A phylogenetic view of diversity
Dan Rosauer
Research School of Biology – Australian National University
Centre for Biodiversity Analysis
Biodiversity in R workshop – Day 3
Gehyra koira
Photo: Matt Summerville
Plan for today
• Why use phylogenies to study spatial patterns of
diversity (lecture)
• Phylogenetic metrics of biodiversity and how to
calculate them (lecture and exercises)
• Modelling diversity below species level (lecture and
exercises)
• Phylogenies in conservation planning including
• Phylogenetic GDM models
• Evaluation
plus food and coffee along the way if you’re good!
and finally, those who want head to the ANU bar at 5
This lecture will cover…
• Brief introduction to phylogenies
• Why use phylogenies to represent
biodiversity?
– for conservation
– for macroecology
– to study biota where taxonomy is inaccurate or
incomplete
A brief introduction to phylogenies and
tree-thinking
A primer on phylogenies
• phylogeny is the true evolutionary history of a group of
organisms
• a phylogenetic tree is a graphical representation of the
ancestor-descendant relationships of a group.
• ideally, a reconstructed phylogeny would be based on
complete, unambiguous fossil series along each lineage
A phylogenetic tree is a hypothesis
about the ancestry of a group.
Hominids
Elephants
Homology: different characters in different species derived from a
common ancestral character
Homologous characters: different but same
human horse
humans
cats
Radius
whales
Ulna
birds
four limbs forelimbs
evolved evolved
into wings
crocodiles
bat
bird whale
Humerus
horses
bats
cat
Carpal
Analagous (convergent) characters: same but different
Using parsimony to reconstruct phylogeny
taxon
character
1
2
3
4
5
6
Outgroup
0
0
0
0
0
0
Species A
0
1
0
0
1
1
Species B
1
1
0
1
0
1
Species C
0
0
1
1
0
0
The most parsimonious tree is the one that requires the fewest character state changes
O
8 changes:
A
B
1
2
5
6
2
4
3
6
C
O
7 changes:
C
3
4
A
B
5
4
6
2
1
O
9 changes:
B
A
1
2
2
4
6
5
6
3
4
C
DNA bases as characters
• the positions of bases must be homologous
• finding homologous positions is complicated by indels (insertions and deletions)
• sequences must be correctly aligned before analysis
Problems with applying parsimony to molecular sequence data:
• simplistic model of character evolution: no allowance for multiple hits or
variation in substitution rates
• long branch attraction: rapidly evolving branches wrongly clustered together
Reconstructing phylogenies
Kinds of Data
Phenotypic data: e.g. morphology, physiology, behaviour
Molecular data: e.g. DNA sequences, protein sequences, DNA
hybridization
Kinds of Methods
Distance-based methods: e.g. Neighbor-joining, UPGMA
Character-based methods: e.g. parsimony, maximum likelihood,
Bayesian
Current methods increasingly rely on models of
the process of molecular evolution
unrooted: no
directionality
cladogram: directional, but no
timescale
sp5
sp4
sp5
sp3
sp4
sp3
sp2
sp1
sp2
sp1
phylogram: branch lengths
reflect amount of evolution
chronogram: branch lengths
calibrated to real time
sp5
sp5
sp4
sp4
sp3
sp3
sp2
sp2
sp1
sp1
40
30
20
10
Millions of years
0
Phylogenetic terms
branches (or edges):
internal branch
terminal branch
sp5
tips
sp4
nodes
clade
ingroup
root, or most
recent common
ancestor (MRCA)
of ingroup
sp3
sp2
sister
species
sp1
sp X
outgroup
A time calibrated phylogeny of whales
Types of uncertainty
• Topology
Confidence estimate
on topology
•
•
Which species / tips are
most closely related
Support values on nodes
• Branch dates / length
•
Confidence intervals on
date / branch length
•
How much change has
occurred along a branch?
How old is node?
A Bayesian posterior distribution of
estimates of the true tree
Xiong et al. 2009
Geophylogeny of katydids in Hawaii
http://kiwi.cs.dal.ca/GenGIS/File:Hawaii.png
Go to: www.onezoom.org
The argument for using evolutionary
diversity in conservation
Terrestrial Mammal Species Richness
Hawkins et al (2012) Journal of Biogeography
Cavia intermedia
Extent of occurrence: 0.04 km2
Biodiversity is distributed
unequally between areas
Vulpes vulpes
Extent of occurrence: ~70,000,000km2
Frog species endemism
Slatyer, Rosauer & Lemckert (2007) Journal of Biogeography
Centres of endemism
– Concentration of species with
small range sizes
– Causes
• Isolation (by physical or
environmental barriers)
• Habitat heterogeneity
• Range contraction
– Why is endemism important?
• Conservation
• Biogeography
Biological diversity is also
distributed unequally between
species
Solenodon paradoxus
S. paradoxus
S. cubanus
Roca et al. (2004)
Vane-Wright et al 1990
Systematics and the Agony of Choice
Biological diversity is distributed unequally
between species as well as between areas
The Wollemi Pine
Knapp et al. 2007 Sys. Biol.
Improving conservation decisions
• 1980’s - growing concern about large scale loss of species
• Scientists were seeking methods to increase conservation
benefit from limited resources
Improving conservation decisions
• 1980’s - growing concern about large scale loss of species
• Scientists were seeking methods to increase conservation
benefit from limited resources
4
A
3
A
B
PD =
1
1
C
2
1
∑L
{c∈C }
c
C
B D
D
PD = 1 + 3 + 2 + 1 = 7
Phylogenetic Diversity (PD)
Faith (1992)
Advantages of PD
• captures more information on
diversity
• robust to taxonomic uncertainty
• doesn’t even require taxonomy
Pomaderris elliptica
Calculating
Phylogenetic Diversity
Faith (1992)
Part of an ITS tree for tribe Pomaderreae (Rhamnaceae) Kellerman et al (2005)
Phylogenetic Diversity - Hylids
Compared to species count
PD shows:
↑ tropics more diverse
↓ Central east coast less diverse
Three uses of trees for conservation
1. Noah’s Ark Problem
• There is not enough room on ‘the ark’ for all taxa. Which
n taxa should one choose?
• Answer
– the set of taxa which capture the most branch length across the
tree (of course it’s more complicated…)
But is this useful in the real world?
Three uses of trees for conservation
2. Evolutionary Distinctiveness (ED)
How much does each species contribute to overall PD?
ED gives a value to tree tips, not areas, by allocating the length of each branch
equally to its descendent tips.
ED
4
3
A
4/1
=
4
B
3 / 1 + 1/3
=
3.33
C
1/1+2/2+1/3 =
2.33
D
1/1+2/2+1/3 =
2.33
1
1
2
1
Variations to account for:
• Abundance
• Extinction Risk
Three uses of trees for conservation
2. Evolutionary Distinctiveness (ED)
How much does each species contribute to overall PD?
ED gives a value to tree tips, not areas, by allocating the length of each branch
equally to its descendent tips.
Collen et al (2011) Phil. Trans
Go to: www.edgeofexistence.org/mammals/
www.edgeofexistence.org
Three uses of trees for conservation
3. Systematic conservation planning
Which areas would one protect, to represent the most
PD in reserves?
4
A
3
A
B
1
1
1
C
B
2
1
3
D
Area
1
2
3
4
Species
2
1
1
2
PD
8
4
4
5
2
C D
4
Three uses of trees for conservation
Preserving the evolutionary
potential of floras in
biodiversity hotspots
Forest et al, 2007. Nature
Cape Floristic Region, South Africa
Selecting areas to maximise
phylogenetic diversity (PD) in
protected areas rather than the
number of taxa results in:
• different choice of areas
• more diversity protected
without increasing the area for
conservation
Red = more PD than expected from named taxa
Blue = less PD than expected from named taxa
Phylogeny matters for conservation
d
… affects which areas are of greatest
conservation importance
PD of plant genera in the Cape Floristic Region
… affects which taxa are of greatest
conservation importance
Collen et al (2011) Phil. Trans
… can help to assess impacts on
biodiversity
in this case for eucalypts
Gonzalez Orozco et al (2016) Nature
Climate Change
Forest et al (2007) Nature
www.edgeofexistence.org
Phylogeny matters for conservation
… because it can detect patterns independent of named taxa
Northern lineages
of H. binoei
Phylogenetic
endemism within
H. binoei
Moritz et al (2016) Molecular Ecology
Oedura
Oedura marmorata
PHYLOGENETIC METRICS CAN
INDICATE EVOLUTIONARY REFUGIA
Stability filters diversity
Identifying refugia where diversity has persisted
through late-Pleistocene climate cycles
Atlantic rainforest
modelled distribution
present
6kya
21 kya
last glacial
maximum
Hypsibolas albomarginatus
Carnaval et al 2009 Science
Carnaval & Moritz 2008
J. Biogeography
Lineage endemism
Rosauer et al (2015) PLOS ONE
Explanations of endemism
Are endemic fauna are concentrated in:
historically stable rainforest habitat?
•
•
•
currently good rainforest habitat?
complex topography?
regions with smaller total area?
Lineage endemism
•
(good habitat connected through time and space)
Dynamic stability of
rainforest over 120ka
Rosauer et al (2015) PLOS ONE
Phylogenetic Endemism of Terrestrial Mammals
A score of 90 is equivalent to:
* 90 million years of evolutionary history restricted a single cell,
* 180 million years spread across 2 cells…
… or any additive combination like this
Rosauer & Jetz (2015) Global Ecology & Biogeography
What predicts PE?
1. Primary productivity
2. Isolation
• Isolation at the Last Glacial Maximum > Current Isolation
ΔAIC = 86
• reverse is true for species endemism
explained better by current isolation
• LGM effect stronger after controlling
for species endemism
So persistent isolation linked to
deeper endemism
Connections at LGM may represent
patterns repeated through many
glacial cycles
Unnamed
species
Deep splits
within species
Gehyra nana
Photo: Stephen Zozaya
Northern Spotted Rock Dtella
Species not
monophyletic
So what do we miss if we
map biodiversity only at
species level?
Summary
Reasons to map spatial patterns of phylogenetic diversity
• Help to better allocate conservation resources when not
everything can be protected – minimise losses from the tree
of life
• Can be more sensitive than species-level analyses to detect
signal of evolutionary processes
• Support spatial analyses (for any purpose) when taxonomy is
inaccurate or incomplete (as it usually is)
There is a phylogenetic version of any
species-based metric of biodiversity.
DISCUSSION