Resolving the avian tree of life: Part II

In my previous post I talked about a landmark paper in avian systematics, that of Hackett et al. (2008), who reconstructed the first phylogeny based on DNA sequence data that included all major bird clades, and all the major Passerine families (aka 'perching birds': things like sparrows, flycatchers, warblers, etc., a clade that includes half of the world's bird species). This phylogeny revealed some broad trends in avian evolution, but only included 169 bird species, chosen to represent the full range of avian variation. There are actually about 10 000 known species of bird. This may seem like a lot, but compared to plants and insects, it's virtually nothing, so reconstructing a full species-level molecular phylogeny for birds is theoretically quite doable. This is the task that Walter Jetz and colleagues took up, and achieved, publishing their findings last year in the journal Nature. Here it is:

The full phylogeny reconstructed by Jetz et al. (2012), with frequency distribution of diversification rate (in centre), and diversification rates plotted per species

If there is a more recent, more complete phylogeny of birds, then you may rightfully ask, why did I even bother mentioning the 2008 study by Hackett's team? The answer is twofold. First, the two papers had different aims. The Jetz study, combining their phylogeny with distribution data for the world's birds, aimed to map the global distribution of bird diversification through time (more details to follow). Second, and somewhat connected, is the fact that the Jetz phylogeny is actually not a truly independent creation. In fact, the authors only had sequence data for 6 663 species, roughly two-thirds of the total. To place the remaining species, they relied purely on previous taxonomic classifications. Also, they did not use the molecular data to reconstruct the full phylogeny. Instead, they relied on an older, less detailed, broad phylogeny of birds. Yip, you guessed it, it's back to Hackett. Jetz et al. used the Hackett phylogeny to assign species to particular clades, and then simply added further detail to that 'backbone'. There are various reasons to be cautious about the results stemming from this approach, as Robert E. Ricklefs and Mark Pagel point out in their respective commentaries on this article in the same issue of Nature. But first, let's have a look at what those results were.

The main aim of the study was to explore the evolutionary history of birds as a whole. As the title of the paper ('The global diversity of birds in space and time') suggests, this involved both a temporal and a spatial aspect. Firstly, they plotted species richness, and diversification rate, against time, and showed that both have increased, particularly in the last 50 million years. This has been driven by certain groups, particularly the passerines, which originated about 60 Ma, as well as gulls, woodpeckers, and waterfowl (ducks and geese). Jetz et al. calculated a diversification rate (DR) per species, which initially seems nonsensical. however, their technique was based on the branch lengths of each species. Although the full story is far more technical, the upshot is that more recently diverged species (i.e. those with shorter branch lengths) will have higher DR values. Thus their next step was to examine the distribution of species with high DR.

Spatial diversity: a, b and c show diversification rate, whereas d and e show relative and absolute richness of the top 25% diversification rate species, respectively. f shows species richness for all birds.

The map above shows their results. One of the key findings is that diversification rate, and areas with a high proportion of rapidly diversifying lineages, do not correlate well with latitude across the globe. Instead, temperate areas, rather than highly diverse tropical areas, seem to be hotbeds of radiation, but only in some areas (e.g. compare southern South America with southern Africa in maps a and d). This suggests that tropical areas are stores rather than major generators of recent diversity, and that birds are continuing to speciate in more recently emerged temperate areas. However, many feel that certain aspects of this study provide good reason to regard it as preliminary, and its results tentative.

One of the strongest criticisms is levied by Pagel, against the use of branch length data to infer DR. There are serious reasons to be skeptical of lineage age estimates, and this approach makes the results highly susceptible to changes in this regard. A further criticism relates to the way in which the phylogeny was constructed. Because the method relied so heavily on a prior phylogeny, we have no way of assessing how confident we can be in this one. As a result, any inferences drawn are based on faith in the older phylogeny, and the molecular data currently available are not put to full use.

While both Ricklefs and Pagel present a strong cautionary viewpoint, both also admit that this is at least a very important first step towards providing answers to the most fundamental questions about avian evolution. Answers that may in turn reveal trends applicable to larger and more inclusive sections of the tree of life.