with the program BAPS. Each shape indicates the boundaries of the area where samples were collected, based on a Poisson-Voronoi tessellation (which I’ll talk about in a later post). Color indicates which population each site belongs to. Only sites 8 and 9 are differentiated and so only they have a color other than green. Sites 1-9 are from the previous study, and are mostly paired high and low altitude populations from a series of mountains. Sites 10, 11, and 12 are from my study, and represent the Blackmount Region (specifically the mountain Meall an Araich), Cairn Gorm, and Glasgow respectively. 

Given that Scotland is really made primarily of frog habitat (i.e. it’s really moist all the time everywhere), it may be that frogs are able to move around most anywhere without any real barriers. Even in areas that are highly developed, like the post-industrial Glasgow, the largest city in Scotland, I was able to find R. temporaria tadpoles on the university campus, and most every researcher I spoke to about the project assured me that if I needed some samples, their garden pond was rife with the little guys. In other areas where land use and development could interrupt watershed conditions to the point that there isn’t suitably wet land between populations, frogs would get cut off. Now these conclusions are decidedly post-hoc, meaning I haven’t tested any of this and you can’t even really call them conclusions at all. They’re much more like musings, or future hypotheses to be tested. But it would certainly seem that something in Scotland is helping facilitate gene flow that wasn’t seen in similar studies in Scandanavia. To definitively identify what that something is though, would require a good deal more work comparing these two regions and how frogs behave in each.




Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s