macroevolution of organismal modularity and integration · 2003-03-07 · evolvability a modularity...

Macroevolution of OrganismalModularity and Integration

Gunther J. EbleBioinformatik, Universität Leipzig

What is macroevolution?

- evolution at and above the species level- large-scale phenotypic evolution- among-species evolution

Macroevolutionary biology: the study of theorigin and sorting of macroevolutionaryvariation

- diversity, origination, and extinction- evolutionary radiations- innovations- morphological disparity- constraints on form and patterning of morphospace- structure of the genotype-phenotype map- selectivity of and recovery from extinction- phylogenetic history and trends- community structure- size and allometry- complexity

Bits and Pieces: isolated yetintegrated

Organismal Modularity:definition and theoretical

justification• Definition: Dissociability of phenotypic wholes into

parts• Justification: organizational and variational semi-

independence of morphological organization;morphology itself is involved in the generation ofnew morphological elements late in ontogeny;morphostatic mechanisms (including physiologicalhomeostasis, regeneration, and repair) rely oninformation conveyed by morphological states.

Organismal Modules:causal roles

- as raw material for combinatorial diversification- as the substrate for changes in integration- as units of hierarchical sorting and selection- modularity as a property of clades

Needed: rigorous documentation of modules andof modularity in a macroevolutionary context.

Metrics of organismal modularity

- Conceptually and statistically a number of explicitmetrics can be advanced, but operationally someproxy metrics appear most useful:

- number of parts, constructional elements,characters- within- and among-module integration- disparity

Disparity: the conceptDisparity: the concept

Taxonomic diversity: sample 1 = sample 2 = 6 spp.Morphological disparity: ∆ sample 1 << ∆ sample 2

SampleSample 1 1 SampleSample 2 2

Empirical morphospacesEmpirical morphospaces

-0,28 -0,80 0,58 0,13 -0,78 0,070,76 0,41 0,10 0,08 0,40 -0,46

-0,73 -1,13 -0,56 -0,31 -1,10 0,50-1,71 4,45 1,07 -0,83 4,37 1,671,56 -0,36 1,18 1,18 -0,35 -2,03

X1 Y1 Z1 X2 Y2 Z2Species ASpecies BSpecies CSpecies DSpecies E

...

... p

n

PC1PC2

EIGENVECTOR EXTRACTIONEIGENVECTOR EXTRACTION

MORPHOSPACEMORPHOSPACEPROJECTIONPROJECTION

DATA MATRIX:DATA MATRIX:n SPECIES x p VARIABLESn SPECIES x p VARIABLES

Disparity = dispersion in morphospace

- Ss (sum of univariate variances)

- Sl (sum of eigenvalues)

Disparity = dispersion in morphospace

- Ss (sum of univariate variances)

- Sl (sum of eigenvalues)

-1

-,75

-,5

-,25

0

,25

,5

,75

1

AXI

S 2

-1 -,75 -,5 -,25 0 ,25 ,5 ,75 1

AXIS 1

Sp. A

Sp. DSp. C

Sp. ESp. B

-1

-,75

-,5

-,25

0

,25

,5

,75

1

AXI

S 2

-1 -,75 -,5 -,25 0 ,25 ,5 ,75 1

AXIS 1

Sp. A

Sp. DSp. C

Sp. ESp. B

Measures of disparityMeasures of disparity

2

n

n

- mean Euclidean distance- mean Euclidean distance

- Procrustes distance- Procrustes distance

AustralopithecusAustralopithecus to Homo to Homo

HomoHomo ergaster ergaster to to HomoHomo erectus erectus

HomoHomo erectus erectus to to Homo sapiensHomo sapiens

∑∆2 = 1.08

∑∆2 = 0.37 ∑∆2 = 0.58

5-Br

6-La

7-On

8-In9-Op

4-Gl3-Na

2-Ns

1-Pr

11-LBoS

10-Ba

Modeling divergence between speciesand between genera

Modeling divergence between speciesand between genera

Morphologicaldiversification

Morphological extinction

(Duboule, 1994)

THE HOURGLASS MODEL

Juvenile disparityJuvenile disparity

AdultdisparityAdultdisparity

-3

-2,5-2

-1,5

-1

-,50

,5

11,5

2PC II

-2 -1,5 -1 -,5 0 ,5 1 1,5 2 2,5

PC I

X JUVENILEO ADULT

Variance

PCI PCII1.06! 0.43

0.54! 1.15

p<0.015 p<0.019 (bootstrap z-test)

x

DEVELOPMENTAL MORPHOSPACEOrder Spatangoida

DEVELOPMENTAL MORPHOSPACEOrder Spatangoida

PCII

Error bars based on 1000 bootstrap replications

.35

.4

.45

.5

.55

.6

.65

C.V

. Mea

n Pa

irwis

e D

ista

nce

LARVAE ADULTS

n=26

n=26

Disparity of Larvae X Disparity of AdultsDisparity of Larvae X Disparity of Adults

Echinoderms as model organisms forthe study of modularity and integration

- Many plates and other skeletal elements- Many types of plates- Distinct body regions and growth fields

Issues to be addressed: relationship with diversity,size, evolvability, trends, and context-dependence

Modularity and TaxonomicDiversity

0

20

40

60

80

100

120

140

160

Diversity o

f Gen

era

10 15 20 25 30 35 40 45 50 55 60Number of Derived Character States

p < 0.018 (Spearman)

t

XXXXXXXX

XXXX

XX-2.5

-1.5

-.5

.5

1.5

2.5

-2.5 -1.5 -.5 .5 1.5 2.5

x

x

-2.5

-1.5

-.5

.5

1.5

2.5

-2.5 -1.5 -.5 .5 1.5 2.5

x

xx

x

-2.5

-1.5

-.5

.5

1.5

2.5

-2.5 -1.5 -.5 .5 1.5 2.5

x

xx

xxx

x x

DIVERSITY! ! ! ! ! ! DISPARITY

Discordances betweendiversity and disparityDiscordances betweendiversity and disparity

Sea urchin disparity and diversitySea urchin disparity and diversityDi spa rit yDi ver sit y

0

5

10

15

20

25

30

35

40

45

0

10

20

30

40

50

60

70

80

Disparity (Total V

aria

nce)

Gen

eric

Div

ersi

ty

J1 J2 J3 K1 K2 K3 K4 K5 Pal

Ate los tomat a

Disp

arity

(tot

al v

aria

nce)

Disp

arity

(tot

al v

aria

nce)

Div

ersit

y (g

ener

a)D

iver

sity

(gen

era)

Geological timeGeological time

Modularity and Body Size Variance

.15.2

.25.3

.35.4

.45.5

.55.6

.65

Variance ln(L

engt

h)

10 15 20 25 30 35 40 45 50 55 60

Number of Derived Character States

p < 0.76 (Spearman)

Modularity and Evolvability

In microevolution: “ability to sometimes produce improvement (Wagner and Altenberg 1996)

In macroevolution: “ability to sometimes produce substantial morphological change”

Evolvability a Modularity a Integration a Disparity

Integration among plastraland nonplastral modules

05

1015202530354045

Disp

arity

(Tot

al V

aria

nce)

J1 J2 J3 K1 K2 K3 K4 K5 PalGeologic Time

Integration within plastral module

Geologic Time

0

5

10

15

20

25

Plas

tral D

ispar

ity (S

um o

f Var

ianc

es)

J1 J2 J3 K1 K2 K3 K4 K5 Pal

Integration within nonplastralmodule

Geologic Time

0

5

10

15

20

25

30

J1 J2 J3 K1 K2 K3 K4 K5 Pal

Non

plas

tral D

ispar

ity (S

um o

f Var

ianc

es)

echinoidasteroidcystoidcamptostromatoid

Modularity and major trends inechinoderm evolution -EAT Theory

Context-dependence of modularity

Pz Post-Pz

>

Modularity as number of plate columns

Context-dependence of modularity

Regular Irregular

Modularity as number of plates

>

Context-dependence of modularity

Regular Irregular

Modularity as number of plate types

<

Context-dependence of modularityModularity as number of growth zones

Constant across echinoids

Is integrationthe converse of modularity?

No,“because the whole is more than the sum of theparts”

Yes,“because, all other things being equal, integrationand parcellation are logical opposites and areinversely correlated”

Is integrationthe converse of modularity?

Wait…

Yes and no: they are logical opposites and tend tobe inversely correlated, but the opposition may notbe symmetric and the correlation imperfect becauseof

1) The geometry of organisms2) The topology of morphospace3) Historical contingency

Is integrationthe converse of modularity?

1) The geometry of organisms

The size and shape of organismal parts affectsconnectivity and the strength of interactions amongparts.

Ex. For homogeneous parts such as serial homologues, modularitymay increase of decrease without change of integration

Is integrationthe converse of modularity?

2) The topology of morphospace

Heterogeneities in morphospace, implyasymmetric transition probabilities: in any particularevolutionary trajectory, changes in modularity orintegration may not be reversible or else have alower probability of reversal.

Is integrationthe converse of modularity?

3) Historical contingency

Modules have potentially different degrees ofentrenchment, and chance may at times lead to lossof modularity as well as of integration (e.g., limbloss).

Future Challenges

- Are morphometric landmarks minimal modules? Ifso, in what sense?

- Can morphospaces themselves be differentiallymodular or integrated? In principle.

- How to address the relationships betweenmodularity, integration, disparity, and complexity ina single framework? Or is more than oneframework needed?