Lecture 1 - Introduction To Geometric Morphometrics

Department of Geological Sciences | Indiana University (c) 2012, P.
David Polly
G562 Geometric Morphometrics
Rychlik, L., G. Ramalhinho, and P. D. Polly.
2006. J. Zool. Syst. Evol. Res. 44: 339-351.
An introduction
Geometric morphometrics
P. David Polly
Department of Geology, Indiana
(Biology and Anthropology)
University, 1001 E. 10th Street, Bloomington, IN 47405, USA
pdpolly@indiana.edu
http://mypage.iu.edu/~pdpolly/
Department of Geological Sciences | Indiana University (c) 2012, P. David Polly
Course website
http://www.indiana.edu/~g562/
Spring 2014 Syllabus
!"#$ %&'() *++(,-.$-#+
17-!an-14 lnLroducLlon Lo CeomeLrlc MorphomeLrlc MeLhods (CMM)
24-!an-14 Applled: lnLroducLlon Lo MaLhemaLlca AsslgnmenL 1
31-!an-14 Aopplled: llrsL CMM Analysls AsslgnmenL 2
7-leb-14 no MeeLlng 8LAulnC 1
14-leb-14 Applled: rocrusLes, rlnclpal componenLs analysls and morphospace AsslgnmenL 3
21-leb-14 ulscusslon: MulLlvarlaLe LesLs and 28-LS 8LAulnC 2
28-leb-14 Applled: MulLlvarlaLe LesLs for CMM (proposals due) AsslgnmenL 4
7-Mar-14 ulscusslon: 8andomlzaLlon and 8ooLsLrap 8LAulnC 3
14-Mar-14 Applled: 8ooLsLrapplng and randomlzaLlon LesLs AsslgnmenL 3
21-Mar-14 no MeeLlng: Sprlng 8reak
28-Mar-14 ulscusslon: CuLllnes and LuMA 8LAulnC 4
4-Apr-14 Applled: CuLllne analysls and LuMA AsslgnmenL 6
11-Apr-14 ulscusslon: hylogeneLlc Lrees and morphomeLrlcs 8LAulnC 3
18-Apr-14 Applled: phylogeneLlc Lrees and morphomeLrlcs AsslgnmenL 7
23-Apr-14 ulscusslon: evoluLlonary and growLh Lra[ecLorles 8LAulnC 6
2-May-14 Applled: MonLe Carlo slmulaLlons AsslgnmenL 7
3-May-14 ro[ecLs uue: Monday, May 3
Computers and Mathematica
Laptops should be brought to class every week. The software we will use is
compatible with both Mac and Microsoft operating systems.
General software: If possible, please install the following on your system:
ImageJ (http://rsbweb.nih.gov/ij/ ), Adobe Creative Suite (especially
Photoshop and Illustrator, http://iuware.iu.edu/ ), Microsoft Office or Mac
iWork or OpenOffice (for spreadsheet programs, http://iuware.iu.edu/ ).
Mathematica: Most of our work will be done in the Mathematica which is a
mathematical and statistical application that does efficient computations, is
flexible for customized analysis, has powerful graphics capabilities, and is easy
to learn (compared to R, for example, which is an equivalent application). You
can purchase a student license for Mathematica 8.04 at the IU Stat/Math
Center at 410 Park Avenue for $30 (cash or check only). They will provide
licence code and download link for installation.
Participation 20%
Assignments 20%
Reading reports 10%
Project proposal 10% (due 7 March)
Project report 40% (due 5 May)
Grading
Projects
The focal assignment for the course is to carry out an original morphometric study
that you will present as a written project paper. Ideally your project will be related to
your broader research interests. During the first half of the course you will develop
your study, with advice if you need/want it.
Project proposals are due mid semester. Proposals consist of a one to two page
description of the aim and subject of your project, the source of your data, the
question your study will try to answer, and the tools youll need to collect your data.
Please feel free to consult me as you develop your proposal.
The final project report should be modelled on a scientific journal paper, with
introduction to the problem, materials and methods, results, discussion,
conclusion, references (cited properly in the journal format of your choice). Target
length is 10-15 pages, including figures, tables, and references.
Supplementary texts
Hammer, . and D.A.T. Harper. 2006. Palaeontological Data Analysis.
Blackwell Publishing, Oxford, United Kingdom.
Zelditch, M.L., D.L. Swiderski, H.D. Sheets, and W.L. Fink. 2004. Geometric
Morphometrics for Biologists: a Primer. Elsevier Academic Press, San Diego,
California.
MacLeod, N. 2012. Palaeo Math 101. A series of essays published in the
Palaeontological Association Newsletter. http://www.palass.org/
modules.php?name=palaeo_math
What are morphometrics?
Any quantitative measurement and
analysis of morphological traits
Mandible Length (cm)
M
a
n
d
i
b
l
e

H
e
i
g
h
t

(
c
m
)

Shrew
Marmot
Geometric morphometrics
the quantitative representation and
analysis of morphological shape using
geometric coordinates instead of
measurements
Component 1
C
o
m
p
o
n
e
n
t

2

Shrew
Marmot
The Major Goal of Morphometrics:
Measuring morphological similarity and difference
Types of geometric representation
Outlines (2D or 3D)
Semilandmarks, sliding landmarks
3D surfaces
Landmarks (2D or 3D)
Polly, P. D. 2008. Adaptive Zones and the Pinniped Ankle: A 3D
Quantitative Analysis of Carnivoran Tarsal Evolution. Pp.
165-194 in (E. Sargis and M. Dagosto, Eds.) Mammalian
Evolutionary Morphology: A Tribute to Frederick S. Szalay.
Springer: Dordrecht, The Netherlands.
Polly, P. D. 2003. Paleophylogeography: the
tempo of geographic differentiation in marmots
(Marmota). Journal of Mammalogy, 84: 369-384.
Definitions
Landmark any point described with cartesian coordinates (x, y, z) used to
represent the shape of a structure.
Landmark (2) any point that can be placed on a biologically or geometrically
homologous point on the structure.
Semi-landmark a point that is placed arbitrarily using an algorithm, often by
defining endpoints at biologically homologous points and placing a specified
number of semilandmarks between them.
0
10
0
10
5
5
X-axis
Y
-
a
x
i
s

(2,5)
(4,8)
(5,2)
(6,8)
(6,2)
Landmarks
Landmarks are coordinate points used to
represent a shape
They are quantified as Cartesian
coordinates (x,y[,z])
At least 3 are required (two points make a
line)
Example analyses: Relative Warps (PCA of
landmarks), Euclidean Distance Matrix
Analysis (EDMA) of distances between
landmarks
0
10
0
10
5
5
X-axis
Y
-
a
x
i
s

Outlines
Outlines are perimeters delimited by many
points
coordinates (x,y[,z]), often converted to
angles
Many points are required to represent a
shape
Example analyses: Semilandmarks,
sliding semilandmarks, Eigenshape (PCA
of outline), Fourier analysis
Polly, P. D. 2008. Adaptive Zones and the Pinniped Ankle: A 3D
Quantitative Analysis of Carnivoran Tarsal Evolution. Pp.
165-194 in (E. Sargis and M. Dagosto, Eds.) Mammalian
Evolutionary Morphology: A Tribute to Frederick S. Szalay.
Springer: Dordrecht, The Netherlands.
Surfaces
Surfaces are the 3D surface of an object
coordinates (x,y,z)
Many points are required to represent a
shape
Example analysis: Eigensurface (PCA of
surfaces), sliding semilandmarks
Advantages to geometric representation
Results can be presented visually as a
shape than tables of numbers
Data are easily collected from digital
photographs
Size is mathematically removed from the
analysis to focus on pure shape
X-axis
Y
-
a
x
i
s

Y
-
a
x
i
s

X-axis
Y
-
a
x
i
s

Vector Plot
Spline Plot
GMM results can be presented graphically
Two examples of graphically
showing those differences
Difference in shape of mandibles
of shrew and marmot
Marmot
Shrew
Traditional morphometrics mixes size and shape
The size of the animal affects all measurements so
that primary morphometric difference between two
taxa is size rather than shape
Marmot
Shrew
0
5
10
15
20
25
M
i
l
l
i
m
e
t
r
e
s

H L H L
Marmot
Shrew
Geometric morphometrics removes size by rescaling
Shapes are enlarged or reduced to achieve a standard,
equal size
Coordinates of rescaled landmarks show difference in
relative position only
X-axis
Y
-
a
x
i
s

Disadvantages of geometric
representation
Size is completely absent from the
analysis, and size may be biologically
relevant
Only single rigid structures can be easily
analyzed
-1.50
-1.00
-0.50
0.00
0.50
1.00
1.50
-4.0 -2.0 0.0 2.0 4.0
Log body mass (kg)
L
o
g

m
e
t
a
b
o
l
i
c

r
a
t
e

(
O
2
/
h
r
/
g
)

data from Eisenberg, 1981
Size is biologically important
and it may be of interest in a morphometric analysis
Polly, P. D. 1998. Variability in mammalian dentitions: size-
related bias in the coefficient of variation. Biological Journal of
the Linnean Society, 64: 83-99.
Size and shape may behave differently
Size or shape data may be appropriate for different analyses
Only single rigid structures can represented with
geometric morphometrics
Okay
Not okay
A short history of geometric morphometrics....
Albrecht Drer (1471 -1528)
On Growth and Form, 1917
DArcy Thompson (1860-1948)
Francis Galton (1822-1911)
1891: starts biometric laboratory at
University College London
Biometric approach to genetics:
regression & correlation
Composite portraiture: photographs of
different subjects combined (through
repeated limited exposure) to produce a
single blended image
Anthropometry & differential psychology:
quantitative analysis of fingerprints
Modern Geometric Morphometric Methods (GMM)
Development of landmark geometrics was
driven by Fred Bookstein (long of University
of Michigan, now Washington and Vienna)
Joined very productively by F. James Rohlf
(Stony Brook)
Ian Dryden, Kanti Mardia, Les Marcus, and
Dennis Slice have been important names
in developing techniques and theory.
Bookstein was originally intent on creating
a truly quantitative way of producing dArcy
Thompson transformation grids.
Bookstein
Steps in a geometric morphometric study
Study design
Data collection
Data standardization
Analysis
Results interpretation
How do you choose landmarks
(or outlines, or surfaces)?
1. The data must reflect a hypothesis
2. The data must represent the shape adequately
3. Landmarks must be present on all specimens
Measurement Error and Sample size
1. Measurement error (ME) always exists in any collection of data, but ME
doesnt matter if it is substantially less than the differences you want to
measure.
2. Sample size required for a particular study depends on the within-group
variation relative to differences between groups.
How many specimens do I need?
Depends on the question being addressed
Depends on the error in your data
You need more specimens when the differences you want to measure are
small compared to the variation within your group (natural or due to error)
For sexual dimorphism in skulls of humans or other primates, 10

individuals of each sex might be enough
For differences in genetic strains of mice where the mutation doesnt

obviously affect the skeleton, 50 individuals of each strain is more realistic
For species that belong to different families or orders, 1 specimen per

species is almost always sufficient
What morphometrics cant answer for you..
Morphometrics does not tell you what large or difference or shape mean
(These are definitions you must supply and your results depend upon them)
Morphometrics does not tell you whether you unwittingly have two
unrecognized groups in a single sample
(Although comparison with known groups may help such an endeavour)
How to identify cladistic characters

(For the first two reasons combined)
Examples of available software
Digitizing landmarks and outlines: tpsDIG, ImageJ
Superimposition: Morpheus (plus integrated in some below)
Outline analysis: Eigenshape, PAST
MANOVA: Statistica, PAST
Discriminant functions, CVA: Statistica, PAST
Principal components analysis of landmarks: tpsRELW, PST
Construction of trees: PHYLIP, PAUP, NTSYSpc, PAST
Simulations: Mathematica, R
Links and downloads at SUNY Stony Brook morphometrics site:
http://life.bio.sunysb.edu/morph/
Equipment: 2D outlines and coordinates
High-quality digital cameras
(resolution doesnt matter as much as the possibility of
lens distortion: test your camera first by photographing a
piece of graph paper and looking for fish eye distortion)
Calipers or scale bar
Equipment: 3D outlines and coordinates
Reflex Microscope for collecting
three-dimensional landmarks, outlines and
measurements (good for objects the size
of a cat skull down to things about 2-5 mm
long)
Microscribe robotic arm for collecting 3D
landmarks and measurements (good for
objects the size of a human skull down to
a rat skull)
3D surfaces
Microscan Laser scanner for scanning
surfaces (good for objects the size of a cat
skull down to ones about 2-3 cm long)
NextEngine laser scanner (good for objects
the size of a horse skull down to a single
tarsal bone)

Lecture 1 - Introduction To Geometric Morphometrics

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Lecture 1 - Introduction To Geometric Morphometrics

Transféré par

Droits d'auteur :

Formats disponibles

Department of Geological Sciences | Indiana University (c) 2012, P.

Depends on the question being addressed

Depends on the error in your data

For sexual dimorphism in skulls of humans or other primates, 10

For differences in genetic strains of mice where the mutation doesnt

For species that belong to different families or orders, 1 specimen per

How to identify cladistic characters

Vous aimerez peut-être aussi