August 20, 2015 | David F. Coppedge

Darwin's Tree Am-Bushed

Darwin’s Tree of Life looks more like a bush, evolutionists find in bird data; that inference is probably widespread.

Jonathan Wells (Icons of Evolution, 2000) axed it 15 years ago, but it’s taking a long time to hit ground. Eight years ago, Eugene Koonin called it a “bush of life” (10/08/07). Was nobody paying attention? Why is it news now when Uppsala University says the Tree of Life is a Bush of Life?

New species evolve whenever a lineage splits off into several. Because of this, the kinship between species is often described in terms of a ‘tree of life’, where every branch constitutes a species. Now, researchers at Uppsala University have found that evolution is more complex than this model would have it, and that the tree is actually more akin to a bush.

The “bush of life” metaphor implies that common ancestry is a myth, or diversification is rapid, or both. Using data from the diversification of birds, the Uppsala researchers failed to find a tree: i.e., a branching, bifurcating pattern of ancestors and descendants.

‘We can see that the very rapid rate at which various bird species started evolving once the dinosaurs went extinct, i.e. around 65 million years ago, meant that the genome failed to split into separate lineages during the process of speciation’, Hans Ellegren says.

Now, it appears that to maintain the Darwinian picture requires believing in the lack of evidence. There’s a name for this lack: “incomplete lineage sorting” (ILS).

This is because evolution moved quickly, and many species arose in quick succession. When this happens, different parts of the genome can tell disparate tales of the kinship between the new species. The phenomenon has previously been explained theoretically and is a result of the genetic variation passing from one species to another. If new species then continue to evolve quickly, random chance can end up determining which original genetic variants end up in each lineage. The phenomenon is called incomplete lineage sorting.

But is it plausible to infer the following relationships?

By using the jumping genes, or so-called retrotransposed elements, the Uppsala researchers have found that, for instance, a cuckoo can be more closely related to a hummingbird than a pigeon in a certain part of its genome, while the opposite holds true in another part. The study found numerous examples to corroborate the existence of the phenomenon.

Maybe this is just a problem in the bird family tree. Is that so?

This is one of the first cases in evolutionary research where researchers have been able to document and quantify incomplete lineage sorting far back in time. It is likely a far more common occurrence than previously thought.

What does the original paper in PLoS Biology have to say?  Surprise; the same pattern exists in the mammal tree, and is probably widespread in many lineages.

We conclude that Neoaves diversification is more complex than can be shown in fully bifurcating trees and exhibits a dynamic picture of ILS. The timing of the highly ILS-affected initial super-radiation coincides with the K-Pg extinction of nonavian dinosaurs and archaic birds, suggesting that the abrupt availability of ecological niches was followed by near-simultaneous population isolations via specializations and led to several network-like relationships. The subsequent, decelerated adaptive radiations of waterbirds and raptorial landbirds exhibit less ILS and likely took place after the K-Pg boundary. Interestingly, this time span is similar to placental mammal diversification, which was accompanied by localized and less pronounced ILS than shown here for Neoaves. Finally, and contrary to the expectation that complete genomes will permit full resolution of phylogenies, our genome-level analyses of rare genomic changes yield a broadly bifurcating species tree of Neoaves with local network-like reticulations that probably lie in the anomaly zone. Our study thus provides empirical evidence for a locally confined “hard” polytomy, and we predict that future genome-wide studies of ILS in other adaptive radiations will reveal further examples where a fully bifurcating, universal species tree is an oversimplification of the underlying complexity of speciation.

In this conclusion, they do claim to see a “broadly bifurcating species tree” in birds; the bushy parts “probably lie in the anomaly zone.” That belief, however, relies on various assumptions and tree-building software techniques: e.g., that homoplasy (convergent evolution) was low, and that “bootstrap” support is valid (see excerpt after the commentary*). It doesn’t obviate the fact that maintaining faith in Darwin’s tree requires believing in a “large number of near-simultaneous speciation events of an ancestral species.” The “bushy” appearance is certainly one valid interpretation, they admit: “However, the high bootstrap support (>90%) for some alternative bifurcations could also mean that there are several comparably likely relationships, thus resembling a local network.

Roland G. Roberts does his best to preserve evolutionary ancestry in his commentary on the paper in the same issue of PLoS Biology, but has to admit there are “intractable tangles” in the data. Taking the data at face value, “The prevalence of ILS in the bird phylogeny can be captured more accurately by representing it as a network rather than as a classic tree with simple sequential bifurcations,” he says. One problem with trying to maintain a tree interpretation is that it requires a huge amount of macroevolutionary innovation in a very short period of time. That’s why Casey Luskin at Evolution News & Views last year referred to the “bird explosion” as another challenge to neo-Darwinism downstream of the more famous Cambrian explosion.

In the press release, two of the authors prefer the bush interpretation. They warn other phylogenists, “The more complex kinship patterns that result from this phenomenon [ILS] mean that the Tree of Life should often be understood as a Bush of Life.

Without a tree, Darwinism is dead. You can’t have multiple lineages appearing abruptly and simultaneously. You can’t have disparate groups sharing some similarities that imply one tree, but other traits that imply other trees. Casey Luskin at ENV explains why incomplete lineage sorting is used as an ad hoc, after-the-fact gimmick to maintain belief in common ancestry. Ann Gauger at ENV explains how evolutionists remove inconvenient convergences from the data set by fiat. So what do the data say, in the Cambrian explosion, the bird big bang, and all the other explosive episodes of novelty? William of Ockham would look at it and say, “creation” is the explanation.

Rumor has it that Jonathan Wells is working on an update to his classic Icons of Evolution, showing more icons needing iconoclasm.


*Complexity of the Neoavian Radiation

The probability for the occurrence of ILS depends on Ne in relation to the time between consecutive speciation events [29,30], with Ne correlating positively and time negatively with the expected extent of ILS, respectively. The observed complex genealogical fates of ancestral RE insertion polymorphisms during the initial super-radiation (Fig 4D) therefore suggest that the onset of neoavian diversification was characterized by a large number of near-simultaneous speciation events of an ancestral species with large Ne. Considering the sheer amount of differing allelic combinations that are possible to result from stochastic sorting of ancestral biallelic genetic variation after up to 17 speciation events (Fig 4D), we hypothesize that such complex signals might overrule the underlying species tree-concordant signal, because the latter can be expected to occur rarely under the complex sorting scenario envisioned (cf. Figs 1A–1B and 3A). Considering all theoretically possible RE presence/absence patterns in a five-taxon tree (Fig 1C–1E), ILS across four speciation events requires allelic sorting in each of the descendant lineages, permitting 22 different character distributions that are discordant with the species tree (Fig 6A, S5 Table). Under the model of stochastic sorting of polymorphisms of RE presence/absence (Fig 6) or other types of biallelic variation (e.g., single nucleotides), the probability for the occurrence of hemiplasy surpasses 90% after an ILS duration of seven speciation events (Fig 6D, S5 Table). This may explain why the deepest neoavian bifurcations receive various alternative topologies in the different genome-scale sequence trees of Jarvis et al. [4] (Fig 5D). However, the high bootstrap support (>90%) for some alternative bifurcations could also mean that there are several comparably likely relationships, thus resembling a local network. Alternatively, Salichos & Rokas recently proposed that bootstrapping in phylogenomic analyses can lead to strong support for bifurcations even in the light of strong conflict [39]. Even if one of these genome-scale bifurcating trees reflects the actual neoavian species tree, the verification of such a phylogenetic hypothesis remains challenged by the underlying complex discordances. Finally, the nearly star-shaped topology of this super-radiation (Figs 4A and 5D) may reflect population complexity of the ancestral species, especially if the succession of population isolation during explosive speciation happened in disagreement with prior population structure [40,41].


  • tjguy says:

    ILS. That’s a new one for me. Add it to all the other theory saving devices evolutionists dream up to try to keep their hypothesis from sinking – things like convergence, horizontal gene transfer, co-option of parts, non-evolution in the fossil record – called “stasis”, etc. And then there are all the theory saving ad hoc explanations of cosmology as well! Geology has its share as well as they try and explain away the Cambrian Explosion.

Leave a Reply