Mapping genetic and taxonomic species diversity at regional and continental scales

1
Mapping genetic and taxonomic species diversity
at regional and continental scales

• Shawn Laffan

Sophie Bickford
CSIROCentre for Plant Biodiversity Research
2
Species Diversity

• Conservation prioritisation
• Evolution dispersion
• These are geographic phenomena
• multi-scaled
• expensive to directly sample
• Need a source of data

3
Museum data

• Rich source of geocoded information
• 6 million plant specimens in Australian Herbaria
• Australian Virtual Herbarium approx. 40
  digitised in 2002
• every specimen description has a geographic
  coordinate or reference
• No true absence data
• Variable sampling effort

4
Data set

• Genus Pultenaea
• Eastern Australian species used
• 91 species
• 3 clades used
• Genetic similarity data for 49 species
• Aggregated to 50 km resolution
• reduce effects of variable sampling effort
  roadmap effect
• Lamberts conic conformal projection
• 5799 samples (after aggregation)

5
Species richness
6
Genetic diversity

• Use of a taxonomy assumes it is reliable
• but taxonomy is constantly revised
• implies equal similarity between all species
• If we know the genetic similarity of species then
  we can soften the effects of taxonomy
• number of base pairs in common
• two sections of trnL-F and ndhF chloroplast genes
• base pairs 900 trnL-F, 674 ndhF
• matrix of similarities

7
Genetic diversity

• We know where the species occur
• we can therefore map genetic variation
• useful for evolutionary history
• Explicitly spatial
• mean and coefficient of variation of genetic
  similarity between all species within a local
  neighbourhood
• Implicitly spatial
• agglomerative clustering with average genetic
  similarity as similarity metric

8
Local genetic diversity, all species
9
Local genetic diversity, all species
Local genetic diversity by clade
10
Implicitly spatial

• Agglomerative clustering of cells
• Allows for disjoint distributions
• Taxonomic similarity metric between two cells
• Jaccard distance
• Shared presence divided by total number of
  species in two cells
• a / (a b c)
• a Num species occurring in both cells
• b c Num species occurring in only one cell

11
Jac 3 / (3 4 4) 3 / 11 0.27

• ACT Brumbies
• B. greganii
• B. larkhamii
• B. gitaeuii
• B. mortlockii
• B. rathboneii
• B. gerardii
• B. roffii

• Australian Wallabies
• B. greganii
• B. larkhamii
• R. flatleyii
• B. mortlockii
• W. tuqirii
• R. sailorii
• W. rogersii

12
Agglomerative clustering linkage heuristic

• Similarity of newly aggregated classes
• Unweighted Pair-Group Method
• UPGM-A
• average similarity of merged classes
• common species counted more than once
• staircase effect
• UPGM-R
• recalculate similarity matrix from first level
• each species counts only once in each merger

13
Average Linkage
Recalculated Linkage
Similarity
Similarity
14
(No Transcript)
15
(No Transcript)
16
(little variation within clades 2, 3a 3b)
17
(No Transcript)
18
What next?

• Cross-validation for genetic analyses
• More sophisticated measures of genetic similarity
• able to cope with more diverse groups
• mean CofV not ideal measures for non-symmetric
  distributions
• Software distribution / web analysis service
• http//www.biodiverse.unsw.edu.au