A talk presenting my work recently published in PLoS One, at the Evolution meeting in 2012, in Ottawa. Examples of morphological differentiation illustrated with colorful pictures of a group known to many. You can find the published paper here, without the pocket monsters: Bapst DW (2013) When Can Clades Be Potentially Resolved with Morphology? PLoS ONE 8(4): e62312. doi:10.1371/journal.pone.0062312 http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0062312 Obviously all the Pokemon are copyright of Nintendo of America. But how useful pedagogically they were!

1. When Do We Lack Resolvable Clades?
Cladogram from Mitchell et al., 2007
Dave Bapst, University of Chicago
The Influence of Morphological Differentiation

2. Studying the Tree of Life
We can use
molecular
information to work
out how living
lineages are related…

3. Studying the Dead Branches
…paleontologists are
tasked with figuring
out the relationships
of extinct lineages…

4. Studying the Dead Branches
…generally based on
incomplete morphological
information.

5. Paleotrees Sometimes Poorly Resolved
Majority Rule Consensus Tree with
87 Graptolite Species (94 Characters)
40 observed nodes
86 possible nodes
= 46.5% resolved
Cladogram from Mitchell et al., 2007

6. Paleotrees Sometimes Poorly Resolved
Majority Rule Consensus Tree with
87 Graptolite Species (94 Characters)
40 observed nodes
86 possible nodes
= 46.5% resolved
Cladogram from Mitchell et al., 2007
Strict Consensus Trees from 17
Recent Analyses of Fossil Taxa
31 - 97% resolved
Mean of 77% resolved

7. Polytomies in Morphological Trees
• Incomplete datasets
• Homoplastic characters
• Intrinsically unresolvable relationships
– Non-dichotomous branching (Maddison, 1989)
– Modes of morphological differentiation and the
sampling of ancestors
• No synapomorphies to further resolve taxa
• Norell, 1992; Smith, 1994; Wagner and Erwin, 1995;
Lamboy, 1996
• How often should we expect an intrinsic lack
of synapomorphies to group taxa?

8. Polytomies in Morphological Trees
• Incomplete datasets
• Homoplastic characters
• Intrinsically unresolvable relationships
– Non-dichotomous branching (Maddison, 1989)
– Modes of morphological differentiation and the
sampling of ancestors
• No synapomorphies to further resolve taxa
• Norell, 1992; Smith, 1994; Wagner and Erwin, 1995;
Lamboy, 1996
• How often should we expect an intrinsic lack
of synapomorphies to group taxa?
…Wait, hold on!
Let’s review modes of
morphological differentiation…

9. Morphological divergence rapid on geologic
timescales in fossil-rich groups
• Temporal lineages static in systematic
characters
When divergence unassociated with
branching events (cladogenesis) that
differentiation pattern is termed anagenesis

10. Morphological Differentiation
Patterns with Branching
Cryptic Cladogenesis
• daughter lineages are not
immediately morphologically
distinguishable

11. Budding Cladogenesis
• where one daughter lineage
immediately diverges
morphologically from the
ancestor.
Wagner and Erwin (1995) found
support for this particular
pattern.

12. This morphological pattern can create
polytomies. To see this, let’s
incompletely sample the fossil record…

13. …and we would distinguish these four
morphotaxa. Yet, no further groups can be
formed which would share any derived
traits to the exclusion of other taxon units.

14. Not sampling ancestor does not allow for
(Vaporeon, Jolteon) clade to be resolved
An intrinsically unresolvable clade would
persist in the data as long as future
descendants of all three are considered

15. Cryptic cladogenesis
produces unresolvable
clades similar to budding

16. Bifurcating Cladogenesis
• Both daughter lineages become
morphologically distinct and the
ancestral morphotaxon does not
persist
Relationships among ancestor and
descendants are polytomy as
long as all three are sampled

17. Intrinsically Unresolvable Clades
• Possible negative effect on morph-based studies
– Lack of synapomorphies may lead to falsely inferred
relationships based on homeoplasy
– Unresolved consensus trees or weakly supported
nodes
• Misinterpreted as homoplastic or inadequate character
data

18. Intrinsically Unresolvable Clades
• Possible negative effect on morph-based studies
• How often should we expect intrinsically
unresolvable clades in real datasets?
– Identifying true polytomies difficult to do robustly
– Few empirical analyses: Wagner and Erwin (1995)
budding over bifurcation and anagenesis

19. Intrinsically Unresolvable Clades
• Possible negative effect on morph-based studies
• How often should we expect intrinsically
unresolvable clades in real datasets?
• Simulations necessary under a broad range of
differentiation patterns and sampling regimes
necessary to determine extent of phenomenon

20. Birth-Death Simulations with paleotree
• Modeled branching, extinction, anagenesis and sampling events
in the fossil record as Poisson processes
– R library paleotree (Bapst, in press, Methods in Ecology and Evolution)
• Simulate only shifts in morphology, rather than individual traits
– Shifts occur at origination of each morphologically differentiated taxon
– Taxa descended from each shift denote each resolvable set of taxa (clade)
– Ratio of resolvable clades to potential clades is resolvable proportion
• 13 Combinations of Modes of Morphological Differentiation
– Branching and extinction rates set equal (λ = μ)
– Anagenesis rate equal to diversification rates (= λ = μ) or zero
– One, two or three cladogenesis modes (with equal probabilities)
– Complexes of cryptic lineages always collapsed

21. Resolvable Proportion of Clades
Fully Extinct Clades
Complete Sampling
1000 runs, ~100 taxa
• With total sampling of fossil record, low
resolvability under budding / bifurcation

22. Resolvable Proportion of Clades
Fully Extinct Clades
Complete Sampling
1000 runs, ~100 taxa
• As expected, lower resolvability when some
cladogenesis is cryptic

23. Resolvable Proportion of Clades
Fully Extinct Clades
Complete Sampling
1000 runs, ~100 taxa
• Also as expected, increase in resolvability with
anagenesis (rate same as div rates)

24. Resolvable Proportion of Clades
Fully Extinct Clades
Incomplete Sampling
Samp Rate = λ = μ
1000 runs, ~100 taxa
• At sampling rate analogous to shelly marine invert records,
more clades are resolvable but most patterns predict 1/3rd
to 1/5th of observed clades to be intrinsically unresolvable

25. Resolvable Proportion of Clades
Fully Extant Clades
Complete Sampling
at Present-Day
1000 runs, 50-300 taxa
No Fossil Taxa
• Even with no sampling in the fossil record, well-sampled
extant-only morph datasets will have unresolvable clades
(except when all branching is via bifurcation)

26. Additionally…
• Some patterns often produce large intrinsic
polytomies when long-lived morphotaxa produce
many descendants
• Unresolvable clades present at all node depths
– Evolutionary history cannot erase deep polytomies
– Does selectively sampling among higher taxa change
this?
• High rates of anagenesis or, paradoxically, poor
sampling can increase intrinsic resolvability
– Under cryptic cladogenesis, almost no observed
durations for morphotaxa in datasets with few
unresolvable clades

27. Conclusions
• When should we expect intrinsically unresolvable
clades in morphological systematics?
– At least in groups where morphotaxa are consistently
sampled across geologic time
– Present in both fossil and extant analyses
• Having partially unresolved relationships may be
an expected outcome in morph analyses, rather
than a symptom of inadequate data
– Need analyses to quantify phylogenetic uncertainty
Thanks to M. Foote, D. Jablonski, M. Webster,
S. Kidwell, M. Pennell, E. King, A. Krug, J.
Brown and C. Belanger for comments…And
thanks to all the Pokemon!

30. As rate of anagenesis increases relative to div rates, number of resolvable clades increases but
the mean duration of observed taxon ranges drops, no longer reflecting extinction rate.
~600 extinct clades with ~100 sampled morphotaxa

31. Under very poor preservation (sampling rate = tenth of div rates), some models predict that
almost all clades will be resolvable, but not all models of differentiation.

34. Inferring Relationships from
Morphology: Characters
Wings Feathers Bones
Does not
Meta.
Flightless
Head
Crest
Pidgeot 1 1 1 1 0 1
Butterfree 1 0 0 0 0 NA
Farfetch’d 1 1 1 1 1 0
Dodrio 0 1 1 1 1 1
• Relationships among fossil taxa are pieced
together used shared derived morphological
features that are shared by only a portion of taxa
• These are called synapomorphies
• N.B. the lack of three heads is a plesiomorphy

35. • Well supported clades are taxa that share
many characters to the exclusion of other taxa
– The set of relationships that infers the least
changes is generally taken as the best solution
– Many groups often have multiple best solutions
Feathers, bones,
loss of metamorphisis
Wings, Flight
Inferring Relationships from
Morphology: Using Parsimony

36. • Relationships that cannot be resolved become
non-bifurcating polytomies
– Polytomies are generally seen as a failure of the
cladistic analysis (not enough information)
– For a roote fully-bifurcating tree of N taxa, there
would be N-1 nesting clades/nodes
Polytomies:
A Failure to Infer Relationships?
No clear
relationships
among these
three taxa

37. Could we resolve the relationships if
we don’t sample the ancestor?

38. Four Legs On Fire Electrified Scaly
Flameon 1 1 0 0
Jolteon 1 0 1 0
Vaporeon 1 0 0 1

39. Let’s incompletely sample this record…

40. …And we might obtain these three
morphotaxa.

41. Four Legs
Shellder
on Tail
Shellder on
Head
Slowpoke 1 0 0
Slowbro 1 1 0
Slowking 1 0 1

42. Note, in practice, the indep. acquisition of
bipedality and a commensual Shellder
would probably cause Slowbro and Slowking
to be incorrectly united as sister taxa
-The difficulties of defining homology!
If we never sampled the
ancestor, we would just have
a bifurcation.

43. Four Legs On Fire Electrified Scaly
Evee 1 0 0 0
Flameon 1 1 0 0
Jolteon 1 0 1 0
Vaporeon 1 0 0 1

44. Four Legs On Fire Electrified Scaly
Flameon 1 1 0 0
Jolteon 1 0 1 0
Vaporeon 1 0 0 1
We’d have this
problem even if
we didn’t sample
the ancestor!

45. Pure Budding Sims:
50-60% of the nodes
Bifurcating and Anagenesis
most resolved

46. Wagner and Erwin found support for mixture of budding
cladogenesis and anagenesis in forams and gastropods. Here,
when rates of budding and anagenesis are equal, 65-75% of
clades are resolvable.
(Apropos of nothing, W&E’s gastropod tree is ~60% resolved.)

47. As sampling rates INCREASE, proportion of
resolvable clades DECREASE in all scenarios
Change strongest with bifurcation
Thus, expect well sampled groups to be more poorly resolved in fossil record

48. Budding cladogenesis drops number of
resolvable clades in extant-only analyses

49. Dave Bapst
University of Chicago
Geophysical Sciences
When Do We Lack Resolvable Clades?