Biodiversity conservation using Phylogenetics on a global scale Klaas Hartmann TAFI, University of Tasmania

1
Biodiversity conservation using Phylogenetics on
a global scaleKlaas HartmannTAFI, University of
Tasmania
2
Extinction

• gt100,000 extincts per year
• 100-1000 times background rate
• 39 of IUCN Red List species are endangered

3
Charismatic Megafauna
4
Biodiversity measures

• Species richness
• Species definition unclear
• Species distinctiveness not considered
• Phylogenetic diversity (PD)
• Dan Faith and Ross Crozier

5
The Noahs Ark Problem

• Species have a survival probability which can be
  increased at a cost
• Objective maximise future expected PD
• Some algorithms to produce optimal solutions have
  been developed

• K. H. and M. Steel. (2006). Maximimizing
  phylogenetic diversity in biodiversity
  conservation greedy solutions to the Noah's Ark
  problem. Systematic Biology 55(4), 644-651.
• K. H. and M. Steel. Phylogenetic diversity From
  combinatorics to ecology. Book chapter for
  Reconstructing evolution New mathematical and
  computational approaches (eds. O. Gascuel and M.
  Steel) Oxford University Press
• T. Gernhard, K. H. and M. Steel. Stochastic
  properties of generalised Yule models, with
  biodiversity applications. Journal of
  Mathematical Biology

6
NAP with uncertain parameters
7
Issues

• Too complex
• Difficult to integrate with existing approaches

8
Species Specific Indices

• An SSI attributes a single value to each species
• Some are easy to understand
• Examples
• Pendant edge
• Fair proportion
• Equal splits
• Shapley value

9
SSI vs PD

• D.W. Redding, K. H., A. Mimoto, D. Bokal, M.
  Devos and A.O. Mooers. Evolutionarily distinct
  species capture more phylogenetic diversity than
  expected. Journal of Theoretical Biology 251,
  606-615.

10
(No Transcript)
11
(No Transcript)
12
Bird EDGE

• 9,787 species
• The Data
• Half of the species have sequence information
• All have taxonomic information
• Hundreds of (conflicting) expert trees
• How do we combine this information???

13
Bird EDGE approach

• Species are divided into patches
• All expert trees for a patch are combined
• Taxonomic information is used to enforce
  monophyletic genera where possible
• The constrained patch trees are run with a
  modified version of mrBayes 3.1.2

14
Blue Fern

• Two rack IBM Blue Gene/L
• 4096 cores
• 1 Terabyte RAM
• 11.2 Teraflops
• 53kW power consumption
• One run takes about 5 cpu years

15
Bird EDGE approach

• A BEAST skeleton tree is used to provide
  probability distributions of the root age for
  each patch
• EDGE Indices are produced

16
Acknowledgements

• Financial contributors
• University of Canterbury (NZ)
• Allan Wilson Centre for Molecular Ecology and
  Evolution (NZ)
• Google Inc. (USA)
• Simon Fraser University (Vancouver)
• I am gratefully indebted to
• Arne Mooers
• Mike Steel
• Walter Jetz
• David Redding
• Gavin Thomas
• Tanja Gernhard
• Too many others to list!