Haplotypes and Understanding Haplotype Networks
Okay so lamarckwaswrong asked me to help out with an issue regarding the interpretation of haplotype networks. I thought this might be a bit of a general interest query, so I decided to make a more complete post about what they are, how they work, and how to interpret them.
So first off, what is a haplotype? Well basically, a haplotype is a unique sequence of bases over a region of the genome. Individuals that share a haplotype have the exact same sequence for this entire region, and are therefore inferred to be closely related to one another. Usually a haplotype sequence is relatively short, because the longer the sequence length used for comparison between populations, the lower the chance that individuals within and between populations will be identical across the full extent of that sequence - you lose any informative value. On the other hand, if the sequence is too short, there is no variation at all except over huge time periods, and again it is uninformative. A happy medium must be found.
You can use haplotypes of a sensible size to compare individuals between populations in order to infer which populations are most closely related to which others, and thereby also to infer the recent evolutionary history of diverse and dispersed populations. It is much more informative in this respect than microsatellites, because it can use an actually informative region of genome, rather than tandem repeats, to infer relatedness, and is therefore less likely to be due to repeated convergence on the same unique sequence.
Haplotypes are now one of the foremost tools in phylogeography, the recently developed field of mapping evolutionary and geographical history against one another. They can be displayed as either a heirarchical tree (a phylogeny) or as what is called a 'haplotype network'. The latter is most informative when the ancestral and derived haplotypes exist simultaneously (Gehring et al. 2012). It consists of a series of coloured balls (pie diagrams) and sticks:
In figure 1, from Ratsoavina et al. (2013, full text available here), each ball represents a haplotype. The size of the ball is proportional to the number of individuals sampled belonging to that haplotype. The colour represents, in this case, a locality. Lines connect each haplotype to its most similar relative. A grey circle is an inferred median. Bars represent mutational steps between haplotypes.
But this is by no means the only way to display a haplotype network. Take, for example, this network of Heterixalus betsileo frogs, published by Gehring et al. (2012, full text available here):
Figure 2 is exceedingly complicated. Here, the haplotype has been inferred from a nuclear gene (Rag-1), and in this figure it is being mapped against mitochondrial lineage. So as before, each ball represents a haplotype, the sequential number of which is indicated either within or beside it, with size proportional to the number of individuals belonging to that haplotype, connected to the most similar haplotype by a line. The colour represents the mitochondrial lineage to which individuals represent, and the slices of pie indicate sampling localities. Here, black circles represent intermediate missing or unsampled steps. Dashed lines represent equally probable alternative connections.
This second figure, as it is shown here, is geographically uninformative, but instead is highly informative about the relationship between nuclear and mitochondrial lineages. Fortunately, earlier in this same paper, Gehring et al. (2012) showed a haplotype network of the mitochondrial lineages of H. betsileo:
This is far more informative, and has been contextualised also with a map of eastern Madagascar and the phylogeny of the lineages. By doing so, this figure allows us to see that there is a some extent of geographic correlation between haplotypes and lineages within these frogs, and thereby to infer the evolutionary history of this species.
In summary, there are a few commonalities between all haplotype networks: (1) each circle represents a unique haplotype; (2) the diameter of the circle is usually proportional to the number of individuals of that haplotype sampled; (3) haplotypes are connected to one another based on their similarity; and (4) inferred missing or unsampled steps are represented, either by perpendicular lines or black balls. Choice of what colouration indicates varies between purposes.
Haplotype networks are sometimes hard to interpret, but they are extremely useful tools. Phylogeography has the potential to address a great many questions about population dynamics, colonisation, and perhaps most excitingly, incipient speciation.
Gehring, P. S., M. Pabijan, J. E. Randrianirina, F. Glaw, and M. Vences. 2012. The influence of riverine barriers on phylogeographic patterns of Malagasy reed frogs (Heterixalus). Molecular phylogenetics and evolution 64:618-632.
Ratsoavina, F. M., M. Vences, and E. E. Louis. 2012. Phylogeny and phylogeography of the Malagasy leaf-tailed geckos in the Uroplatus ebenaui group. African Journal of Herpetology 61:143-158.
