Trends in Ecology & Evolution
ReviewDeciphering ancient rapid radiations
Introduction
Phylogenies are crucial to our understanding and explanation of the origin and evolution of the major adaptations and lineages of organisms on Earth. Yet, despite recent advances in the availability of data and methodologies for investigating the historical relationships among organisms, many major phylogenetic patterns remain poorly resolved. Rapid evolutionary radiations have been proposed to explain poorly resolved phylogenies in many groups of organisms, including aphids, black flies, bees, birds, turtles, mammals and higher plants (Table 1).
Classic tales of radiation include the Cambrian explosion of animal phyla, the Cretaceous origin of angiosperms, the diversification of birds and mammals, and controversies over the origins of highly social behavior in the bees. Identifying which of these cases of diversification represent genuine rapid radiations, and which do not, has broad ramifications in evolutionary biology. However, deciphering patterns of ancient radiation is inherently problematic, often requiring an array of data sources and analytical techniques at the cutting edge of current knowledge. Why is the evolutionary history of these radiations so much more difficult to reconstruct than it is for many other phylogenetic questions?
Section snippets
Some phylogenies are harder to estimate than others
As the methodology of phylogeny estimation has become more sophisticated, diverse and efficient, our ability to unravel successfully the phylogenetic histories of many groups of organisms has improved. With the advent of molecular systematics, the evolutionary relationships of many organisms (e.g. placental mammals 1, 2), are now thought to be understood with a high degree of confidence. This is despite considerable controversy remaining over the advantages and limitations of various
Difficulties underlying ancient rapid radiations
The biology of genomes and gene sequences also conspires to make the phylogenetic reconstruction of radiations difficult. Their evolutionary properties and the problems that they cause can be straightforward to diagnose, but when co-occurring, they can easily confound one another and complicate the resolution of rapid radiations.
As the time between divergences becomes shorter, it eventually enters the time span of lineage-sorting problems (Figure 2), where individual gene trees might not
Ancient rapid radiation or inadequate data?
Ancient rapid radiations lead to reconstructed phylogenies with low support for basal relationships. However, poor internal branch support in phylogenies can also be obtained by: (i) using molecular or morphological data that are not variable enough at the appropriate level; (ii) having data sets that strongly conflict with one another; (iii) applying inappropriate phylogenetic methods and substitution models; or (iv) not having enough data to solve the problem. So when does it appear that
Eliminating character conflict as an explanation
Short branch lengths (i.e. low support) in phylogenies can result from different characters or data sources providing support for conflicting trees, rather than from the absence of support. Because this kind of conflict among characters can stem from causes unrelated to the absence of sufficient time for support to accumulate in clades, it can be useful to first test whether short internal branches are due to character conflict. If they are, it is likely that additional data will more strongly
How much data will be required to resolve relationships?
If there has been simultaneous diversification of multiple taxa within a group, the phylogeny is more accurately represented by a polytomy rather than by a bifurcating tree [30]. Thus, we might expect in these cases that no amount of data will resolve a bifurcating tree. If instead the time between lineage splits is short but at least detectable, it is possible to estimate (i.e. extrapolate) how much additional data it would take to resolve the branching order.
Several statistical tests for
Outstanding questions
The unearthing of ancient rapid radiations has led to several questions that are still unresolved. For instance, are there ancient patterns of divergence that are impossible to resolve using molecular sequence data? Theoretical explorations 22, 23 suggest that there are, but empirical studies are still largely data limited. If sequence data do not suffice, will rare genomic changes such as gene content, presence–absence of biochemical pathways, gene arrangements, intron and transposon
The value of data exploration
Modern phylogenetic analysis has become an increasingly complex task as the focus of study widened to encompass the full range of hierarchical questions. It is evident that many phylogenetic histories will require significant scientific effort to decipher, not necessarily because our methods are still poor but because the histories are truly challenging to recover. In some cases, we will exceed the limits of resolution of certain classes of data and many phylogenetic methods. It will be
Acknowledgements
The original motivation for this review came from molecular phylogenetic research on microgastrine braconid wasps funded originally by USDA NRI grant 9501893 and continued under NSF grant DEB 0316566. A Sabbatical Leave Grant to J.B.W. from the Allan Wilson Centre facilitated the writing of the review during February–April 2006, as well as valuable input from Sydney Cameron, Barbara Holland, Daniel Huson, David Penny and Mike Steel.
Glossary
- Consensus network
- a phylogenetic network that displays the splits found in source trees that have the same (completely overlapping) set of taxa; thresholds can be implemented to display all the splits in all trees, or only the splits that occur in x% of source trees; a network version of a consensus tree.
- Hadamard transform
- a discrete Fourier transform that enables easy reversible translation between distances and site patterns (partition frequencies); given a tree with branch lengths, one can
References (67)
Should phylogenetic models be trying to ‘fit an elephant’?
Trends Genet.
(2005)- et al.
Should we be worried about long-branch attraction in real data sets? Investigations using metazoan 18S rDNA
Mol. Phylogenet. Evol.
(2004) - et al.
Trees for bees
Trends Ecol. Evol.
(2001) - et al.
The place of Amborella within the radiation of angiosperms
Trends Plant Sci.
(2005) Gene trees and species trees are not the same
Trends Ecol. Evol.
(2001)The probability of topological concordance of gene trees and species trees
Theor. Popul. Biol.
(2002)- et al.
A phase transition for a random cluster model on phylogenetic trees
Math. Biosci.
(2004) Chloroplast genome phylogenetics: why we need independent approaches to plant molecular evolution
Trends Plant Sci.
(2005)The use of genome-level characters for phylogenetic reconstruction
Trends Ecol. Evol.
(2006)- et al.
Phylogenetic signal in the COI, 16S, and 28S genes for inferring relationships among genera of Microgastrinae (Hymenoptera: Braconidae): evidence of a high diversification rate in this group of parasitoids
Mol. Phylo. Evol.
(1999)