The optimization of landmark data: a three-dimensional approach

CATALANO, SANTIAGO; GOLOBOFF, PABLO; GIANNINI, NORBERTO

San Miguel de Tucuman, Argentina

Simposio; XXVII Annual Meeting of the Willi Hennig Society; 2008

Willi Hennig Society

  Since the implementation of Farris optimization toanalyze continuous characters in TNT, there has been arenewed interest in using such characters in phyloge-netic analyses, and in trying to extend phylogeneticanalysis to quantitative descriptions of shape. Allprevious approaches were based on the optimizationof values from some sort of unidimensional shapedescriptor from the terminals. We consider that a morenatural method should be based on explicitly estimatingoptimal ancestral shapes. Particularly, in the case oflandmark configurations, we propose a generalizationof Farris optimization to more than one dimension. Inthis approach, each landmark in the ancestor can beplaced on any position in a space that presents thesame dimensionality as that on which the landmarkdata were defined. Under a 3D-optimization, theoptimal landmark positions of ancestors are thosewhich minimize the distances between homologouslandmarks along branches. Our approach synthesizeswell-known elements of phylogenetic analysis into anew perspective for the problem of phylogenetic anal-ysis of shape.