Evolutionarily distinctive species often capture more phylogenetic diversity than expected

https://doi.org/10.1016/j.jtbi.2007.12.006Get rights and content

Abstract

Evolutionary distinctiveness measures of how evolutionarily isolated a species is relative to other members of its clade. Recently, distinctiveness metrics that explicitly incorporate time have been proposed for conservation prioritization. However, we found that such measures differ qualitatively in how well they capture the total amount of evolution (termed phylogenetic diversity, or PD) represented by a set of species. We used simulation and simple graph theory to explore this relationship with reference to phylogenetic tree shape. Overall, the distinctiveness measures capture more PD on more unbalanced trees and on trees with many splits near the present. The rank order of performance was robust across tree shapes, with apportioning measures performing best and node-based measures performing worst. A sample of 50 ultrametric trees from the literature showed the same patterns. Taken together, this suggests that distinctiveness metrics may be a useful addition to other measures of value for conservation prioritization of species. The simplest measure, the age of a species, performed surprisingly well, suggesting that new measures that focus on tree shape near the tips may provide a transparent alternative to more complicated full-tree approaches.

Introduction

With increasing extinction there is a pressing need to effectively prioritize species for conservation. Many non-exclusive currencies are used, e.g. threat status, ecological importance, social or intrinsic value, and financial cost (for discussion, see Crozier, 1992; Weitzman, 1998; Andelman, 2004; Avise, 2005). Here we focus on the evolutionary distinctiveness of species in the context of their conservation. In particular, we examine the trade-off between prioritizing the most evolutionary distinctive species in a tree and prioritizing sets of species that best represent the whole tree.

Conservation biologists have approached the goal of representing a phylogenetic tree from two angles. Both approaches use information about the relatedness among tips (usually species), but one (phylogenetic diversity, PD) is a group measure, while the other (evolutionary distinctiveness) is a species-specific property. To illustrate the connections between the two approaches, consider the order Sphenodontia. This order contains the two species of tuatara and is sister to Squamata (Snakes, Lizards, Amphisbaenians), which contains ∼6200 species. From a macroevolutionary perspective, if one species from each order were equally threatened, priority should go to a tuatara before any lizard, snake, or amphisbean species, because both tuatara species are highly distinctive and so contain a disproportionately large proportion of the PD contained within the two orders. However, if only two species of the ∼6202 species were to be preserved (an unlikely scenario), the tree would be best represented with a set that included only one of the two tuataras, and one Squamate. More generally, any subset that did not contain one of the tuataras would be suboptimal.

The idea of comparing the relative PD represented by sets of species in order to prioritize sets that contribute more unique evolution was pioneered by Vane-Wright et al. (1991) and by Faith (1992). The PD of a set of species is generally measured as the sum of the branch lengths of the tree containing those species and the root (see Fig. 1 and Faith and Baker, 2006). Sets of species with maximal PD can be found using simple algorithms (Steel, 2005; Pardi and Goldman, 2005; Minh et al., 2006), and the approach has also been useful when ranking predefined areas (which define subsets) for conservation (see, e.g. Forest et al., 2007). The PD approach has also been extended to include species survival probabilities and conservation costs and budgets (Weitzman, 1998; Hartmann and Steel, 2006, Hartmann and Steel, 2007; Pardi and Goldman, 2007).

In parallel, systematists have proposed metrics for how much unique evolution a particular species contributes to some larger set (again, see Vane-Wright et al., 1991; see also May, 1990; Nixon and Wheeler, 1992; Pavoine et al., 2005; Redding, 2003; Redding and Mooers, 2006; Isaac et al., 2007). Early attempts to attribute a score of evolutionary distinctiveness to individual species (May, 1990; Nixon and Wheeler, 1992; Vane-Wright et al., 1991) used only tree topology, and relied on the fact that basal and evolutionary isolated species have fewer nodes between the tip and the root. Recent workers (Isaac et al., 2007; Pavoine et al., 2005; Redding and Mooers, 2006; Steel et al., 2007; Weitzman, 1998) have suggested distinctiveness measures (outlined below) that use both topology and internal branch lengths to measure species isolation. All such measures have one thing in common: they give species that have many and closer relatives less value than they give species with fewer and more distant relatives.

Evolutionary distinctiveness measures and PD approaches differ in substantial ways. PD is uninformative for any one species on an ultrametric tree—all single species are the same distance from the root and so receive the same value. Many current conservation approaches (e.g. endangered species lists) rely on having species ranked in order of priority. Current PD approaches offer no such order. To overcome this, species within any optimal set chosen could be ordered by arranging them according to their evolutionary distinctiveness, or alternatively, species could be chosen according to natural species-specific indices that ensure optimal sets (DB, MdV, KH, unpublished results). However, there will be as many possible rankings of species produced by these approaches as there are PD maximizing solutions for the set of candidate species, and thus may be difficult to implement at the management level.

More importantly, the amount of PD saved is only optimal if all the species that are selected are subsequently protected. If any species in the selection are lost, new optimal sets are possible. Finally, it may be difficult to find optimal sets of species if there are large numbers of species to prioritize, and other complex factors such as cost of conserving individual species are considered (Weitzman, 1998).

The recently developed species-specific measures of evolutionary distinctiveness may, in comparison, represent a flexible and transparent conservation tool to promote “evolutionary value” in the current legislative climate. However, and importantly, they have not been designed to capture total PD. If sets of evolutionary distinctive species did capture substantial PD, then the species-specific measures would be doubly useful, highlighting the most individually distinctive species and helping preserve more of the tree of life. This is the focus of the present study. We first outline the distinctiveness measures we tested, and then consider aspects of the underlying tree that might affect the relationship between distinctiveness measures and PD.

Section snippets

The measures

There exists a small group of older, node-based measures of distinctiveness (reviewed in Pavoine et al., 2005). This group suffers from poor resolution among species (i.e. it gives rise to many tied values) and also loses any information conferred by branch lengths (Pavoine et al., 2005). However, preliminary analyses (results not shown) identified May's (1990) modification of Vane-Wright et al.'s (1991) measure (VW) as the best node-based measure for capturing PD, and so we used it as a basis

Tree shape

Tree shape is likely to be an important factor in determining how effectively distinctiveness measures capture PD. We outline three measures of tree shape and how we think they will affect the PD ‘capture rate’.

The balance of the tree towards the root (Ic; Colless, 1982; Heard, 1996) will dictate whether, when randomly selecting species, some internal branches are more likely to be chosen than others. Repeatedly selecting closely related species will decrease the total amount of PD represented,

Methods

Our primary data set consisted of 5000 simulated “Yule trees” (Yule, 1924) with 16 tips, created using Bio::Phylo package (Vos, 2006). In order to test the sensitivity of the findings to the process model used, we also simulated “Hey trees” (Hey, 1992): these are the tree shapes expected under the Moran (1951) coalescent and have Yule topologies, but different waiting times, with more splits occurring near the present. Using Yule trees in this situation is conservative, as the principal way

Results

For our small simulated trees, all five distinctiveness metrics perform well at capturing PD for the majority of tree shapes, and generally capture significantly more of the tree than would a random sample (Table 1). The extra amount of the tree that would be captured by selecting species using an distinctiveness measure as opposed to randomly selecting them, e.g. the difference between random selection and ES, ranges from 4% to 9%.

When two species are chosen from the trees, the measures

Discussion

We highlight four main findings: first, for the small trees considered here, all five distinctiveness metrics perform well at capturing PD for the majority of tree shapes, and generally capture significantly more of the tree than would randomly selecting species (Table 1). The absolute improvement, however, was modest. For example, on the tree of Aves (phylotaxonomy based on Monroe and Sibley, 1993), selecting 50% of the species in the tree using ES captures a further 8 billion years of

Conclusion: implications for conservation

The criterion most often used to prioritize species for conservation is threat status (Possingham et al., 2002). While threatened species tend to come from species-poor groups (Purvis et al., 2000), threat status may not be much more effective at capturing PD than choosing species at random (Redding and Mooers, 2006). Likewise, evolutionary distinctiveness and threat are only very weakly correlated for birds and mammals (AOM and DWR, unpublished observations). If one conservation goal is the

Acknowledgments

We thank Mike Steel and IRMACS for fostering our collaboration, the fab*-lab at SFU for commenting on some of the ideas presented here, and Carolyn Huston for statistical advice. This work was funded by NSERC, Canada, Allan Wilson Centre and the New Zealand Marsden Fund. D.B. and M.D. were supported by Bojar Mahar's Canada Research Chair in Graph Theory at SFU.

References (45)

  • D.H. Colless

    Phylogenetics: the theory and practice of phylogenetic systematics. Book review

    Syst. Zool.

    (1982)
  • D.P. Faith et al.

    Phylogenetic diversity (PD) and biodiversity conservation: some bioinformatics challenges

    Evol. Bioinform. Online

    (2006)
  • F. Forest et al.

    Preserving the evolutionary potential of floras in biodiversity hotspots

    Nature

    (2007)
  • Haake, C.-J., Kashiwada, A., Su, F.E., 2005. The Shapley value of phylogenetic trees. IMW Working Paper No....
  • K. Hartmann et al.

    Maximizing phylogenetic diversity in biodiversity conservation: greedy solutions to the Noah's Ark problem

    Syst. Biol.

    (2006)
  • K. Hartmann et al.

    Phylogenetic diversity: from combinatorics to ecology. Book chapter for: Reconstructing Evolution: New Mathematical and Computational Approaches

  • S.B. Heard

    Patterns in phylogenetic tree balance with variable and evolving speciation rates

    Evolution

    (1996)
  • J. Hey

    Using phylogenetic trees to study speciation and extinction

    Evolution

    (1992)
  • Isaac, N.J.B., Turvey, S.T., Collen, B., Waterman, C., Baillie, J.E.M., 2007. Mammals on the EDGE: conservation...
  • F. Matsen

    A geometric approach to tree shape statistics

    Syst. Biol.

    (2006)
  • R.M. May

    Taxonomy as destiny

    Nature

    (1990)
  • M.A. McPeek et al.

    Clade age and not diversification rate explains species richness among animal taxa

    Am. Nat.

    (2007)
  • Cited by (99)

    View all citing articles on Scopus
    1

    These authors contributed equally.

    View full text