Opinion
Diversity dynamics: molecular phylogenies need the fossil record

https://doi.org/10.1016/j.tree.2010.05.002Get rights and content

Over the last two decades, new tools in the analysis of molecular phylogenies have enabled study of the diversification dynamics of living clades in the absence of information about extinct lineages. However, computer simulations and the fossil record show that the inability to access extinct lineages severely limits the inferences that can be drawn from molecular phylogenies. It appears that molecular phylogenies can tell us only when there have been changes in diversification rates, but are blind to the true diversity trajectories and rates of origination and extinction that have led to the species that are alive today. We need to embrace the fossil record if we want to fully understand the diversity dynamics of the living biota.

Section snippets

Molecular phylogenies and rates of diversification

Understanding the patterns and processes of diversification has long been of interest to paleontologists as we use the fossil record to document biodiversity change through time [1]. However, interest in diversity dynamics among biologists and the greater public has grown, particularly as we become aware of the impact of humans on the biosphere. Among biologists, the study of biodiversity dynamics was invigorated by the proposal that the rates and processes of diversification could be inferred

Describing patterns and inferring processes from molecular phylogenies

To study patterns of diversification, Pybus and Harvey [12] introduced the γ statistic (Box 1). This is a simple tool for determining if a molecular phylogeny is consistent with a constant rate of diversification, or if there has been a decrease in the diversification rate (inferred when molecular phylogenies have significantly negative γ values). There is now a sizeable literature which indicates that many clades have decreasing diversification rates 12, 15, 16, 17, in fact for about half of

Can diversification dynamics be gleaned without fossils?

As pointed out by Ricklefs [2], a molecular phylogeny, by virtue of starting with the living biota, must show a pattern of increasing diversity with time regardless of the true diversity dynamics of the clade. This perceptual bias is reflected in the methods developed for inferring diversity dynamics from molecular phylogenies. These typically assume that the speciation rate is on average higher than, or occasionally equal to 10, 19, the extinction rate. However, it is clear from the fossil

‘Time-traveling’ with computer simulation

As discussed above, if there is extinction, computer simulations show that diversity-dependent diversification can be detected in a molecular phylogeny only if the initial speciation rate is sufficiently high to ameliorate the erosive effect of the extinction. A metric that quantifies how high the initial speciation needs to be is the LiMe ratio (the ratio of the initial speciation rate to the equilibrium extinction rate) [21]. For example, if the speciation rate is linearly density-dependent

Is there too much evidence of diversity-dependent diversification?

The results from computer simulations described above suggest there are too many phylogenies that show decreasing rates of diversification for the process of density dependence to be ubiquitous. This counter-intuitive conclusion is readily explained. If diversity-dependent diversification was the rule for all clades, and if extinction is pervasive (as suggested by the fossil record), we would expect to see evidence of density dependence only in the subset of clades that had large LiMe values

Do decreasing rates of diversification indicate clades in decline?

Pybus and Harvey [12] recognized that a failure to sample all the species in a clade will result in more negative γ values than would otherwise be measured. To date, the literature has considered under-sampling only in the context of species that we know are extant, but that we have failed to sample. However, if several taxa have recently become extinct, there is also a failure to sample a subset of taxa. In this case there is no remedy for the under-sampling; the missing taxa are beyond reach.

Accessing the past through the fossil record

Computer simulation shows that there are many ways to generate similarly shaped phylogenies. At present, the only reliable way to discriminate among the various possibilities is to directly access the past, i.e. through the fossil record. To illustrate the power of using the fossil record to discriminate between various diversification scenarios, we consider just one example: the diversification of the Cetacea. The rich and well-understood fossil record of the Cetacea shows a very different

Cetacean diversity dynamics: the molecular phylogenetic perspective

Analysis of molecular phylogenetic data for 87 of the 89 currently recognized species of the Cetacea indicates a non-significantly negative γ value [27] (Figure 2a), and numerical analysis of the phylogeny suggests the clade has undergone three specific increases in diversification rate. Thus, the molecular phylogeny suggests exponential growth. Ignoring human impacts, this implies that, in terms of diversification, this is the best of all possible times to be a cetacean. Finally, numerical

Cetacean diversity dynamics: the paleontological perspective

The fossil record of cetaceans is relatively complete, has been recently re-evaluated 28, 29, 30, and is readily accessed through the Paleobiology Database. Nonetheless, the incomplete preservation typical of most fossils makes distinguishing closely related species difficult. For this reason, the analysis of the cetacean fossil record is based on fossil genera (as is commonly the case in paleontological studies) [31]. In the analysis below, we assume that there is no substantial difference in

Why is there conflict between molecular phylogenies and the fossil record?

The reason why the patterns seen in the fossil record and those inferred from molecular phylogenies can be so different is simple. The diversity we see at present is a consequence of the long-term balance between origination and extinction, but this long-term net rate of diversification inferred from a molecular phylogeny need bear no relationship to the current or any past rate of diversification. The cetacean data demonstrate this well. In this case, the net rate of diversification over the

Dealing with complex temporal trajectories

Many molecular phylogenies have patterns of lineage accumulation that are too complex to be adequately characterized by the relatively simple γ statistic or even by more complex tools (e.g. [34]). For example, a phylogeny of Australian and southern African legumes [35] shows an initial diversification followed by a long plateau with essentially no lineage accumulation, followed by a steady and relatively rapid rate of accumulation of lineages to the present. Simple diversification models cannot

Conclusions

Much has been claimed about the rates and patterns of diversification of the living biota from the analysis of molecular phylogenies, but we feel that only a subset is reliable. The most reliable statements concern the identification of clades with changes in their rates of diversification [12]. If one is willing to accept that decreasing rates of diversification are due to diversity-dependent diversification, then it appears that the only way to explain an observed decrease in diversification

Acknowledgements

We are grateful to Jonathan Losos for encouragement, the Paleobiology Discussion group at Harvard University, as well as Lee Hsiang Liow, Nathalie Nagalingum, John Alroy and Swee Peck Quek for discussion, and Mark Uhen for help with the cetacean fossil record. We also thank Paul Harvey, Dan Rabosky, Michael Foote and Gene Hunt for their helpful reviews of the manuscript. This is Paleobiology Database Publication 114.

Glossary

Boundary-crosser method
used to estimate the diversity at the boundary of adjacent geological time intervals by counting the number of taxa that must have crossed the boundary because they are known before and after it. This method has the desirable property of assuring the co-existence of taxa.
Chronogram
a phylogeny with branch lengths adjusted so that they are proportional to absolute time. Also known as a ‘time tree’.
Crown group
a monophyletic clade that contains all extant members of the clade

References (57)

  • R.E. Ricklefs

    Estimating diversification rates from phylogenetic information

    Trends Ecol. Evol.

    (2007)
  • J. Phillips

    Life on the Earth: Its Origin and Succession

    (1860)
  • S. Nee

    Tempo and mode of evolution revealed from molecular phylogenies

    Proc. Natl. Acad. Sci. U.S.A.

    (1992)
  • S. Nee

    Extinction rates can be estimated from molecular phylogenies

    Philos. Trans. R. Soc. Lond., B, Biol. Sci.

    (1994)
  • S. Nee

    The reconstructed evolutionary process

    Philos. Trans. R. Soc. Lond., B, Biol. Sci.

    (1994)
  • P.H. Harvey

    Phylogenies without fossils

    Evolution

    (1994)
  • A. Purvis

    Phylogenetic approaches to the study of extinction

    Annu. Rev. Ecol. Syst.

    (2008)
  • D.L. Rabosky

    Extinction rates should not be estimated from molecular phylogenies

    Evolution

    (2010)
  • E.A. Thompson

    Human Evolutionary Trees

    (1975)
  • J. Hey

    Using phylogenetic trees to study speciation and extinction

    Evolution

    (1992)
  • B. Rannala et al.

    Probability distribution of molecular evolutionary trees: a new method of phylogenetic inference

    J. Mol. Evol.

    (1996)
  • O.G. Pybus et al.

    Testing macro-evolutionary models using incomplete molecular phylogenies

    Proc. R. Soc. Lond. B. Biol. Sci.

    (2000)
  • W.P. Maddison

    Estimating a binary character's effect on speciation and extinction

    Syst. Biol.

    (2007)
  • M.E. Alfaro

    Lineage-specific diversification rates and high turnover in the history of jawed vertebrates

    Proc. Natl. Acad. Sci. U.S.A.

    (2009)
  • M.A. McPeek

    The ecological dynamics of clade diversification and community assembly

    Am. Nat.

    (2008)
  • A.B. Phillimore et al.

    Density-dependent cladogenesis in birds

    PLoS Biol.

    (2008)
  • D.L. Rabosky et al.

    Explosive evolutionary radiations: decreasing speciation or increasing extinction through time?

    Evolution

    (2008)
  • D.L. Rabosky

    Ecological limits on clade diversification in higher taxa

    Am. Nat.

    (2009)
  • D.L. Rabosky

    Heritability of extinction rates links diversification patterns in molecular phylogenies and fossils

    Syst. Biol.

    (2009)
  • D.L. Rabosky

    Ecological limits and diversification rate: alternative paradigms to explain the variation in species richness among clades and regions

    Ecol. Lett.

    (2009)
  • T.B. Quental et al.

    Extinction during evolutionary radiations: reconciling the fossil record with molecular phylogenies

    Evolution

    (2009)
  • L.J. Revell

    Underparameterized model of sequence evolution leads to bias in the estimation of diversification rates from molecular phylogenies

    Syst. Biol.

    (2005)
  • A. Purvis

    Temporal patterns in diversification rates

  • Liow, L.H. et al. When can decreasing diversification rates be detected with molecular phylogenies and the fossil...
  • C.R. Marshall

    Explaining latitudinal diversity gradients

    Science

    (2007)
  • M. Foote

    Symmetric waxing and waning of marine animal genera

    Paleobiology

    (2007)
  • M.E. Steeman

    Radiation of extant Cetaceans driven by restructuring of the oceans

    Syst. Biol.

    (2009)
  • M. Uhen et al.

    Diversity estimates, biases, and historiographic effects; approaches to resolving cetacean diversity in the Tertiary

    Palaeontol. Electronica

    (2007)
  • Cited by (293)

    View all citing articles on Scopus
    View full text