Analisis de Datos en Sistematica

A report on a workshop on Geometric Morphometrics

The Societad Latinoamericana de Teriologia held its first congress, I CO.LA.TER. between December 7 and Dec. 11, 1992, in the extraordinary tropical envirnment of the Universidad Simon Bolivar, near Caracas. During the congress, a very important meeting for Latinoamerican mammalogists which will lead to the birth of the new journal 'Marmosiana'. A workshop on geometric morphometrics was included in the program.

The workshop took place at the Computer Center of the University (with the invaluable help of Marcos Palma), where 25 people spent two & one half days with hands on computers and minds warping. The people were conducted through the principles of geometric morphometrics by Marco Corti, with the assistance of Roberto Cipriani and Marcelo Sancez.

Participants were from Venezuela, Chile, Colombia, Cuba, Spain and U.S.A. All received a booklet with some words introducing geometric morphometrics, followed by a step by step commented walk through the workshop, and by a list of SAS programs (a copy with the data can be obtained from me or Les Marcus by E-mail and will be put on SBBIOVM shortly so you can FTP it if you want). This was used for an example data set recorded from a collection of four Italian chromosomal races of the house mouse (Mus musculus domesticus) which form a monophyletic group. 14 landmarks were collected digitized on each mandible.

During the first day participants transformed raw data into shape coordinates. They made a good number of plots of individual landmarks and of configurations, analyzed differences in (centroid) size and in uniform component, and then ran univariate and multivariate statistics on shape coordinates. All this was done in SAS, with a brief introduction to the package. The day terminated with fun (needed after 4 hours of SAS), with an introduction to NTSYS, with dendrograms from Mahalanobis D, 2-D plots of CVA, and 3-D rotating plots.

The second day was entirely for Procrustes ('old' GRF) and for Thin Plate Spline. Consensus configurations were prepared using Least Squares Procrustes after comparisons with Resistant Fit, after having explored all the options in the program. Consensus were then submitted to Relative Warp analysis, exploring the use of different alpha values and including or excluding affine. The day terminated with individual play on TPS program, making landmarks with the mouse, and observing how changes in the position of landmarks can affect the uniform and the non-uniform shape change with its Principal Warps.

There are two comments on the workshop that should be made.

First, the program was really dense, a complete landmark study, but participants were so brilliant that - apparently - they all could really walk through this complex stuff. Landmarks conquered Latin America!

The second comment is more inherent to the debate wich is going on in (and out of) MORPHMET: this is the first time - as far as I know - that a workshop on geometric morphometrics was run by someone who was originally a 'user', although with a great interest in understanding details and experimenting with the new ideas that come out day by day. The only formula shown during the workshop was that of the fundamental solution of the biharmonic equation. Principal, Partial, and Relative Warps, and the reason why they are so good, were explained using a 'user to user' language. Naturally, this did not pretend to be more than a general introduction to geometric morphometrics (remember all this was done in 8 hours), and the people wishing to use this approach for their particular biological problem need a better understanding.

One difference from previous workshops was that the emphasis was on a real biological problem, with a picture of the phylogenetic tree and the chromosomal races' ranges always in front of participants to be compared with printouts and results in order to formulate hypothese - which participants were constantly asked to do.

I was and am very curious to receive feedback from this experience which was really intense for the participants, the assistents and the conductor. The original intention was to provoke interest in the tools by experiencing them in a most exciting way for the mind.

Marco Corti, Dipartimento di Biologia Animale e dell'Uomo, Universita' di Roma 'La Sapienza', Via Borelli 50, 00161 Roma, Italy. E-mail: CORTI@ITCASPUR.bitnet